Animal Models of Episodic Memory

Jonathon D. Crystal

Department of Psychological & Brain Sciences
Indiana University

Reading Options

---

Abstract

People retrieve episodic memories about specific earlier events that happened to them. Accordingly, researchers have sought to evaluate the hypothesis that nonhumans retrieve episodic memories. The central hypothesis of an animal model of episodic memory is that, at the moment of a memory assessment, the animal retrieves a memory of the specific earlier event. Testing this hypothesis requires the elimination of nonepisodic memory hypotheses. A number of case studies focus on the development of animal models of episodic memory, including what-where-when memory, source memory, item-in-context memory, and unexpected questions. Compelling evidence for episodic memory comes from studies in which judgments of familiarity cannot produce accurate choices in memory assessments. These approaches may be used to explore the evolution of cognition.

Keywords: episodic memory, what-where-when memory, source memory, binding, item-in-context memory

Author Note: Jonathon D. Crystal, Department of Psychological & Brain Sciences, Indiana University, Bloomington, IN 47405-7007.

Correspondence should be addressed to Jonathon D. Crystal at jcrystal@indiana.edu.

Acknowledgments: This work was supported by National
Institute on Aging grant AG044530 and AG051753.

---

Introduction

Fundamental aspects of human cognition raise a natural question, namely, How widely distributed are elements of cognition among nonhuman animals? Exploring the distribution of cognitive processes in animals may provide insight into the evolution of cognition (Emery & Clayton, 2004; Gallistel, 1990). This review focuses on episodic memory (see Table 1). Students of human memory focus on episodic memory because it stores personal past experiences of an individual. In this respect, episodic memory is self-referencing. By contrast, other memory systems store facts without retaining other features that accompany memory storage. Moreover, students of human memory have been concerned with subjective experiences that are thought to accompany episodic memory retrieval in people (Tulving, 1985, 1987). However, documenting behavioral expression of a putative subjective experience is problematic in animals. This review advocates that it is profitable to focus on the content of episodic memories, rather than the subjective experiences that may accompany episodic memory.

The central hypothesis of an animal model of episodic memory is that, at the moment of a memory assessment, the animal remembers back in time and retrieves a memory of the earlier event or episode (Crystal, 2013b, 2016a, 2016b). An important alternative explanation exists whenever the animal can solve the memory test without remembering back to the specific earlier event. In this review, I focus on judgments of familiarity as a primary nonepisodic memory alternative. According to this view, the presentation of a stimulus gives rise to a memory trace that passively decays as a function of time. Because the age of memories can be detected from a comparison of memory trace strengths, an animal may solve a new–old memory test by following a relatively simple rule such as, Choose the item that currently generates the lowest level of familiarity. Critically, an animal that uses judgments of relative familiarity need not retrieve an episodic memory of the earlier event (see details that follow). Other approaches have sought to document that animals can discriminate: combinations of item–place–context (Eacott & Norman, 2004; Kart-Teke, De Souza Silva, Huston, & Dere, 2006), the sequential order in which events (e.g., odors, objects) are presented (Dere, Huston, & De Souza Silva, 2005a, 2005b; Eacott, Easton, & Zinkivskay, 2005; Ergorul & Eichenbaum, 2004; Fortin, Wright, & Eichenbaum, 2004; Hunsaker, Lee, & Kesner, 2008; Kart-Teke et al., 2006; Kesner & Hunsaker, 2010; Kesner, Hunsaker, & Warthen, 2008), trial-by-trial records of information (Devkar & Wright, 2016; Kheifets, Freestone, & Gallistel, 2017; Wright, 2007), and the elements of configural learning (Iordanova, Burnett, Aggleton, Good, & Honey, 2009; Iordanova, Burnett, Good, & Honey, 2011; Iordanova, Good, & Honey, 2008). For related reviews, see Dere, Dere, De Souza Silva, Huston, and Zlomuzica (2017) and Eacott and Easton (2010).

Important to note, a range of assessments are likely needed to eliminate all candidate nonepisodic memory alternatives by relying on a strategy of converging lines of evidence (Crystal, 2009). In the sections that follow, I describe a number of case studies using rats that develop animal models of episodic memory. In each case, the central hypothesis just described is tested against a familiarity (nonepisodic memory) hypothesis.

Animal Models of Episodic Memory: Some Case Studies

This section reviews a number of case studies that focus on the development of animal models of episodic memory. The cases include what-where-when memory, source memory, item-in-context memory, and unexpected questions.

What-Where-When Memory

In a now classic paper, Clayton and Dickinson (1998) described the first evidence of episodic memory in a nonhuman using scrub jays (for reviews, see Clayton, Bussey, & Dickinson, 2003; Clayton, Bussey, Emery, & Dickinson, 2003; Clayton & Emery, 2015; Clayton, Salwiczek, & Dickinson, 2007; Griffiths, Dickinson, & Clayton, 1999); the phenomenon of episodic memory likely predates the more recent application of behavioral definitions of episodic memory. Food-storing scrub jays cached either peanuts followed by wax worms or, on other occasions, worms followed by peanuts; they retrieved the caches after a short or long retention interval. For some birds, the worms decayed after the long retention interval, and for other birds fresh worms were provided; peanuts did not decay, and worms were always fresh after the short retention interval. The birds learned to prefer the worm rather than peanut cache sites when the worms were fresh but reversed this preference when the worms were decayed. These data are consistent with the hypothesis that jays are sensitive to what (food type), where (location in the tray), and when (time of caching and recovery).

Clayton and colleagues focused on memory for what, where, and when an event occurred. Babb and Crystal (2005, 2006a, 2006b) and Naqshbandi, Feeney, McKenzie, and Roberts (2007) adapted this approach for use with rats. In these experiments, a distinctive flavor (e.g., chocolate) was encountered at a randomly selected arm in an eight-arm radial maze during a study episode, in addition to standard “chow”-flavored food. In a subsequent test of memory, the distinctive flavor sometimes replenished and chow never replenished. Notably, replenishment of the distinctly baited location replenished after one delay (e.g., a long retention interval such as 6 hr), whereas the distinctly baited location did not replenish after a different delay (e.g., a short retention interval such as 1 hr). Evidence for what-where-when memory comes from the observation that rats learned to revisit the distinctively baited location at a higher rate after the replenishment delay than after the nonreplenishment delay. Roberts and colleagues (2008) argued that our approach could be explained by judgments of relative familiarity. According to this view, because presentation of an event gives rise to a memory trace that decays as a function of time, memory trace strength is different after short and long delays. Thus, animals might have passed previous tests of what-where-when memory without remembering the episode by merely revisiting when the memory trace was at its typical level of decay (and not visiting when the memory trace was at a different level of decay). Zhou and Crystal (2009, 2011) responded to this criticism by equating familiarity across experimental conditions. The key innovation was the use of constant delays between encoding and memory assessments, thereby rendering familiarity signals nondiagnostic of replenishment/nonreplenishment. The critical question is, Can rats solve a what-where-when memory problem when familiarity does not provide diagnostic information?

To this end, Zhou and Crystal (2009) tested rats using an eight-arm radial maze (Figure 1A). At encoding, the rats were given access to four randomly selected arms, and one of these locations was randomly selected to provide access to chocolate-flavored food (all other arms in the maze provided standard chow-flavored food); the initial encoding opportunity provided the rats with their first helping of chocolate. After a brief retention interval delay, all eight arms were accessible in a memory assessment. Previously unvisited locations provided access to chow (and rats accurately avoided revisits to depleted chow locations). The location that provided chocolate in the encoding phase replenished additional chocolate in the memory assessment depending on the time of day at which the encoding phase occurred. For some rats, chocolate replenished when it had been encountered in the morning; for other rats, chocolate replenished in the afternoon. Chow locations never replenished. The replenishment location provided the rats with a second helping of chocolate. Of importance, the delay between encoding and memory assessment was constant (approximately 2 min during the initial training). Therefore, any judgments about the familiarity of earlier events (e.g., navigating, finding food, eating chocolate or chow, etc.) were constant in replenishment and nonreplenishment conditions. We proposed that, at the moment of memory assessment, the rats retrieved an episodic memory of the earlier event, including what happened (flavor), where it occurred (location), and when (i.e., the time of day) the event took place. Consistent with this proposal, the rats were more likely to revisit the replenishment location in the memory assessment relative to the nonreplenishment location (Figure 2A).

To test the episodic memory hypothesis, we conducted a number of experiments (Figure 1; Zhou & Crystal, 2009). In one experiment, we phase shifted the light onset in the colony to put time of day and time since light onset in the colony in conflict (Figure 1B). According to an episodic memory hypothesis, the rats remember the earlier event, including the time of day at which the event occurred; an animal could use a circadian representation of time to remember the time of day at which the event occurred, and we refer to this as the time-of-day hypothesis. However, according to a nonepisodic memory hypothesis, the animal may time the interval between light onset in the colony and the time at which replenishment/nonreplenishment occurs; we refer to this interval timing proposal as the how-long-ago hypothesis. A phase shift simulates what people commonly experience with jet lag, namely, that one’s circadian rhythm continues immediately after a time zone shift (Crystal, 2012). Notably, circadian rhythms do not immediately adjust to a new schedule (sunrise, mealtimes, etc.), thereby giving rise to the phenomenon of jet lag. Similarly, if the rats were using a circadian representation of time, then immediately after the phase shift in which light onset occurred earlier than normal, they would still treat a session conducted in the morning as a “morning” session because the circadian rhythm would not yet have adjusted. In contrast, if the rats were timing an interval with respect to light onset in the colony, a single change in light onset would produce an immediate change in the interval (cf. a stopwatch is reset at any time and does not exhibit jet lag). We put these two hypotheses in conflict by arranging the magnitude of the phase shift (change in light cycle) so that an animal that used interval timing would treat a session conducted in the morning as if it were an “afternoon” session after the phase shift. The rats revisited chocolate locations in accordance with time of day immediately after the phase shift occurred (Figure 2B).

In two additional experiments (Figure 1C–1D), we tested the hypothesis that the rats remembered the time of encoding (rather than merely being reactive at the time of the memory assessment). For example, we unexpectedly increased the delay between encoding and memory (7 hr instead of 2 min; Figure 1C). In another experiment, we put time-of-encoding and time-of-test predictions in conflict (Figure 1D). In this experiment, a session started at the usual time for a late session and ended at the usual time of an early session (note the unusual late followed by early). In this situation, a rat that was remembering the start time would behave differently than a rat that was reacting to the time at the moment of the test (by revisiting or withholding revisits as appropriate for early vs. late sessions). The rats revisited chocolate locations in accordance with the time of encoding (Figure 2C–2E). This work showed that rats remember the time of day at which a study episode occurred, in addition to what and where the event happened. Note that the use of constant delays between encoding and memory assessments rendered familiarity signals nondiagnostic of replenishment/nonreplenishment.

Source Memory

We developed an animal model of source memory (Crystal & Alford, 2014; Crystal, Alford, Zhou, & Hohmann, 2013; Crystal & Smith, 2014; Smith, Dalecki, & Crystal, 2017; Smith, Slivicki, Hohmann, & Crystal, 2017; Smith et al., 2016; reviewed in Crystal, 2016a; see Basile & Hampton, 2017, for an example of source memory in rhesus monkeys). Source memory is memory for the origin of episodic memories (Janowsky, Shimamura, & Squire, 1989; Johnson, Hashtroudi, & Lindsay, 1993; Mitchell & Johnson, 2009). In our approach, rats foraged in a radial maze for distinctive flavors of food that replenished or failed to replenish at its recently encountered location according to a source-information rule (Figure 3A). The source memory of eating chocolate pellets was manipulated by the experimenter placing the rat at the food trough of an arm that dispensed chocolate (an experimenter-generated event), whereas the rat encountered chocolate on its own at a food trough on a different arm (a self-generated event); these arms were randomly selected, and rats discovered chow-flavored pellets at two other randomly selected arms. After a retention interval, the rats discovered chow-flavored pellets at the other four arms in a memory assessment. The arm where the rat discovered chocolate on its own provided additional chocolate in the memory assessment (replenishment), whereas the arm where the rat was placed by the experimenter did not provide additional chocolate (nonreplenishment) in some experiments; in other experiments, the replenishment contingency was reversed. Chow-baited locations never replenished. Thus, the rat needed to remember the source of chocolate (self-generated vs. experimenter-generated information). Important to note, a single retention interval produced constant familiarity, which could not be used to predict replenishment. Rats revisited the replenishment location at a higher rate than the nonreplenishment location while avoiding revisits to chow locations (Figure 3B). These data are consistent with the hypothesis that rats remember the source of encoded information (Crystal & Alford, 2014; Crystal, Alford, et al., 2013; Crystal & Smith, 2014; Smith, Slivicki, et al., 2017; Smith et al., 2016). Moreover, source memory is quite long-lasting (Figure 3C), surviving retention-interval challenges of at least 1 week (Crystal & Alford, 2014; Crystal, Alford, et al., 2013; Crystal & Smith, 2014).

We used our source memory approach to test the hypothesis that rats remember episodic memories as bound representations (Crystal & Smith, 2014). The binding hypothesis proposes that the source memory for the event is stored with the remaining elements of the episodic event in an integrated manner. Another possibility is that memory consists only of unconnected features, which we refer to as the unbound-feature hypothesis. Binding functions to disambiguate similar episodes (i.e., episodes that share some, but not all, features) from one another.

Rats were presented with the opportunity to encode multiple features of an event, namely, what-where-source-context features: what (food flavor), where (maze location), source (self-generated or experimenter-generated food seeking), and context (spatial cues in the room where the event occurred). The what-where-source encoding occurred in one room, followed immediately by a second what-where-source encoding in a second room. After a retention interval, one flavor replenished at the self-generated location but not at the experimenter-generated location independently in a memory assessment in each room; the order of room presentations was randomly selected each day. For comparison, we assessed memory for one event (i.e., study and test in the same room). By increasing the memory load, we presented the rats with multiple overlapping features that can be fully disambiguated only by remembering that one study event occurred in one particular context (one room), whereas the other event occurred in a different context (another room). To produce potential interference, we used two identical radial mazes, with each arm pointing in the same orientation in two rooms that had similar geometric cues and a range of overlapping visual cues (Figure 4).

Binding multiple events into separate episodic memories would allow a rat to disambiguate similar events. Thus, bound representations of separate episodes predict successful performance with both memory loads. By contrast, the unbound-feature hypothesis predicts that retrieving information about two relatively similar events is expected to produce interference between events if at least some of the features overlap (see Figure 4).

The rats revisited the replenishing chocolate location in the memory assessment at a higher rate than the nonreplenishment chocolate location when we used a memory load of two rooms, at a level of proficiency similar to that observed when the memory load was one room (Crystal & Smith, 2014). Moreover, source-memory performance was resistant to interference from highly similar episodes and survived long retention intervals (about 1 week; Crystal & Smith, 2014). These results suggest that multiple episodic memories are each structured as bound representations.

Item-in-Context Memory

Because familiarity cues are pervasive (presentation of a stimulus always gives rise to a familiarity signal), it would be valuable to develop a method to dissociate familiarity and episodic memory solutions to a memory problem, a technique that we recently developed (Panoz-Brown et al., 2016). In our approach, rats were presented with odor-infused lids placed atop a container that could be baited with a reward. Novel odors were rewarded, whereas old (i.e., familiar) odors were not rewarded. To dissociate episodic memory from judgments of relative familiarity, we presented all of the odors in each of two distinctive contexts (arenas that differed in size, pattern, extra-arena cues, etc.). Important to note, rewards were given whenever the item (i.e., odor) was new to each context (Figure 5A). Rats could use episodic memory to remember the presentation of each item and the context in which it had been previously presented (Eichenbaum, 2007). Alternatively, the rats could choose new items by avoiding the familiar odors (i.e., a nonepisodic memory hypothesis). To dissociate item-in-context and familiarity hypotheses, we unexpectedly transitioned between the contexts (e.g., Context A → B → A). Critically, we identified sequences of odor presentations across the unexpected context transitions that predict above chance performance for item-in-context memory and below chance performance for selecting the least familiar item.

To dissociate episodic memory from familiarity judgments, we identified sequences of odors that put familiarity cues and item-in-context memory in conflict. For a particular pair of odors (e.g., strawberry and blueberry, depicted as red and blue in Figure 5A), we presented one item (strawberry) but not the other (blueberry) in the first context. Next, both items were presented in the second context (notably strawberry followed by blueberry). Finally, the memory assessment occurred upon return to the first context. In the memory assessment, the rats were confronted with a choice between strawberry and blueberry. Blueberry is the correct choice based on item in context because it has not yet been presented in the first context; indeed, blueberry is rewarded when chosen in this test, and our measure of accuracy is the proportion of choices of the rewarded item. Important to note, prior to the memory assessment, blueberry was presented more recently than strawberry (see Figure 5A). Because strawberry would be less familiar relative to blueberry in the memory assessment, an animal that relied on judgments of relative familiarity would choose the strawberry (i.e., following the rule “avoid familiar items”). By our measure of accuracy, such a choice would result in accuracy below chance. By contrast, an animal that relied on item-in-context memory would choose blueberry in the memory assessment, which would in turn result in above chance accuracy. Notably, this memory assessment dissociates item-in-context memory (above chance) from judgments of relative familiarity (below chance).

To test whether the rats were relying on item-in-context episodic memory or nonepisodic judgments of familiarity, we examined the rats’ accuracy in the initial memory assessments (i.e., before receiving feedback from rewards in the novel condition). When the identity of items in context was put in conflict with familiarity cues, initial performance was above chance (80% ± 6%; mean ± standard error of the mean [SEM]; chance = 50%) using 32 odors and context transitions that ranged from two (shown in Figure 5A) to 15; with each new number of context transitions, we re-created novel conditions because it was not possible for the rat to anticipate a new transition between contexts. High accuracy (Figure 5B) provides compelling evidence that rats relied on episodic item-in-context memory rather than judgments of familiarity.

Incidental Encoding and Unexpected Questions

One problem with many animal approaches to episodic memory is that training generates expectations, which may lead to memories of planned actions (Singer & Zentall, 2007; Zentall, 2005, 2006; Zentall, Clement, Bhatt, & Allen, 2001; Zentall, Singer, & Stagner, 2008). Zentall and colleagues have argued that when information is encoded for use in an expected memory test, explicitly encoded information may generate a planned action; thus, at the time of the test, the remembered action can occur successfully without remembering any earlier episodes. The central hypothesis of an animal model of episodic memory is that, at the moment of memory assessment, the animal remembers back in time to the event or episode (Crystal, 2013b, 2016a, 2016b); the focus on retrieving a memory of the earlier event is the key element that makes an animal model of episodic memory episodic. Therefore, carrying forward information that is needed at a future test while not specifically retrieving a memory of the earlier episode represents a serious threat to the episodic-memory hypothesis. Thus, it is necessary to rule out the hypothesis that accurate performance in the test is based on a planned action generated when information was explicitly encoded rather than a memory of the episode (Crystal, 2013b; Singer & Zentall, 2007; Zentall et al., 2001; Zentall et al., 2008; Zhou & Crystal, 2011). Notably, it is possible that animals may have solved previous tests of episodic memory by using learned semantic rules without remembering the episode. Formally, learned rules stored in semantic memory, a nonepisodic memory system devoted to storing generic facts (Tulving, 1993), could be used to generate a planned action. By contrast, when information is encoded incidentally, it is not possible to transform information into an action plan because the nature of the subsequent memory test is not yet known. Thus, accurate performance observed in an unexpected test after incidental encoding would suggest that this performance is based on memory of the earlier episode (i.e., retrieval of an episodic memory; Singer & Zentall, 2007; Zentall et al., 2001; Zentall et al., 2008; Zhou & Crystal, 2011). Zentall and colleagues (Singer & Zentall, 2007; Zentall et al., 2001; Zentall et al., 2008) have demonstrated that pigeons pass this episodic-memory test.

We used Zentall’s approach to test the hypothesis that rats can answer an unexpected question after incidental encoding (Zhou, Hohmann, & Crystal, 2012). To this end, we enabled incidental encoding by embedding two different tasks within the same radial maze (Figure 6A); a subset of arms were reserved for one task, and the other task used the remaining arms (shading in Figure 6A highlights the assignment of specific arms to the two different tasks, but all of the arms in the actual maze were white). In one task, the rats foraged for food (five-arm radial maze task) as in the standard eight-arm radial maze task (Olton & Samuelson, 1976), except only five arms were used. Three arms were randomly selected from the set of five arms to be baited with a food pellet in the study phase; next, five arms were accessible and an additional pellet was baited at each of the two arms not yet visited during the daily trial; the five arms shown in gray in Figure 6A were reserved for the five-arm task. In a second task, the rats learned the “reporting” skill (T-maze task) that would be used later in the unexpected question; the three arms shown in black in Figure 6A were reserved for the T-maze task. In the T-maze task, rats were rewarded for selecting a left/right turn after being presented with a sample of food or no food, respectively; one arm was designated as the sample arm where the animals obtained a food (six-pellet) or no-food (zero-pellet) sample after interrupting a photobeam in the sample arm (using the bottom arm shown in Figure 6A); next the two choice arms were available, and additional food (six pellets) could be obtained by a left turn or right turn (the rewarded turn was contingent on the identity of the sample—food vs. no food—and was counterbalanced across rats).

To assess the ability of rats to answer an unexpected question, we allowed rats to initially forage for food in the five-arm radial maze task (using the three top arms shown in Figure 6A), thereby affording the opportunity to incidentally encode either the presence (food probe) or the absence (no-food probe) of food. After the rat exited one of the top arms in Figure 6A, the rat was confronted with the opportunity to report in the T-maze task (via its left/right turn into an arm shown as black in Figure 6A) whether it remembered encountering the presence or absence of food in the five-arm task; the uninterrupted transition from the five-arm task foraging to a T-maze choice phase was possible because the two tasks were embedded in the same radial maze. A rat that incidentally encoded the availability of food would be able to successfully answer an unexpected question by retrieving a memory of the earlier episode. By contrast, a rat without episodic memory is expected to be unable to answer an unexpected question after incidental encoding; therefore, the probability of left and right turns is expected to be equally likely in the absence of episodic memory. The rats answered the unexpected question with a level of accuracy similar to that observed in T-maze training (Figure 6B).

To test the hypothesis that answering an unexpected question requires episodic memory, we asked whether it is hippocampal dependent; extensive evidence suggests that the hippocampus is a critical processing center for episodic memory (e.g., (Eichenbaum, 2000, 2017; Eichenbaum, Yonelinas, & Ranganath, 2007; Nyberg et al., 1996). If answering an unexpected question after incidental encoding requires episodic memory, then temporary inactivation of the hippocampus should selectively impair the ability of rats to answer an unexpected question without impacting the ability to answer an expected question. To assess accuracy in answering an unexpected question, we used a no-food probe. To assess accuracy in answering an expected question, we used a control procedure that combined elements of the T-maze task while equating other features of the no-food probe; we referred to this control condition as a rotation probe. As in the T-maze task (but unlike the no-food probe), the rotation probe presented a no-food sample followed immediately by the opportunity to turn left or right. Thus, this control procedure can be solved by remembering a planned action without remembering the episode; because the rotation probe can be solved without remembering the episode, we expect that performance on the rotation probe will not be impaired by temporary inactivation of the hippocampus. To equate the control procedure with other aspects of the no-food probe, the rotation probe offered a no-food sample, and the sample was presented in the arm opposite to that used in training (i.e., rotated 180o with respect to the usual T-maze sample location, using the top-center arm shown in Figure 6A); this rotation is equivalent to the average rotation in the no-food probe. Thus, the no-food and rotation probes varied the episodic-memory demands while equating rotation and absence of food.

Next we surgically implanted cannulae bilaterally aimed at the CA3 region of the hippocampus to temporarily inactive this region with lidocaine. Accuracy was reestablished following surgical recovery, demonstrating that surgery did not disrupt performance. Following local infusion of lidocaine bilaterally into CA3, accuracy in answering the unexpected question was significantly reduced relative to baseline (Figure 6C), whereas accuracy in answering the expected question was not impaired. The selective reduction of accuracy on unexpected questions could be attributed to effects of lidocaine infusion, because accuracy was not impaired relative to baseline by infusions of vehicle (Figure 6C).

In summary, the suppressive effect of lidocaine on memory was selective for unexpected questions. Accuracy was significantly reduced in the unexpected—relative to expected—question conditions following lidocaine infusion. Important to note, impairment in answering the unexpected question was selective to inactivation of the hippocampus with lidocaine when an episodic memory needed to be retrieved.

Comparative Studies of Episodic Memory

Comparisons across species and appreciation of adaptive specializations of memory are at the heart of the earliest efforts to study episodic memory in animals. Of course, Clayton and colleagues’ use of scrub jays capitalized on the extraordinary memory and cognition profile of this food-storing corvid. Moreover, a range of cognitive abilities in corvids (including episodic memory) has led to a prominent proposal that corvids are “feathered apes”—the notion that corvids and apes evolved cognitive abilities along convergent evolution (Emery & Clayton, 2004). Next I summarize a range of comparative studies of episodic memory.

Episodic memory has been investigated in a number of nonhuman primates, including monkeys (Basile & Hampton, 2017; Devkar & Wright, 2016; Hampton, Hampstead, & Murray, 2005; Hampton & Schwartz, 2004; Hoffman, Beran, & Washburn, 2009) and apes (Kano & Hirata, 2015; Martin-Ordas, Berntsen, & Call, 2013; Menzel, 1999; Schwartz, Colon, Sanchez, Rodriguez, & Evans, 2002; Schwartz & Evans, 2001; Schwartz, Hoffman, & Evans, 2005). In addition, episodic memory has been investigated in invertebrates, including cuttlefish and honeybees (Jozet-Alves, Bertin, & Clayton, 2013; Pahl, Zhu, Pix, Tautz, & Zhang, 2007; Zhang, Schwarz, Pahl, Zhu, & Tautz, 2006). The approaches described in the preceding sections with rat subjects may be used in comparative studies in future research.

Conclusions

A number of approaches to test the central hypothesis of episodic memory were described. Compelling evidence for episodic memory is obtained when judgments of familiarity cannot produce accurate choices in memory assessments (equating familiarity cues in what-where-when and source memory; dissociating episodic memory and familiarity in item-in-context memory). These techniques may be used to explore the evolution of cognition in other species. This review focused on recent advances in developing animal models of episodic memory. Notably, understanding the elements of cognition in animals may also be advanced by developing a range of other memory models, such as prospective memory (Beran, Evans, Klein, & Einstein, 2012; Beran, Perdue, Bramlett, Menzel, & Evans, 2012; Crystal, 2013a; Crystal & Wilson, 2015; Wilson & Crystal, 2012; Wilson, Pizzo, & Crystal, 2013), retrieval practice (Crystal, Ketzenberger, & Alford, 2013), and working memory (Bratch et al., 2016; Roberts, Guitar, Marsh, & ­MacDonald, 2016).

A number of lines of evidence suggest that animals provide a valuable model of episodic memory. Episodic memory involves a range of elements, rather than a single defining feature (Tulving, 1983). Thus, the multiple approaches reviewed here may be used to identify the elements of episodic memory in animals. Some animals may have a wide range of elements of episodic memory that correspond to human episodic memory, whereas other animals may be found to have some elements but not other elements (Crystal, 2009). Exploration of a wide range of elements of episodic memory may be used in comparative studies to test hypothesis about the evolution of episodic memory (Allen & Fortin, 2013; Emery & Clayton, 2004). In addition, the development of rodent models of episodic memory can be combined with well-developed animal models of human diseases that afflict memory, such as Alzheimer’s disease. Notably, research with genetic models of Alzheimer’s disease typically use simple measures of learning and memory as behavioral endpoints (O’Leary & Brown, 2008; Palop et al., 2003; Pennanen, Wolfer, Nitsch, & Götz, 2006; Roberson et al., 2007; Stepanichev, Zdobnova, Zarubenko, Lazareva, & Gulyaeva, 2006; Stepanichev et al., 2004; Timmer et al., 2008; Yates et al., 2008). It is more appropriate to use measures of episodic memory in these models because the deterioration of episodic memory is the most debilitating aspect of Alzheimer’s disease (Bäckman et al., 1999; Butters, Granholm, Salmon, Grant, & Wolfe, 1987; Egerhazi, Berecz, Bartok, & Degrell, 2007; Le Moal et al., 1997; Liscic, Storandt, Cairns, & Morris, 2007). Clinical trials for Alzheimer’s therapeutics have consistently failed despite being preceded by promising preclinical studies (Becker & Greig, 2010; Carlsson, 2008; Jacobson & Sabbagh, 2011; Mangialasche, Solomon, Winblad, Mecocci, & Kivipelto, 2010; Mullane & Williams, 2013; Schneider & Lahiri, 2009). Important to note, the advent of new gene editing technologies such as CRISPR (Cong et al., 2013; Wang et al., 2013) has increased the availability of genetic models of diseases in rats. Integration of advanced assessments of episodic memory with cutting-edge genetic models of Alzheimer’s disease and assessments of neuropathology may represent a key ingredient to promote successful translation.

References

1. Allen, T. A., & Fortin, N. J. (2013). The evolution of episodic memory. Proceedings of the National Academy of Sciences, 110(Suppl. 2), 10379–10386. doi:10.1073/pnas.1301199110

2. Babb, S. J., & Crystal, J. D. (2005). Discrimination of what, when, and where: Implications for episodic-like memory in rats. Learning & Motivation, 36, 177–189. doi:10.1016/j.lmot.2005.02.009

3. Babb, S. J., & Crystal, J. D. (2006a). Discrimination of what, when, and where is not based on time of day. Learning & Behavior, 34, 124–130. doi:10.3758/BF03193188

4. Babb, S. J., & Crystal, J. D. (2006b). Episodic-like memory in the rat. Current Biology, 16, 1317–1321. doi:10.1016/j.cub.2006.05.025

5. Bäckman, L., Andersson, J. L., Nyberg, L., Winblad, B., Nordberg, A., & Almkvist, O. (1999). Brain regions associated with episodic retrieval in normal aging and Alzheimer’s disease. Neurology, 52, 1861–1870. doi:10.1212/WNL.52.9.1861

6. Basile, B. M., & Hampton, R. R. (2017). Dissociation of item and source memory in rhesus monkeys. Cognition, 166, 398–406. doi:10.1016/j.cognition.2017.06.009

7. Becker, R. E., & Greig, N. H. (2010). Why so few drugs for Alzheimer’s disease? Are methods failing drugs? Current Alzheimer Research, 7, 642–651. doi:10.2174/156720510793499075

8. Beran, M. J., Evans, T. A., Klein, E. D., & Einstein, G. O. (2012). Rhesus monkeys (Macaca mulatta) and capuchin monkeys (Cebus apella) remember future responses in a computerized task. Journal of Experimental Psychology: Animal Behavior Processes, 38, 233–243. doi:10.1037/a0027796

9. Beran, M. J., Perdue, B. M., Bramlett, J. L., Menzel, C. R., & Evans, T. A. (2012). Prospective memory in a language-trained chimpanzee (Pan troglodytes). Learning and Motivation, 43(4), 192–199. doi:10.1016/j.lmot.2012.05.002

10. Bratch, A., Kann, S., Cain, J. A., Wu, J.-E., Rivera-Reyes, N., Dalecki, S., … Crystal, J. D. (2016). Working memory systems in the rat. Current Biology, 26, 351–355. doi:10.1016/j.cub.2015.11.068

11. Butters, N., Granholm, E., Salmon, D. P., Grant, I., & Wolfe, J. (1987). Episodic and semantic memory: A comparison of amnesic and demented patients. Journal of Clinical and Experimental Neuropsychology, 9, 479–497. doi:10.1080/01688638708410764

12. Carlsson, C. M. (2008). Lessons learned from failed and discontinued clinical trials for the treatment of Alzheimer’s disease: Future directions. Journal of Alzheimer’s Disease, 15, 327–338. doi:10.3233/JAD-2008-15214

13. Clayton, N. S., Bussey, T. J., & Dickinson, A. (2003). Can animals recall the past and plan for the future? Nature Reviews Neuroscience, 4(8), 685–691. doi:10.1038/nrn1180

14. Clayton, N. S., Bussey, T. J., Emery, N. J., & Dickinson, A. (2003). Prometheus to Proust: The case for behavioural criteria for “mental time travel.” Trends in Cognitive Sciences, 7(10), 436–437. doi:10.1016/j.tics.2003.08.003

15. Clayton, N. S., & Dickinson, A. (1998). Episodic-like memory during cache recovery by scrub jays. Nature, 395(6699), 272–274. doi:10.1038/26216

16. Clayton, N. S., & Emery, N. J. (2015). Avian models for human cognitive neuroscience: A proposal. Neuron, 86(6), 1330–1342. doi:10.1016/j.neuron.2015.04.024

17. Clayton, N. S., Salwiczek, L. H., & Dickinson, A. (2007). Episodic memory. Current Biology, 17(6), R189–R191. doi:10.1016/j.cub.2007.01.011

18. Cong, L., Ran, F. A., Cox, D., Lin, S., Barretto, R., Habib, N., … Zhang, F. (2013). Multiplex genome engineering using CRISPR/Cas systems. Science, 339(6121), 819–823. doi:10.1126/science.1231143

19. Crystal, J. D. (2009). Elements of episodic-like memory in animal models. Behavioural Processes, 80, 269–277. doi:10.1016/j.beproc.2008.09.009

20. Crystal, J. D. (2012). Sensitivity to time: Implications for the representation of time. In E. A. Wasserman & T. R. Zentall (Eds.), The Oxford handbook of comparative cognition (pp. 434–450). New York, NY: Oxford University Press.

21. Crystal, J. D. (2013a). Prospective memory. Current Biology, 23(17), R750–751. doi:10.1016/j.cub.2013.07.081

22. Crystal, J. D. (2013b). Remembering the past and planning for the future in rats. Behavioural Processes, 93, 39–49. doi:10.1016/j.beproc.2012.11.014

23. Crystal, J. D. (2016a). Animal models of source memory. Journal of the Experimental Analysis of Behavior, 105, 56–67. doi:10.1002/jeab.173

24. Crystal, J. D. (2016b). Comparative cognition: Action imitation using episodic memory. Current Biology, 26(23), R1226–R1228. doi:10.1016/j.cub.2016.10.010

25. Crystal, J. D., & Alford, W. T. (2014). Validation of a rodent model of source memory. Biology Letters, 10(3), 20140064. doi:10.1098/rsbl.2014.0064

26. Crystal, J. D., Alford, W. T., Zhou, W., & Hohmann, A. G. (2013). Source memory in the rat. Current Biology, 23(5), 387–391. doi:10.1016/j.cub.2013.01.023

27. Crystal, J. D., Ketzenberger, J. A., & Alford, W. T. (2013). Practicing memory retrieval improves long-term retention in rats. Current Biology, 23(17), R708–709. doi:10.1016/j.cub.2013.07.044

28. Crystal, J. D., & Smith, A. E. (2014). Binding of episodic memories in the rat. Current Biology, 24, 2957–2961. doi:10.1016/j.cub.2014.10.074

29. Crystal, J. D., & Wilson, A. G. (2015). Prospective memory: A comparative perspective. Behavioural Processes, 112, 88–99. doi:10.1016/j.beproc.2014.07.016

30. Dere, E., Dere, D., de Souza Silva, M. A., Huston, J. P., & Zlomuzica, A. (2017). Fellow travellers: Working memory and mental time travel in rodents. Behavioural Brain Research. Advance online publication. doi:10.1016/j.bbr.2017.03.026

31. Dere, E., Huston, J. P., & De Souza Silva, M. A. (2005a). Episodic-like memory in mice: Simultaneous assessment of object, place and temporal order memory. Brain Research Protocols, 16, 10–19. doi:10.1016/j.brainresprot.2005.08.001

32. Dere, E., Huston, J. P., & De Souza Silva, M. A. (2005b). Integrated memory for objects, places, and temporal order: Evidence for episodic-like memory in mice. Neurobiology of Learning and Memory, 84, 214–221. doi:10.1016/j.nlm.2005.07.002

33. Devkar, D. T., & Wright, A. A. (2016). Event-based proactive interference in rhesus monkeys. Psychonomic Bulletin & Review, 23, 1474–1482. doi:10.3758/s13423-016-1005-x

34. Eacott, M. J., & Easton, A. (2010). Episodic memory in animals: Remembering which occasion. Neuropsychologia, 48, 2273–2280. doi:10.1016/j.neuropsychologia.2009.11.002

35. Eacott, M. J., Easton, A., & Zinkivskay, A. (2005). Recollection in an episodic-like memory task in the rat. Learning & Memory, 12, 221–223. doi:10.1101/lm.92505

36. Eacott, M. J., & Norman, G. (2004). Integrated memory for object, place, and context in rats: A possible model of episodic-like memory? The Journal of Neuroscience, 24, 1948–1953. doi:10.1523/JNEUROSCI.2975-03.2004

37. Egerhazi, A., Berecz, R., Bartok, E., & Degrell, I. (2007). Automated Neuropsychological Test Battery (CANTAB) in mild cognitive impairment and in Alzheimer’s disease. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 31, 746–751. doi:10.1016/j.pnpbp.2007.01.011

38. Eichenbaum, H. (2000). A cortical-hippocampal system for declarative memory. Nature Reviews Neuroscience, 1(1), 41–50. doi:10.1038/35036213

39. Eichenbaum, H. (2007). Comparative cognition, hippocampal function, and recollection. Comparative Cognition & Behavior Reviews, 2, 47–66. doi:10.3819/ccbr.2008.20003

40. Eichenbaum, H. (2017). Prefrontal-hippocampal interactions in episodic memory. Nature Reviews Neuroscience, 18(9), 547–558. doi:10.1038/nrn.2017.74

41. Eichenbaum, H., Yonelinas, A. P., & Ranganath, C. (2007). The medial temporal lobe and recognition memory. Annual Review of Neuroscience, 30, 123–152. doi:10.1146/annurev.neuro.30.051606.094328

42. Emery, N. J., & Clayton, N. S. (2004). The mentality of crows: Convergent evolution of intelligence in corvids and apes. Science, 306, 1903–1907. doi:10.1126/science.1098410

43. Ergorul, C., & Eichenbaum, H. (2004). The hippocampus and memory for “what,” “where,” and “when.” Learning & Memory, 11, 397–405. doi:10.1101/lm.73304

44. Fortin, N. J., Wright, S. P., & Eichenbaum, H. (2004). Recollection-like memory retrieval in rats is dependent on the hippocampus. Nature, 431(7005), 188–191. doi:10.1038/nature02853

45. Gallistel, C. R. (1990). The organization of learning. Cambridge, MA: MIT Press.

46. Gallistel, C. R. (2009). The importance of proving the null. Psychological Review, 116, 439–453. doi:10.1037/a0015251

47. Griffiths, D., Dickinson, A., & Clayton, N. S. (1999). Episodic memory: What can animals remember about their past? Trends in Cognitive Sciences, 3(2), 74–80. doi:10.1016/S1364-6613(98)01272-8

48. Hampton, R. R., Hampstead, B. M., & Murray, E. A. (2005). Rhesus monkeys (Macaca mulatta) demonstrate robust memory for what and where, but not when, in an open-field test of memory. Learning & Motivation, 36, 245–259. doi:10.1016/j.lmot.2005.02.004

49. Hampton, R. R., & Schwartz, B. (2004). Episodic memory in nonhumans: What, and where, is when? Current Opinion in Neurobiology, 14, 192–197. doi:10.1016/j.conb.2004.03.006

50. Hoffman, M. L., Beran, M. J., & Washburn, D. A. (2009). Memory for “what,” “where,” and “when” information in rhesus monkeys (Macaca mulatta). Journal of Experimental Psychology: Animal Behavior Processes, 35, 143–152. doi:10.1037/a0013295

51. Hunsaker, M. R., Lee, B., & Kesner, R. P. (2008). Evaluating the temporal context of episodic memory: The role of CA3 and CA1. Behavioural Brain Research, 188, 310–315. doi:10.1016/j.bbr.2007.11.015

52. Iordanova, M. D., Burnett, D. J., Aggleton, J. P., Good, M., & Honey, R. C. (2009). The role of the hippocampus in mnemonic integration and retrieval: complementary evidence from lesion and inactivation studies. European Journal of Neuroscience, 30, 2177–2189. doi:10.1111/j.1460-9568.2009.07010.x

53. Iordanova, M. D., Burnett, D. J., Good, M., & Honey, R. C. (2011). Pattern memory involves both elemental and configural processes: Evidence from the effects of hippocampal lesions. Behavioral Neuroscience, 125, 567–577. doi:10.1037/a0023762

54. Iordanova, M. D., Good, M. A., & Honey, R. C. (2008). Configural learning without reinforcement: Integrated memories for correlates of what, where, and when. Quarterly Journal of Experimental Psychology, 61, 1785–1792. doi:10.1080/17470210802194324

55. Jacobson, S. A., & Sabbagh, M. N. (2011). Investigational drugs for the treatment of AD: What can we learn from negative trials? Alzheimer’s Research & Therapy, 3, 14. doi:10.1186/alzrt73

56. Janowsky, J. S., Shimamura, A. P., & Squire, L. R. (1989). Source memory impairment in patients with frontal lobe lesions. Neuropsychologia, 27, 1043–1056. doi:10.1016/0028-3932(89)90184-X

57. Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114, 3–28. doi:10.1037/0033-2909.114.1.3

58. Jozet-Alves, C., Bertin, M., & Clayton, N. S. (2013). Evidence of episodic-like memory in cuttlefish. Current Biology, 23(23), R1033–R1035. doi:10.1016/j.cub.2013.10.021

59. Kano, F., & Hirata, S. (2015). Great apes make anticipatory looks based on long-term memory of single events. Current Biology, 25, 2513–2517. doi:10.1016/j.cub.2015.08.004

60. Kart-Teke, E., De Souza Silva, M. A., Huston, J. P., & Dere, E. (2006). Wistar rats show episodic-like memory for unique experiences. Neurobiology of Learning and Memory, 85, 173–182. doi:10.1016/j.nlm.2005.10.002

61. Kesner, R. P., & Hunsaker, M. R. (2010). The temporal attributes of episodic memory. Behavioural Brain Research, 215, 299–309. doi:10.1016/j.bbr.2009.12.029

62. Kesner, R. P., Hunsaker, M. R., & Warthen, M. W. (2008). The CA3 subregion of the hippocampus is critical for episodic memory processing by means of relational encoding in rats. Behavioral Neuroscience, 122, 1217–1225. doi:10.1037/a0013592

63. Kheifets, A., Freestone, D., & Gallistel, C. R. (2017). Theoretical implications of quantitative properties of interval timing and probability estimation in mouse and rat. Journal of the Experimental Analysis of Behavior, 108, 39–72. doi:10.1002/jeab.261

64. Le Moal, S., Reymann, J. M., Thomas, V., Cattenoz, C., Lieury, A., & Allain, H. (1997). Effect of normal aging and of Alzheimer’s disease on episodic memory. Dementia And Geriatric Cognitive Disorders, 8, 281–287. doi:10.1159/000106645

65. Liscic, R. M., Storandt, M., Cairns, N. J., & Morris, J. C. (2007). Clinical and psychometric distinction of frontotemporal and Alzheimer dementias. Archives of Neurology, 64, 535–540. doi:10.1001/archneur.64.4.535

66. Mangialasche, F., Solomon, A., Winblad, B., Mecocci, P., & Kivipelto, M. (2010). Alzheimer’s disease: Clinical trials and drug development. Lancet Neurology, 9, 702–716. doi:10.1016/S1474-4422(10)70119-8

67. Martin-Ordas, G., Berntsen, D., & Call, J. (2013). Memory for distant past events in chimpanzees and orangutans. Current Biology: CB, 23, 1438–1441. doi:10.1016/j.cub.2013.06.017

68. Menzel, C. R. (1999). Unprompted recall and reporting of hidden objects by a chimpanzee (Pan troglodytes) after extended delays. Journal of Comparative Psychology, 113(4), 426–434. doi:10.1037/0735-7036.113.4.426

69. Mitchell, K. J., & Johnson, M. K. (2009). Source monitoring 15 years later: What have we learned from fMRI about the neural mechanisms of source memory? Psychological Bulletin, 135, 638–677. doi:10.1037/a0015849

70. Mullane, K., & Williams, M. (2013). Alzheimer’s therapeutics: Continued clinical failures question the validity of the amyloid hypothesis—but what lies beyond? Biochemical Pharmacology, 85, 289–305. doi:10.1016/j.bcp.2012.11.014

71. Naqshbandi, M., Feeney, M. C., McKenzie, T. L. B., & Roberts, W. A. (2007). Testing for episodic-like memory in rats in the absence of time of day cues: Replication of Babb and Crystal. Behavioural Processes, 74, 217–225. doi:10.1016/j.beproc.2006.10.010

72. Nyberg, L., McIntosh, A., Cabeza, R., Habib, R., Houle, S., & Tulving, E. (1996). General and specific brain regions involved in encoding and retrieval of events: What, where, and when. Proceedings of the National Academy of Sciences of the United States of America, 93, 11280–11285. doi:10.1073/pnas.93.20.11280

73. O’Leary, T. P., & Brown, R. E. (2008). P1–117: Visuo-spatial learning and memory in an aged double transgenic mouse model of Alzheimer’s disease. Alzheimer’s and Dementia, 4(4, Suppl. 1), T241–T241. doi:10.1016/j.jalz.2008.05.704

74. Olton, D. S., & Samuelson, R. J. (1976). Remembrance of places passed: Spatial memory in rats. Journal of Experimental Psychology: Animal Behavior Processes, 2, 97–116. doi:10.1037/0097-7403.2.2.97

75. Pahl, M., Zhu, H., Pix, W., Tautz, J., & Zhang, S. (2007). Circadian timed episodic-like memory—A bee knows what to do when, and also where. Journal of Experimental Biology, 210(Pt. 20), 3559–3567. doi:10.1242/jeb.005488

76. Palop, J. J., Jones, B., Kekonius, L., Chin, J., Yu, G.-Q., Raber, J., … Mucke, L. (2003). Neuronal depletion of calcium-dependent proteins in the dentate gyrus is tightly linked to Alzheimer’s disease-related cognitive deficits. Proceedings of the National Academy of Sciences of the United States of America, 100, 9572–9577. doi:10.1073/pnas.1133381100

77. Panoz-Brown, D. E., Corbin, H. E., Dalecki, S. J., Gentry, M., Brotheridge, S., Sluka, C. M., Wu, J.-E., & Crystal, J. D. (2016). Rats remember items in context using episodic memory. Current Biology, 26, 2821–2826. doi:10.1016/j.cub.2016.08.023

78. Pennanen, L., Wolfer, D. P., Nitsch, R. M., & Götz, J. (2006). Impaired spatial reference memory and increased exploratory behavior in P301L tau transgenic mice. Genes, Brain & Behavior, 5, 369–379. doi:10.1111/j.1601-183X.2005.00165.x

79. Roberson, E. D., Scearce-Levie, K., Palop, J. J., Yan, F., Cheng, I. H., Wu, T., … Mucke, L. (2007). Reducing endogenous tau ameliorates amyloid β-induced deficits in an Alzeimer’s disease mouse model. Science, 316(5825), 750–754. doi:10.1126/science.1141736

80. Roberts, W. A., Feeney, M. C., MacPherson, K., Petter, M., McMillan, N., & Musolino, E. (2008). Episodic-like memory in rats: Is it based on when or how long ago? Science, 320(5872), 113–115. doi:10.1126/science.1152709

81. Roberts, W. A., Guitar, N. A., Marsh, H. L., & MacDonald, H. (2016). Memory systems in the rat: effects of reward probability, context, and congruency between working and reference memory. Animal Cognition, 19, 593–604. doi:10.1007/s10071-016-0964-7

82. Rouder, J., Speckman, P., Sun, D., Morey, R., & Iverson, G. (2009). Bayesian t tests for accepting and rejecting the null hypothesis. Psychonomic Bulletin & Review, 16, 225–237. doi:10.3758/pbr.16.2.225

83. Schneider, L. S., & Lahiri, D. K. (2009). The perils of Alzheimer’s drug development. Current Alzheimer Research, 6, 77–78. doi:10.2174/156720509787313871

84. Schwartz, B. L., Colon, M. R., Sanchez, I. C., Rodriguez, I., & Evans, S. (2002). Single-trial learning of “what” and “who” information in a gorilla (Gorilla gorilla gorilla): implications for episodic memory. Animal Cognition, 5, 85–90. doi:10.1007/s10071-002-0132-0

85. Schwartz, B. L., & Evans, S. (2001). Episodic memory in primates. American Journal of Primatology, 55, 71–85. doi:10.1002/ajp.1041

86. Schwartz, B. L., Hoffman, M. L., & Evans, S. (2005). Episodic-like memory in a gorilla: A review and new findings. Learning and Motivation, 36, 226–244. doi:10.1016/j.lmot.2005.02.012

87. Singer, R. A., & Zentall, T. R. (2007). Pigeons learn to answer the question “Where did you just peck?” and can report peck location when unexpectedly asked. Learning & Behavior, 35, 184–189. doi:10.3758/BF03193054

88. Smith, A. E., Dalecki, S. J., & Crystal, J. D. (2017). A test of the reward-value hypothesis. Animal Cognition, 20, 215–220. doi:10.1007/s10071-016-1040-z

89. Smith, A. E., Slivicki, R. A., Hohmann, A. G., & Crystal, J. D. (2017). The chemotherapeutic agent paclitaxel selectively impairs learning while sparing source memory and spatial memory. Behavioural Brain Research, 320, 48–57. doi:10.1016/j.bbr.2016.11.042

90. Smith, A. E., Xu, Z., Lai, Y. Y., Kulkarni, P. M., Thakur, G. A., Hohmann, A. G., & Crystal, J. D. (2016). Source memory in rats is impaired by an NMDA receptor antagonist but not by PSD95-nNOS protein–protein interaction inhibitors. Behavioural Brain Research, 305, 23–29. doi:10.1016/j.bbr.2016.02.021

91. Stepanichev, M. Y., Zdobnova, I., Zarubenko, I., Lazareva, N., & Gulyaeva, N. V. (2006). Differential effects of tumor necrosis factor-alpha co-administered with amyloid β-peptide (25-35) on memory function and hippocampal damage in rat. Behavioural Brain Research, 175, 352–361. doi:10.1016/j.bbr.2006.09.006

92. Stepanichev, M. Y., Zdobnova, I. M., Zarubenko, I. I., Moiseeva, Y. V., Lazareva, N. A., Onufriev, M. V., & Gulyaeva, N. V. (2004). Amyloid-β(25-35)-induced memory impairments correlate with cell loss in rat hippocampus. Physiology & Behavior, 80, 647–655. doi:10.1016/j.physbeh.2003.11.003

93. Timmer, N., van Dijk, L., Bergervoet, N., Kiliaan, A., van der Zee, I., de Waal, R., & Verbeek, M. (2008). P4-312: Treatment of APPswe/PS1dE9 mice with Enoxaparin improves cognitive impairment. Alzheimer’s and Dementia, 4(4, Suppl. 1), T763–T763. doi:10.1016/j.jalz.2008.05.2382

94. Tulving, E. (1983). Elements of episodic memory. New York, NY: Oxford University Press.

95. Tulving, E. (1985). Memory and consciousness. Canadian Psychology/Psychologie canadienne, 26, 1–12. doi:10.1037/h0080017

96. Tulving, E. (1987). Multiple memory systems and consciousness. Human Neurobiology, 6, 67–80.

97. Tulving, E. (1993). What is episodic memory? Current Directions in Psychological Science, 2, 67–70. doi:10.1111/1467-8721.ep10770899

98. Wang, H., Yang, H., Shivalila, C. S., Dawlaty, M. M., Cheng, A. W., Zhang, F., & Jaenisch, R. (2013). One-step generation of mice carrying mutations in multiple genes by CRISPR/cas-mediated genome engineering. Cell, 153, 910–918. doi:10.1016/j.cell.2013.04.025

99. Wilson, A. G., & Crystal, J. D. (2012). Prospective memory in the rat. Animal Cognition, 15(3), 349–358. doi:10.1007/s10071-011-0459-5

100. Wilson, A. G., Pizzo, M. J., & Crystal, J. D. (2013). Event-based prospective memory in the rat. Current Biology, 23, 1089–1093. doi:10.1016/j.cub.2013.04.067

101. Wright, A. A. (2007). An experimental analysis of memory processing. Journal of the Experimental Analysis of Behavior, 88, 405–433. doi:10.1901/jeab.2007.88-405

102. Yates, D. M., Mitchell, J. C., Ariff, B., Stephenson, J., Miller, C. C. J., & McLoughlin, D. M. (2008). P1–058: The neuronal X11[alpha] adaptor protein improves cognitive function of Tg2576 (APPswe) mice. Alzheimer’s and Dementia, 4(4, Suppl. 1), T222–T223. doi:10.1016/j.jalz.2008.05.644

103. Zentall, T. R. (2005). Animals may not be stuck in time. Learning & Motivation, 36, 208–225. doi:10.1016/j.lmot.2005.03.001

104. Zentall, T. R. (2006). Mental time travel in animals: A challenging question. Behavioural Processes, 72, 173–183. doi:10.1016/j.beproc.2006.01.009

105. Zentall, T. R., Clement, T. S., Bhatt, R. S., & Allen, J. (2001). Episodic-like memory in pigeons. Psychonomic Bulletin & Review, 8, 685–690. doi:10.3758/BF03196204

106. Zentall, T. R., Singer, R. A., & Stagner, J. P. (2008). Episodic-like memory: Pigeons can report location pecked when unexpectedly asked. Behavioural Processes, 79, 93–98. doi:10.1016/j.beproc.2008.05.003

107. Zhang, S., Schwarz, S., Pahl, M., Zhu, H., & Tautz, J. (2006). Honeybee memory: A honeybee knows what to do and when. Journal of Experimental Biology, 209, 4420–4428. doi:10.1242/jeb.02522

108. Zhou, W., & Crystal, J. D. (2009). Evidence for remembering when events occurred in a rodent model of episodic memory. Proceedings of the National Academy of Sciences of the United States of America, 106, 9525–9529. doi:10.1073/pnas.0904360106

109. Zhou, W., & Crystal, J. D. (2011). Validation of a rodent model of episodic memory. Animal Cognition, 14, 325–340. doi:10.1007/s10071-010-0367-0

110. Zhou, W., Hohmann, A. G., & Crystal, J. D. (2012). Rats answer an unexpected question after incidental encoding. Current Biology, 22, 1149–1153. doi:10.1016/j.cub.2012.04.040

Ingredients for Understanding Brain and Behavioral Evolution: Ecology, Phylogeny, and Mechanism

Stephen H. Montgomery*

Department of Zoology, University of Cambridge

Adrian Currie*

Center for the Study of Existential Risk, University of Cambridge

Dieter Lukas

Department of Zoology, University of Cambridge

Neeltje Boogert

Centre for Ecology and Conservation, University of Exeter

Andrew Buskell

Department of History and Philosophy of Science, University of Cambridge

Fiona R. Cross

School of Biological Sciences, University of Canterbury

International Centre of Insect Physiology and Ecology

Sarah Jelbert

Department of Psychology, University of Cambridge

Shahar Avin

Center for the Study of Existential Risk, University of Cambridge

Rafael Mares

Department of Anthropology, University of California, Davis
Smithsonian Tropical Research Institute

Ana F. Navarrete

Centre for Biodiversity, University of St Andrews

Shuichi Shigeno

Department of Biology and Evolution of Marine Organisms, Stazione Zoologica Anton Dohrn

Corina J. Logan

Department of Zoology, University of Cambridge

Reading Options

---

Abstract

Uncovering the neural correlates and evolutionary drivers of behavioral and cognitive traits has been held back by traditional perspectives on which correlations to look for—in particular, anthropocentric conceptions of cognition and coarse-grained brain measurements. We welcome our colleagues’ comments on our overview of the field and their suggestions for how to move forward. Here, we counter, clarify, and extend some points, focusing on the merits of looking for the “best” predictor of cognitive ability, the sources and meaning of “noise,” and the ways in which we can deduce and test meaningful conclusions from comparative analyses of complex traits.

Keywords: brain measures, cognition, behavior, noise

Author Note: Corina J. Logan, Department of Zoology, University of Cambridge, Downing Street, Cambridge, CB2 3EJ, United Kingdom.

Correspondence concerning this article should be addressed to Corina J. Logan at cl417@cam.ac.uk.

Acknowledgments: Stephen H. Montgomery and Adrian Currie share co–first authorship.

---

Response

With some notable exceptions, the study of brain–behavior relationships in a comparative, phylogenetic context has been marked by (a) anthropocentrism regarding which behaviors are regarded as important; (b) a focus on coarse-grained neuroanatomical traits, most notably brain size; and (c) a reliance on somewhat slippery notions with contentious definitions like “intelligence” and “cognition.” In our review (Logan et al., 2018), we highlight that brains and behavior are variable, both within species and across taxa. This heterogeneity undermines the use of coarse-grained, anthropocentric measures. Instead, we argue that correlations between neural and behavioral traits in cross-taxa contexts should be tackled using a two-pronged strategy that combines the power of comparative analyses to detect generalizable evolutionary trends, with the depth of understanding provided by detailed studies of ecologically relevant traits in particular species.

Herculano-Houzel (2018) and Serpico and Frasnelli (2018) provide thoughtful commentaries that push our arguments in differing directions. First, Herculano-Houzel (2018) emphasizes the importance of structural variation among brains of different species, particularly in neural number, but questions the two-pronged approach we recommend. Second, Serpico and Frasnelli suggest that we may have been too hasty in dismissing coarse-grained, anthropocentric measures and present a case study in which they argue that such measures are successful. In this response, we wish to emphasize a few points of difference in opinion, clarify our view, and extend the new ideas the commentators have raised.

The studies that interest us establish and explore correlations between neural properties, on one hand, and behavioral or cognitive traits, on the other. We argue that to properly understand their relationships and dependencies, both sides should be considered in finer grained detail. The black box of “brain size” should be opened up to more specific brain structures and neuronal measures, and behavior should be ecologically meaningful and quantifiable. Herculano-Houzel (2018) focuses on the former point, providing a lucid description of the importance of neuron number. Her view is informed by what she calls “embrained cognitive evolution” (p. 93): the idea that cognitive evolution should be understood in light of within-brain neural structures. She contrasts embrained cognitive evolution with “embodied cognition,” the idea that if we are to truly understand cognition we must not restrict our notions of cognition or investigations of cognitive operations to the vault of the skull. Where the latter asks us to extend cognition to include bodily processes, the former asks us to understand cognitive evolution in terms of internal neural processes.

We wholly agree that embrained cognition is an improvement from focusing on whole brain size or volume. After all, neurons and their synaptic connections play a critical role in producing behavioral phenotypes. We do want to note two points of difference, however. First, Herculano-Houzel (2018) suggests that neuronal number is the best correlate on which to focus, where we recommend a plurality of within-brain correlates. Second, our view is not simply embrained cognition: We also emphasize ecological and phylogenetic context.

For Herculano-Houzel (2018), cortical neuron number is “the best predictor of quantitative differences in cognitive performance across species” (p. 92), implying that it should have priority over other measures. Our view goes beyond choosing one brain measure as a one-size-fits-all variable because we see this approach as making unvalidated assumptions about both the proximate and ultimate basis of behavioral evolution (Logan et al., 2018). The “cognitive performance” to which Herculano-Houzel refers comes from interspecific studies of “self-control” (Herculano-Houzel, 2017; data from MacLean et al., 2014). In our review, we argue that such proxies of cognition are unsuitable for intra- and interspecific correlations (Tables 1 and 2, Logan et al., 2018). Further, despite noting that “cerebral and cerebellar cortices gain neurons in tandem across mammalian species” (p. 92), which implies they are functionally interdependent and may both contribute to cognitive evolution (see also Barton, 2012; Barton & Venditti, 2014), the analysis performed in Herculano-Houzel (2017) does not account for this potential interdependence. Moreover, we have argued that the pursuit of the “best” predictor of cognition is itself a mistaken endeavor. Behavior is complex and heterogenous, and as such we may expect a range of variables to contribute to different aspects of behavioral evolution. Treating traits as “predictor” variables implies a strategy for trying to deduce cognitive ability from another, more easily quantifiable trait (i.e., using neural number to infer cognitive capacity). We worry that such a strategy is entangled with our desire to rank animals by their cognitive prowess, typically with human performance as the stick against which all others are measured. It certainly does not follow from this that neural number–behavioral correlations (or other correlations) are uninformative, but rather that we should be cautious of this strategy falling into anthropocentrism. Approaching behavioral and cognitive traits from an ecological and phylogenetic perspective, we hope, guards against this.

Herculano-Houzel (2018) suggests that we imply brain size remains a “legitimate” predictive variable and instead argues that it should be considered a descriptive variable because body mass scales only with some organs, and brain size does not track neuron number in consistent ways across taxa. In our review, we argue that brain size should not be viewed as either a predictive or a descriptive variable but as a variable phenotype. As evolutionary biologists, we should be interested in the proximate and ultimate causes of this variation. We describe how heterogeneity in brain structure and composition differentially effects brain size, and we argue that understanding how brains vary internally is an integral part of understanding brain size variation. The question follows naturally, What determines that internal variation? Answering this requires understanding both developmental aspects of brain evolution and the ultimate selection pressures shaping variation in specific brain components or networks. We do not believe the relatively crude analysis of total brain size and ill-defined behaviors provides an adequate shortcut to addressing these questions.

Herculano-Houzel (2018) contests our suggestion that brain size is a noisy trait and argues instead that it is easily measurable. However, Herculano-Houzel adopts a narrow view of what “noise” is. We defined noise as any feature that affects whether a measurement is transferable outside of the context in which it was measured; that is, “noise” amounts to exogenous confounding effects (e.g., unvalidated proxies leave an open question of whether the behavior of interest was actually approximated; a measured behavior depends on the internal state of an individual and their perception of the environment, which are not accounted for in the behavioral measure; Currie & Walsh, 2018; Logan et al., 2018). When we argue that brain size is noisy, we are not saying that it is difficult to isolate and thus measure, which is a distinct problem, but rather we are pointing to limitations on the ability of researchers to extrapolate from these measurements; such limitations arise from the exogenous confounding factors endemic in reasoning about complex, evolved systems. Herculano-Houzel’s (2018) argument that the use of brain size as a predictor of cognition is problematic due to variation in neuronal density is therefore consistent with what we have in mind; we think both phenotypic heterogeneity and phylogenetic and ecological context are important sources of noise in studies of brain size.

That aside, we do think brain size is more difficult to measure than appearance suggests: There is a non-negligible degree of measurement error. For example, cetacean species vary widely in the amount of “nonbrain” tissue found within the cranium, which introduces extensive variability when comparing endocranial volume within cetaceans and across mammalian orders (Ridgway, Carlin, Van Alstyne, Hanson, & Tarpley, 2016). Ridgway and colleagues also described how the weight of a brain can change depending on how long it has been immersed in a fixative solution, which is rarely controlled across data sets. These, and other effects such as age and sex, are some of the ways that measurement error can occur and vary in interspecific databases of “average” brain size. Currently, large comparative data sets of brain size are often based on small numbers of individuals; therefore, the extent of these effects on the results of comparative analyses is unknown.

Herculano-Houzel’s (2018) vision for the future of comparative studies of brains and cognition focuses on the emergence of new databases that systematically acquire information on brain composition across species, building on improved methodologies that balance precision and accuracy, particularly in the context of counting neuron numbers. We would happily ascribe to this future, and we are excited about the potential of the field. However, we caution against allowing this kind of analysis to fall into the same old traps—particularly, expecting to find a single brain trait to be the “best” at predicting a behavioral or cognitive variable of interest. We again emphasize phylogenetic and ecological context (see also the “contextual null” in Mikhalevich 2015). We are interested in the kinds of behaviors that make ecological sense for the clade at hand, and the kinds of predictions we make about brain–behavior relationships should emerge from that phylogenetic and ecological context. We must also be open to the possibility that any relationships found may be specific to that context. As such, our approach involves testing the predictions, and understanding the functional basis, of correlations derived from comparative studies within a target subset of species. Herculano-Houzel objects to this strategy, but she has misinterpreted our argument. She takes it that the two-pronged approach fails to consider scale: interspecies correlations are not invalidated because the correlations don’t hold within a species (p. 93). However, we did not necessarily intend for intra- and interspecific correlations to be compared, or for intraspecific studies to be a necessary validation of phylogenetic studies. We suggest in Figure 3 (Logan et al., 2018) that we should use the conclusions drawn from interspecific correlations to generate hypotheses that can be tested within a subset of species, potentially including pairs of phenotypically divergent species, where fine-grained detail can be added to the coarse level of analyses performed in comparative studies. Contrary to Herculano-Houzel’s claim that this would require focusing on lower quantitative levels of phenotypic variation, we suggest that the optimal approach would be to target a subset of species that capture the phenotypic or ecological range sampled by a broader comparative analysis. This two-pronged strategy merely reflects a pragmatic trade-off between limitations on the level of functional and behavioral detail that can be collected for each species and the number of species that can be studied.

We would, however, not exclude intraspecific studies of phenotypic variables from this approach. ­Herculano-Houzel (2018) reminds us that “brains are self-organizing systems that assimilate their environmental and life histories into their structure and function” p. 93, which may obscure genuine associations between neural and behavioral traits in intraspecific analyses. We acknowledge that plasticity is a key component of neural and behavioral development. However, phylogenetic correlations stem from selection acting on individuals within populations over time. For traits to coevolve across phylogenies, selection must have acted on intraspecific variation at some time in the past; therefore, intraspecific variation in a neural phenotype must be at least partially heritable and have had a fitness/behavioral effect in some ecological context that relates to the results of the interspecific comparisons. Identifying that context in extant populations seems like a legitimate way to follow up on interspecific correlations. But we agree with Herculano-Houzel that failing to find the same interspecific correlations at the intraspecific level does not invalidate the interspecific correlations, in part because the selection pressures and variation in extant populations does not need to be consistent with historical trends. We would also point out that environmental effects contribute to interspecific variation as well, and may themselves vary over time. This is perhaps a further reason why we should strive to follow up coarse-level phylogenetic analyses with more in-depth study in targeted species. Our point is that we should not be satisfied with simple brain–behavior correlations; we should aim to understand the complex relationship between brains, behavior, and ecology in detail.

In their commentary, Serpico and Frasnelli (2018) argue that the traditional approach (e.g., coarse-grained brain measurements, anthropocentric behavioral traits) can be successful in some contexts. A great virtue of their argument is a shift from considering whether one approach is better generally, to asking how to identify the contexts in which we should expect particular approaches to be suitable or not; building a picture of such expectations fits well with our “no one-size-fits all” approach to brain–behavior correlations.

Serpico and Frasnelli (2018) present the study of brain lateralization as a vindication of the traditional approach. They argue that even though lateralization presents differently across animals, there are nonetheless quite general things to say about asymmetries in brain function at broad phylogenetic levels. We suspect that, rather than revealing the merits of coarse-grained anthropocentric measures, this example potentially demonstrates the merits of combining broad comparative studies with detailed, fine-grained understandings of selected species. The knowledge of lateralization that Serpico and Frasnelli present comes from a functional (“bottom-up”) approach that has generated extensive amounts of data from targeted species that display lateralization (see Rogers, Vallortigara, & Andrew, 2013, for examples). This not only resulted in the conclusion that lateralization is a common trait across many taxonomic groups, as opposed to being a human-specific adaptation, but also revealed more about the cellular basis of brain lateralization and its behavioral relevance. To us, this seems like a good example of how the bottom-up approach can build depth into our understanding of a trait whose variation had initially been investigated in coarse detail across broader taxonomic levels.

We argued that understanding both the proximate and ultimate basis of behavioral evolution requires avoiding anthropocentric approaches in favor of framing arguments around the ecological context of the species under study (Logan et al. 2018). Serpico and Frasnelli (2018) view the study of lateralization as a counterexample. We are neither experts nor historians of lateralization, but our impression is that much of this field was initially driven by a desire to understand variation in lateralization in humans. As such, “success” in this context intrinsically relates to an anthropocentric aim. Nevertheless, as Serpico and Frasnelli note, even though lateralization has subsequently been observed in many lineages, we still do not understand whether this represents repeated independent innovations to a similar problem or whether lateralization inherently arises in all brains, a question we argue could be better addressed by switching to further fine-grained studies in multiple lineages. Finally, although lateralization is not itself a behavior or a cognitive ability, there are potentially ecological contexts in which behavioral variation due to lateralization has significant fitness effects. Understanding these contexts across species would link the study of neural traits, behaviors, and fitness in this example and help elucidate the evolution of the trait of interest.

We are optimistic about our ability to uncover the proximate basis, ecological context, and selective pressures shaping animal behavior and cognition. The field is already progressing at pace via a combination of novel methods of phenotyping behavior and new tools for identifying the genetic and neural variation underpinning behavioral variation. To make the most of these advances, it is time to set aside dated concepts about what brain size means and what cognition is. We should be explicit about assumptions in the field and set out to design ways to test and improve upon those assumptions. When properly applied and contextualized, the combination of detailed bottom-up studies with the potential for a phylogenetic “top-down” approach to test the generality of brain–behavior associations is a powerful route toward understanding brain–behavior relationships in ecologically relevant contexts.

References

1. Barton, R. A. (2012). Embodied cognitive evolution and the cerebellum. Philosophical Transactions of the Royal Society B: Biological Sciences, 367(1599), 2097–2107. doi:10.1098/rstb.2012.0112

2. Barton, R. A., & Venditti, C. (2014). Rapid evolution of the cerebellum in humans and other great apes. Current Biology, 24, 2440–2444. doi:10.1016/j.cub.2014.08.056

3. Currie, A., & Walsh, K. (2018). Newton on Islandworld: Ontic-driven explanations of scientific method. Perspectives on Science, 26, 119–156. doi:10.1162/POSC_a_00270

4. Herculano-Houzel, S. (2017). Numbers of neurons as biological correlates of cognitive capability. Current Opinion in Behavioral Sciences, 16, 1–7. doi:10.1016/j.cobeha.2017.02.004

5. Herculano-Houzel, S. (2018). Embodied (embrained?) cognitive evolution, at last! Comparative Cognition & Behavior Reviews, 13, 91–94. doi:10.3819/ccbr.2018.130009

6. Logan, C. J., Avin, S., Boogert, N., Buskell, A., Cross, F. R., Currie, A., … Montgomery, S. H.. (2018). Beyond brain size: Uncovering the neural correlates of behavioral and cognitive specialization. Comparative Cognition & Behavior Reviews, 13, 55–90. doi:10.3819/ccbr.2018.130008

7. MacLean, E. L., Hare, B., Nunn, C. L., Addessi, E., Amici, F., Anderson, R. C., … Zhao, Y. (2014). The evolution of self-control. Proceedings of the National Academy of Sciences, 111(20), E2140–E2148. doi:10.1073/pnas.1323533111

8. Mikhalevich, I. (2015). Experiment and animal minds: why the choice of the null hypothesis matters. Philosophy of Science, 82(5), 1059–1069.

9. Ridgway, S. H., Carlin, K. P., Van Alstyne, K. R., Hanson, A. C., & Tarpley, R. J. (2016). Comparison of dolphins’ body and brain measurements with four other groups of cetaceans reveals great diversity. Brain, Behavior and Evolution, 88, 235–257. doi:10.1159/000454797

10. Rogers, L. J., Vallortigara, G., & Andrew, R. J. (2013). Divided brains: The biology and behaviour of brain asymmetries. New York, NY: Cambridge University Press. doi:10.1002/ajhb.22485

11. Serpico, D., & Frasnelli, E. (2018). Where the standard approach in comparative neuroscience fails and where it works: General intelligence and brain asymmetries. Comparative Cognition & Behavior Reviews, 13, 95–98. doi:10.3819/ccbr.2018.130010

Where the Standard Approach in Comparative Neuroscience Fails and Where It Works: General Intelligence and Brain Asymmetries

Davide Serpico

Department of Philosophy and Educational Sciences
University of Turin

Department of Antiquity, Philosophy, and History
University of Genoa

Elisa Frasnelli

School of Life Sciences
University of Lincoln

Reading Options

---

Abstract

Although brain size and the concept of intelligence have been extensively used in comparative neuroscience to study cognition and its evolution, such coarse-grained traits may not be informative enough about important aspects of neurocognitive systems. By taking into account the different evolutionary trajectories and the selection pressures on neurophysiology across species, Logan and colleagues suggest that the cognitive abilities of an organism should be investigated by considering the fine-grained and species-specific phenotypic traits that characterize it. In such a way, we would avoid adopting human-oriented, coarse-grained traits, typical of the standard approach in cognitive neuroscience. We argue that this standard approach can fail in some cases, but can, however, work in others, by discussing two major topics in contemporary neuroscience as examples: general intelligence and brain asymmetries.

Keywords: general intelligence, brain asymmetries, comparative neuroscience

Author Note: Davide Serpico, Department of Philosophy and Educational Sciences, University of Turin, Via Verdi, 8, Turin, Italy.

Correspondence concerning this article should be addressed to Davide Serpico at davide.serpico@edu.unito.it.

Acknowledgments: We thank Anna Wilkinson for inviting us to contribute a commentary on this paper.

---

What do coarse-grained and taxon-neutral traits, such as brain size and intelligence, tell us about neurocognitive systems? What characteristics should a behavioral proxy have to allow us to properly compare cognitive abilities across different taxa? In their article “Beyond Brain Size: Uncovering the Neural Correlates of Behavioral and Cognitive Specialization,” Logan and colleagues address these questions by reviewing the literature on the relationship between brain size and cognition. In the light of empirical and theoretical considerations, the authors suggest that, to understand how brain and cognition evolve, comparative biologists should focus on fine-grained, taxon-specific phenotypic traits within the relevant ecological and adaptive context of a given species. This would help to avoid reification, defined by the authors as mistaking an operationalized target of measurement with a real, causally meaningful object.

Although we agree with this conclusion, there are two aspects analyzed by Logan and colleagues that we think may deserve further consideration. The first is the common tendency in comparative research to assess human-oriented phenotypic traits in nonhuman species; the second is the tendency, which is also widespread, to adopt coarse-grained traits, such as brain size, instead of fine-grained and more informative traits.

Let us start by considering the first one. A general premise of Logan and colleagues’ article is that comparative behavioral research is often characterized by a sort of “anthropocentric perspective.” However, as the authors argue, human behavioral traits—and the related neural bases—are not necessarily shared by other species. This argument can be understood as a criticism against the so-called Great Chain of Being, a secular ideology that assumes a natural hierarchy of organisms with humankind at the top and then, successively, lower animals from primates to bacteria (see, e.g., Sternberg, 2017). Logan and colleagues clarify how misleading this interpretation of the tree of life is by highlighting the importance of the ecological and adaptive context of a species: The ability of the organisms to achieve their goals should be evaluated within the range of challenges they face within their natural environment. Therefore, phenotypic traits should be identified by accounting for what is actually “meaningful” for a given species rather than what is meaningful for humans.

The second central point of the target article concerns the use of general, coarse-grained phenotypic traits and proxy measures to study cognition. Brain size represents, in the authors’ view, a noninformative measure of neurocognitive systems. Indeed, such a broad measure cannot disentangle from one another important aspects such as the dimensions of specific brain areas, the neuron density, and the connectivity patterns between neurons, which may be more informative than brain size about the properties of a neurocognitive system. Hence, it is not surprising that correlational analyses of brain size tend to produce spurious associations at both the intraspecific and interspecific levels (reviewed by Logan et al.).

We recognize that these criticisms could rule out misleading assumptions in comparative behavioral research. However, although the authors’ arguments seem to be applicable to most research about high-level psychological constructs, we believe they may miss the mark in respect with structural features of nervous systems and the related behavioral manifestations. To explain our concerns about the approach proposed by Logan and colleagues, we use as examples two topics in neuroscience: general intelligence and brain asymmetries. These two examples elucidate where the coarse-grained and human-oriented comparative approach fails and where instead it can be appropriate and helpful.

The authors’ argument against anthropocentric comparative research sounds effective in the case of general intelligence and its putative underlying mechanism, namely, the g factor. General intelligence represents a psychological trait assessed by psychometric IQ tests that generally recruit linguistic, mathematical, logical, and spatial abilities. The g factor, instead, is often understood as a domain-general cognitive mechanism accounting for both individuals’ performance in tests (i.e., the intelligent behavior) and individual intellectual differences within populations (see Burkart, Schubiger, & van Schaik, 2017; Serpico, 2017; Sternberg & Grigorenko, 2002). General intelligence is the subject of heated controversy in regards to two aspects tackled by Logan and colleagues with respect to brain size, that is, the granularity problem and the anthropocentric perspective widespread in comparative research.

First, exactly like brain size, general intelligence represents a coarse-grained evaluation of a cognitive system, regardless of any detail about its structural and functional composition. However, many scholars have argued that intelligence, rather than reflecting a single neurocognitive mechanism, is composed of several distinct and autonomous—but not necessarily ­independent—cognitive mechanisms (see Hampshire, Highfield, Parkin, & Owen, 2012; Serpico, 2017; Sternberg & Grigorenko, 2002; Van der Maas et al., 2006). Logan and colleagues’ argument about brain size is in line with the view that general intelligence, like other high-level psychological constructs, does not represent an informative measure of neurocognitive systems (see Craver, 2009, who discussed the problem of subtyping complex psychological traits into lower-level characteristics).

Second, we think that general intelligence is exactly that sort of human-oriented trait that Logan and colleagues criticize as erroneously generalized to other species. Although the g factor likely plays a role, if any, in human cognition only (indeed, g accounts for the population variance in a battery of IQ tests), many authors have tried to assess its role in other species. For instance, in their review about the evolution of general intelligence, Burkart and colleagues (2017) took the g factor as a domain-general neurocognitive mechanism shared by humans, primates, birds, and rodents. The arguments provided by Logan and colleagues against the anthropocentric viewpoint in comparative studies seem to us well suited to describe, and possibly to rule out, this sort of reification of general intelligence.

By contrast, the arguments provided by the authors seem to be less suited to be applied to structural properties of nervous systems. This is the case of brain and behavioral asymmetries, where a sort of human-oriented, coarse-grained approach seems to be promising and successful.

Traditionally, lateralization (i.e., the different functional specialization of the right and left sides of the nervous system) was considered a uniquely human characteristic, related to handedness and linguistic functions (McManus, 1999). Over the past decades, thanks to research conducted in comparative psychology, neuroscience, and developmental biology, we have realized that lateralization is a widespread phenomenon among vertebrates (Rogers, Vallortigara, & Andrew, 2013). Moreover, there is emerging evidence of behavioral and brain asymmetries in invertebrates, suggesting that lateralization is a feature of simpler as well as more complex neurocognitive systems (reviewed by ­Frasnelli, 2013).

In most vertebrates, for example, the right hemisphere is involved in responding to unexpected and novel stimuli (e.g., predators) and in interacting with conspecifics, and the left hemisphere is specialized in less complex and repetitive tasks (e.g., behavioral routines). Lateralization also characterizes learning and memory both in vertebrates (e.g., in birds; see Andrew, 1999; Cipolla-Neto, Horn, & McCabe, 1982; Clayton, 1993) and in invertebrates. For instance, honeybees present an asymmetrical use of the right and left antennae in learning and recalling olfactory memories: recall of short-term memory is implemented by the right side, whereas recall of long-term memory is possible through the left side (see Letzkus et al., 2006; Rogers & Vallortigara, 2008; for a review, see Frasnelli et al., 2014). This suggests that the shifts between the two sides of the brain from recently acquired information to more integrated and complete long-term records might constitute a considerable advantage for both arthropods and vertebrates. Thus, mechanisms controlling such shifts may have evolved, perhaps independently, in both phyla (see Frasnelli, Vallortigara, & Rogers, 2012).

In sum, despite differences between the nervous systems of humans and other animals, there is good evidence that lateralization represents a structural feature implemented by similar mechanisms across different species and taxa (see Rogers et al., 2013). Thus, thanks to these similarities, animal models have allowed us to uncover the evolution of behavioral and neural asymmetries, and their developmental mechanisms as well.

The take-home message is that the approach proposed by Logan and colleagues might be promising for behavioral traits reflecting higher-level cognitive aspects such as general intelligence but be potentially fragile for behavioral traits reflecting lower-level structural properties of the nervous system. As we have argued, the disagreement about general intelligence relies on its generality and on its dubious value in characterizing the neurocognitive system of nonhuman species. By contrast, what we know about lateralization points at the (partial) adequacy of a human-oriented, coarse-grained approach in comparative research.

We do not mean to assume an anthropocentric viewpoint. Nevertheless, there are good reasons to think that some structural aspects of nervous systems—and their behavioral correlates—are highly conserved across different taxa, regardless of their granularity. Whether the conservation of this kind of traits is due to natural selection, to the modularity of the related genetic processes, or to some sort of developmental constraint in brain morphology is an empirical question that future research must address.

References

1. Andrew, R. J. (1999). The differential roles of right and left sides of the brain in memory formation. Behavioural Brain Research, 98, 289–295. doi:10.1016/S0166-4328(98)00095-3

2. Burkart, J. M., Schubiger, M. N., & van Schaik, C. P. (2017). The evolution of general intelligence. Behavioral and Brain Sciences, 40, 1–24. doi:10.1017/S0140525X16000959

3. Cipolla-Neto, J., Horn, G., & McCabe, B. J. (1982). Hemispheric asymmetry and imprinting: The effect of sequential lesions to the hyperstriatum ventrale. Experimental Brain Research, 48, 22–27. doi:10.1007/BF00239569

4. Clayton, N. (1993). Lateralization and unilateral transfer of spatial memory in marsh tits. Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral Physiology, 171, 799–806. doi:10.1007/BF00213076

5. Craver, C. (2009). Mechanisms and natural kinds. Philosophical Psychology, 22, 575–594. doi:10.1080/09515080903238930

6. Frasnelli, E. (2013). Brain and behavioral lateralization in invertebrates. Frontiers in Psychology, 4, 939. doi:10.3389/fpsyg.2013.00939

7. Frasnelli, E., Haase, A., Rigosi, E., Anfora, G., Rogers, L. J., & Vallortigara, G. (2014). The bee as a model to investigate brain and behavioural asymmetries. Insects, 5, 120–138. doi:10.3390/insects5010120

8. Frasnelli, E., Vallortigara, G., & Rogers, L. J. (2012). Left–right asymmetries of behaviour and nervous system in invertebrates. Neuroscience & Biobehavioral Reviews, 36, 1273–1291. doi:10.1016/j.neubiorev.2012.02.006

9. Hampshire, A., Highfield, R. R., Parkin, B. L., & Owen, A. M. (2012). Fractionating human intelligence. Neuron, 76, 1225–1237. doi:10.1016/j.neuron.2012.06.022

10. Letzkus, P., Ribi, W. A., Wood, J. T., Zhu, H., Zhang, S. W., & Srinivasan, M. V. (2006). Lateralization of olfaction in the honeybee Apis mellifera. Current Biology, 16, 1471–1476. doi:10.1016/j.cub.2006.05.060

11. McManus, I. C. (1999). Handedness, cerebral lateralization, and the evolution of language. In M. C. Corballis & S. E. G. Lea (Eds.), The descent of mind: Psychological perspective on hominid evolution (pp. 194–217). New York, NY: Oxford University Press. doi:10.1093/acprof:oso/9780192632593.003.0011

12. Rogers, L. J., & Vallortigara, G. (2008). From antenna to antenna: Lateral shift of olfactory memory recall by honeybees. PLoS One, 3(6), e2340. doi:10.1371/journal.pone.0002340

13. Rogers, L. J., Vallortigara, G., & Andrew, R. J. (2013). Divided brains: The biology and behaviour of brain asymmetries. New York, NY: Cambridge University Press.

14. Serpico, D. (2018). What kind of kind is intelligence? Philosophical Psychology, 31(2), 232–252. doi:10.1080/09515089.2017.1401706

15. Sternberg, R. J. (2017). It’s time to move beyond the “Great Chain of Being.” Behavioral and Brain Sciences, 40, 45–46. doi:10.1017/S0140525X16001783

16. Sternberg, R. J., & Grigorenko, E. (Eds.). (2002). The general factor of intelligence. How general is it? London, England: Psychology Press.

17. Van Der Maas, H. L., Dolan, C. V., Grasman, R. P., Wicherts, J. M., Huizenga, H. M., & Raijmakers, M. E. (2006). A dynamical model of general intelligence: The positive manifold of intelligence by mutualism. Psychological Review, 113, 842–861. doi:10.1037/0033-295X.113.4.842

Embodied (Embrained?) Cognitive Evolution, at Last!

Suzana Herculano-Houzel

Department of Psychology and Department of Biological Sciences Vanderbilt Brain Institute,
Vanderbilt University

Reading Options

---

Abstract

It is time that brain size stops serving as a black box–type property of brains, “somehow” related to variations in cognitive performance across species. We now know that hidden behind similar brain structure sizes are diverse numbers of neurons and fibers that can differ in function according to experience and environment and that species differences are not a continuation of individual differences. Moving forward in understanding how cognitive evolution is linked to brain evolution requires acknowledging that, just like evolving brains are tied to evolving bodies, changing cognition comes from changing brains—and at multiple levels and timescales, which extend from inherited biological variation to experience and environmental influences that shape each individual brain and turn biological capabilities into actual abilities.

Keywords: brain size, cognition, brain evolution

Author Note: Suzana Herculano-Houzel, 111 21st Avenue South, 37240-7817 Nashville, TN.

Correspondence concerning this article should be addressed to Suzana Herculano-Houzel at suzana.herculano@vanderbilt.edu.

---

Cognition was widely considered to be a stand-alone property of brains until Francisco Varela and colleagues formulated the concept of an embodied mind, after which understanding cognition requires taking into consideration several aspects of the body and organism around the brain, from sensory and motor aspects to bodily interactions with the environment (Varela, Thompson, & Rosch, 1991). Corina Logan and colleagues make a similar appeal to those of us interested in studying brain evolution and its implications for cognition: that there can be no real understanding of cognitive evolution without a preoccupation to take into consideration the diverse makeup of the brain (and the rest of the body), as well as the neuronal basis of the behavior and the ecological context for the species in question.

The authors are conservative when they state that progress ahead will lie in acknowledging the diversity of brain morphologies and behavioral capacities and focusing on specific neuroanatomical and behavioral traits within relevant ecological and evolutionary contexts. Because one of the advantages of writing a commentary is that I don’t have to be as conservative, I would like to go further and suggest that what they propose amounts very much to the evolutionary version of embodied cognition, or how embodied cognition evolves over geological time. Their proposed “shift away from broad-scale analyses of superficial phenotypes” can be understood as requiring exactly the type of analysis that depends on what the relevant circuits underlying a given behavior (or cognition as a whole) are made of and describing how they fit in a species’ brain and ecological niche. “Embrained evolution” would thus be a fitting term for their proposed strategy for understanding how cognition compares across species and how those differences evolve—as opposed to so many of the past strategies that just glaze over what brains are made of, as if that were an unnecessary inconvenience.

In a field that has been riddled with implicit assumptions about what brain size measures and how it correlates with behavioral and cognitive traits (such as the equally ill-defined “intelligence”), Logan and colleagues make a lucid attempt to spell them out and understand the progress, but also the confusion, that those assumptions have brought. Take brain size, for instance. For decades, this was understandably the most practical morphological measurement of anything brain related, but its use was based on the initially explicit (Jerison, 1973) but then progressively more hidden assumption that absolute brain volume and relative volume of brain structures stood universally for absolute and relative numbers of neurons composing the brain and its structures. The hidden assumptions, and the fact that they were obviously incorrect, as later data attested (Herculano-Houzel, 2010, 2011a, 2011b), explain major contradictions in the interpretation of results such as the faster scaling of cerebral cortical over cerebellar volumes across species (Clark, Mitra, & Wang, 2001), whereas their surface areas scale only linearly across the same species (Sultan, 2002). Assuming implicitly that (relatively and absolutely) larger cortices are made of (relatively and absolutely) more brain neurons than larger cerebella, Clark et al. (2001) inferred that the cerebral cortex comes to dominate brain function as larger brains evolve; conversely, and using the same data set but assuming instead that it is surface area that reflects the number of neurons and hence the information-processing capacity of cerebral or cerebellar cortical structures, Sultan (2002) concluded that the two structures gained in processing capability concertedly in evolution. Both can obviously not be true at the same time. So which is it?

Neither, it turns out; being able to estimate numbers of neurons directly, without making assumptions about the volume or surface area of the brain structure they compose, allowed us and our collaborators to determine that there is not a universal relationship between the volume of a cortical structure or its surface area and the number of neurons that compose the structure (Herculano-Houzel, 2010; Jardim-Messeder et al., 2017; Mota & Herculano-Houzel, 2015). Now that absolute numbers of neurons are available and can be compared directly across equivalent structures in different species, the picture that emerges is one in which cerebral and cerebellar cortices gain neurons in tandem across mammalian species, regardless of the two-dimensional or three-dimensional size of the structures, and in which absolute numbers of cortical neurons appear as the best predictor of quantitative differences in cognitive performance across species (Herculano-Houzel, 2017). Most of the data generated so far have been concerning whole structures that are easily definable in a comparable fashion across species (whole cerebral cortex, whole cerebellum, whole olfactory bulbs). But, as those data lay a new foundation that is increasingly consistent in the story it tells, and as the method employed gains traction in the field as comparable in precision and superior in ease of use to stereology (Herculano-Houzel, von Bartheld, Miller, & Kaas, 2015), the numbers of neurons that compose functionally or anatomically identified brain structures are expected to become increasingly available in larger numbers of species. It is the hope of some of us in the field that the growing availability of these data, as well as manifestos such as that of Logan and colleagues, will drive more and more researchers to no longer consider themselves satisfied with reporting just brain size in the species they study, and instead to expand their analyses to systematic investigations of numbers of neurons (or cell subtypes, connections, synapses) in the behaviorally relevant structures across species. In this manner, I hope the day will come soon when databases emerge with systematically acquired data on brain structure composition that can be cross-correlated to behavior across species with as few assumptions as possible standing in the way of understanding the cognitive consequences of brain scaling in evolution.

So far, it has become clear already that normalizing morphometric features for body mass, a procedure that was meant to eliminate common factors, actually introduces more noise to the analysis (Herculano-Houzel, 2017). Using residuals after correction for body mass rather than absolute values was a rescuing measure prompted by the vexing realization that neither the human brain nor its cerebral cortex was the largest of them all (Jerison, 1973; Stephan & Andy, 1964), but it turns out to warp data in a manner that “favors” the relatively small bodied—when body size may actually not be relevant for cognition. Instead, it turns out that regardless of body size or even brain size, humans have the most neurons in the cerebral cortex, and crows and large parrots have just as many neurons in their relatively tiny pallium (the corresponding structure, albeit nonlayered) as the much larger brained and larger bodied macaques—­findings that make sense with their otherwise unexplained similar cognitive abilities (Emery & Clayton, 2004).

I do disagree with Logan et al. in some points. Although the authors spell out many assumptions about brain size that have turned out to be wrong, they still refer to it as a legitimate variable, which I argue it’s not—no more than body mass is informative of its parts, which is to say, really not. Although some organs scale at fairly constant proportions to others (i.e., isometrically), many are largely free to vary, which makes body mass only appear to be a good proxy for body composition (Herculano-Houzel, 2018). Similarly, it must be acknowledged that the size of the brain is the result of its parts—how many neurons of what average size in what structures—and because the parts are to a large extent free to vary in both numbers of neurons and average cell size across species and clades, brain size is not a good universal predictor of how many neurons compose each brain part (Herculano-Houzel, Catania, Manger, & Kaas, 2015; Herculano-Houzel, Manger, & Kaas, 2014). Thus, rather than proposing that “understanding how brain size relates to selection for behavioral complexity or cognition is … a two-step process” (Logan et al., 2018, p. 59), I would urge readers to keep using brain size as a descriptive variable, of course, but consider eliminating it as a ­predictive variable altogether.

 

I also disagree when the authors state that brain size is a noisy variable. It is very easily measurable with high reproducibility across measurements, which makes it far more precise than, say, our best direct estimates of numbers of neurons (which typically come with a method-defined coefficient of variation of 5%). What is noisy is how brain size has been used as a variable: as a direct measurement, an indirect estimate, or a proxy for something else; compared across multiple clades together, as if accounting for phylogenetic relationships were enough to separate what turn out to be clade-specific relationships; or compared across individuals of the same species and sometimes simultaneously across species as if variation in either case amounted to the same thing.

This latter assumption, by the way, takes me to my final commentary. Logan et al. suggest a two-pronged approach to comparative studies of cognition, in which the first prong is comparing behavior and how it relates to brain composition across individuals within same species. The underlying logic is the expectation that whatever variations turn out to be relevant for predicting differences in behavior across species (or ultimately causative) necessarily must also apply across individuals of a same species. The first and most obvious problem with this logic is one of failing to consider scale: Almost by definition (domestic dogs excluded), variation is orders of magnitude larger across species than across individuals of the same species, so the opportunity to detect significant correlations in a small range (the species) is much smaller than across a large range (across species) even if it could be ascertained that they are manifestations at different scales of the same underlying phenomenon. For instance, so far the strong correlation across species between structure mass and the number of neuronal or nonneuronal cells that compose it is not replicated across individuals of the same species (to the contrary, the opposite pattern emerges in how cell density scales with cell number; (Herculano-Houzel, Messeder, Fonseca-Azevedo, & Pantoja, 2015), but that lack of a significant correlation between structure mass and number of neurons across individuals certainly does not invalidate the obvious correlation detected at a larger scale.

Second, and most important, is the problem of failing to consider that brains are self-organizing systems that assimilate their environmental and life histories into their structure and function. Although brains (or hippocampi, or olfactory bulbs, or superior colliculi) with a twofold difference in numbers of neurons could be expected to differ in their information-processing capacity across two species, two individuals of the same species with identical numbers of neurons in the relevant structures may have had their behavior so dramatically shaped by opportunity, practice and any other number of factors that, although they have similar information-processing capabilities, their actual behavioral abilities differ in ways that couldn’t be predicted from their numbers of neurons. A two-pronged approach such as that proposed by Logan et al. is thus fine as long as it does not expect one prong to be a natural extension, or continuation, of the other. Besides embracing a program of “embrained cognitive evolution” that takes brain makeup into consideration, I would urge that comparative and evolutionary studies of cognition start taking great care in separating what are biological capabilities, dependent, for example, on numbers of neurons, or synapses, or connecting fibers, and what are behavioral abilities that incorporate opportunity, innate genetic/physiological variation, environment, and culture, and thus reveal those self-organizing properties of the brain that allow it to assimilate external information into its complexity (Herculano-Houzel, 2016).

References

1. Clark, D. A., Mitra, P. P., & Wang, S. S. H. (2001). Scalable architecture in mammalian brains. Nature, 411, 189. doi:10.1038/35075564

2. Emery, N. J., & Clayton, N. S. (2004). The mentality of crows: Convergent evolution of intelligence in corvids and apes. Science, 306(5703), 1903–1907. doi:10.1126/science.1098410

3. Herculano-Houzel, S. (2010). Coordinated scaling of cortical and cerebellar numbers of neurons. Frontiers in Neuroanatomy, 4(12). doi:10.3389/fnana.2010.00012

4. Herculano-Houzel, S. (2011a). Brains matter, bodies maybe not: The case for examining neuron numbers irrespective of body size. Annals of the New York Academy of Sciences, 1225, 191–199. doi:10.1111/j.1749-6632.2011.05976.x

5. Herculano-Houzel, S. (2011b). Not all brains are made the same: New views on brain scaling in evolution. Brain, Behavior and Evolution, 78, 22–36.

6. Herculano-Houzel, S. (2016). The human advantage: a new understanding of how our brain became remarkable. Cambridge, MA: MIT Press.

7. Herculano-Houzel, S. (2017). Numbers of neurons as biological correlates of cognitive capability. Current Opinion in Behavioral Sciences, 16, 1–7. doi:10.1016/j.cobeha.2017.02.004

8. Herculano-Houzel, S. (2018). Neuronal density in the mammalian cerebral cortex is a major determinant of variation in basal metabolic rate. Manuscript in preparation.

9. Herculano-Houzel, S., Catania, K., Manger, P. R., & Kaas, J. H. (2015). Mammalian brains are made of these: A dataset of the numbers and densities of neuronal and nonneuronal cells in the brain of glires, primates, scandentia, eulipotyphlans, afrotherians and artiodactyls, and their relationship with body mass. Brain, Behavior and Evolution, 86(3–4), 145–163.

10. Herculano-Houzel, S., Manger, P. R., & Kaas, J. H. (2014). Brain scaling in mammalian evolution as a consequence of concerted and mosaic changes in numbers of neurons and average neuronal cell size. Frontiers in Neuroanatomy, 8(77). doi:10.3389/fnana.2014.00077

11. Herculano-Houzel, S., Messeder, D. J., Fonseca-Azevedo, K., & Pantoja, N. A. (2015). When larger brains do not have more neurons: Increased numbers of cells are compensated by decreased average cell size across mouse individuals. Frontiers in Neuroanatomy, 9(64). doi:10.3389/fnana.2015.00064

12. Herculano-Houzel, S., von Bartheld, C. S., Miller, D. J., & Kaas, J. H. (2015). How to count cells: The advantages and disadvantages of the isotropic fractionator compared with stereology. Cell and Tissue Research, 360, 29–42. doi:10.1007/s00441-015-2127-6

13. Jardim-Messeder, D., Lambert, K., Noctor, S., Pestana, F. M., de Castro Leal, M. E., Bertelsen, M. F., … Herculano-Houzel, S. (2017). Dogs have the most neurons, though not the largest brain: Trade-off between body mass and number of neurons in the cerebral cortex of large carnivoran species. Frontiers in Neuroanatomy, 11(118). doi:10.3389/fnana.2017.00118

14. Jerison, H. J. (1973). Evolution of the brain and intelligence. New York, NY: Academic Press.

15. Mota, B., & Herculano-Houzel, S. (2015). Cortical folding scales universally with surface area and thickness, not number of neurons. Science, 349(6243), 74–77. doi:10.1126/science.aaa9101

16. Stephan, H., & Andy, J. O. (1964). Quantitative comparisons of brain structures from insectivores to primates. American Zoologist, 4, 59–74.

17. Sultan, F. (2002). Analysis of mammalian brain architecture. Nature, 415, 133. doi:10.1038/415133b

18. Varela, F. J., Thompson, E., & Rosch, E. (1991). The embodied mind: Cognitive science and human experience. Cambridge, MA: MIT Press.

Beyond Brain Size: Uncovering the Neural Correlates of Behavioral and Cognitive Specialization

Corina J. Logan

Department of Zoology, University of Cambridge

Shahar Avin

Center for the Study of Existential Risk, University of Cambridge

Neeltje Boogert

Centre for Ecology and Conservation, University of Exeter

Andrew Buskell

Department of History and Philosophy of Science, University of Cambridge

Fiona R. Cross

School of Biological Sciences, University of Canterbury and International Centre of Insect Physiology and Ecology

Adrian Currie

Center for the Study of Existential Risk, University of Cambridge

Sarah Jelbert

Department of Psychology, University of Cambridge

Dieter Lukas

Department of Zoology, University of Cambridge

Rafael Mares

Department of Anthropology, University of California, Davis and Smithsonian Tropical Research Institute

Ana F. Navarrete

Centre for Biodiversity, University of St Andrews

Shuichi Shigeno

Department of Biology and Evolution of Marine Organisms, Stazione Zoologica Anton Dohrn

Stephen H. Montgomery

Department of Zoology, University of Cambridge

Reading Options

---

Abstract

Despite prolonged interest in comparing brain size and behavioral proxies of “intelligence” across taxa, the adaptive and cognitive significance of brain size variation remains elusive. Central to this problem is the continued focus on hominid cognition as a benchmark and the assumption that behavioral complexity has a simple relationship with brain size. Although comparative studies of brain size have been criticized for not reflecting how evolution actually operates, and for producing spurious, inconsistent results, the causes of these limitations have received little discussion. We show how these issues arise from implicit assumptions about what brain size measures and how it correlates with behavioral and cognitive traits. We explore how inconsistencies can arise through heterogeneity in evolutionary trajectories and selection pressures on neuroanatomy or neurophysiology across taxa. We examine how interference from ecological and life history variables complicates interpretations of brain–behavior correlations and point out how this problem is exacerbated by the limitations of brain and cognitive measures. These considerations, and the diversity of brain morphologies and behavioral capacities, suggest that comparative brain–behavior research can make greater progress by focusing on specific neuroanatomical and behavioral traits within relevant ecological and evolutionary contexts. We suggest that a synergistic combination of the “bottom-up” approach of classical neuroethology and the “top-down” approach of comparative biology/psychology within closely related but behaviorally diverse clades can limit the effects of heterogeneity, interference, and noise. We argue that this shift away from broad-scale analyses of superficial phenotypes will provide deeper, more robust insights into brain evolution.

Keywords: brain evolution, cognition, comparative method, neuroethology, intelligence

Author Note: Corina J. Logan, Department of Zoology, University of Cambridge, Downing Street, Cambridge, CB2 3EJ, United Kingdom.

Correspondence concerning this article should be addressed to Corina J. Logan at cl417@cam.ac.uk.

Acknowledgments: All authors contributed to the concepts in this article at a workshop organized by Logan in March 2017. Montgomery structured the article. All authors wrote and edited the article. Logan, Montgomery, Boogert, and Mares served as managing editors. All authors approved the final version for submission. We certify that what we have written represents original scholarship. We are grateful for manuscript feedback from Nicky Clayton, Tim Clutton-Brock, Rob Barton, Anna Wilkinson, Alex DeCasien, James Higham, and two anonymous reviewers, and we appreciate Christian Rutz for discussions about New Caledonian crow foraging innovations. We thank our funders: the Isaac Newton Trust and Leverhulme Trust for a Leverhulme Early Career Fellowship to CJL, which funded the workshop on which this article is based; NERC for an Independent Research Fellowship to SHM; the European Research Council (Grant No. 3399933; SAJ); the Royal Society for a Dorothy Hodgkin Research Fellowship to NJB; the Royal Society of New Zealand Marsden Fund (UOC1301; FRC); the National Science Foundation (NSF BCS 1440755; RM); the John Templeton Foundation (AB); and the Templeton World Charity Foundation (AC; Note: The opinions expressed in this publication are those of the author(s) and do not necessarily reflect the views of Templeton World Charity Foundation).

---

Motivation

One of the central motivations for research into brain measurement is its potential to reveal links between neuroanatomical structures and cognitive capabilities. Many debates on the evolution of brains and complex behavior suggestive of advanced cognitive abilities have privileged measures where humans come out on top. This bias has been built into a number of “monolithic” general hypotheses (Barton, 2012) claiming links between measures of absolute or relative brain size and a diverse range of proxy measures of complex behavior, such as “social” intelligence (Dunbar & Shultz, 2007a, 2007b), “cultural” intelligence (Tomasello, 1999; van Schaik & Burkart, 2011; van Schaik, Isler, & Burkart, 2012), “general” intelligence (Burkart, Schubiger, & van Schaik, 2016; Reader, Hager, & Laland, 2011), and behavioral drive (Navarrete, Reader, Street, Whalen, & Laland, 2016; Wyles, Kunkel, & Wilson, 1983). In each of these cases, Homo sapiens emerge as the presumed pinnacle of a trajectory of brain evolution that correlates with increasing behavioral flexibility, intelligence, or socialization. These investigations frequently emphasize the significance of brain size. Yet we now have a more sophisticated brain measurement tool kit available (e.g., data on neuronal density or molecular variables; Montgomery, 2017). However, even with such a powerful tool kit, problems remain in establishing links between brain size and cognitive abilities because the interpretation of the correlated behaviors as more “complex” or “cognitive” remain poorly elucidated (Healy & Rowe, 2007).

Here, we argue that a fruitful approach linking brain measures and cognition involves deemphasizing coarse-grained notions of “intelligence” and whole-brain measurements in favor of (a) taxa-specific measurements of brains and ecologically meaningful behaviors, and (b) “bottom-up” extrapolation of intraspecies measures based on phylogenetic context. Based on our review of the various limitations that have previously been highlighted, we conclude by developing a framework that incorporates bottom-up and top-down approaches to advance the field. Central to this is a movement away from Homo sapiens as the measuring stick for evaluating the neuroanatomical features and behavioral capabilities of other animals.

Aims

We introduce a wide variety of research that examines brains and behavior across various phyla and discuss how lessons learned from disparate taxa can inform the way we interpret brain evolution, even among more familiar taxa such as vertebrates. Our aim is to emphasize the advantages and disadvantages of the different metrics, methods, and assumptions in this field. Our review is structured to first provide an overview of the issues that limit interpretations of brain size studies (which readers may already be familiar with; see the Limitations of Research on Brain Size and Cognition section) and explain why the limitations arise in the context of two concepts borrowed from philosophy of science: noise and interference (see the sections Why Do These Limitations of Brain–Behavior Comparative Studies Arise? and Why Do Limitations in Brain–Behavior Comparative Studies Arise?). We end with our proposed framework for how to move forward in the study of brains and behavior (Beyond Brain Size section).

• Limitations of Research on Brain Size and Cognition section: We review criticisms leveled against comparative studies of brain size.
• Why Do These Limitations of Brain–Behavior Comparative Studies Arise? Noise (p. 62) and Why Do Limitations in Brain–Behavior Comparative Studies Arise? Evidence of Interference (p. 65) sections: We go further by establishing why the recognized limitations arise. By doing so, we show why broad-stroke narratives struggle to capture the wide diversity of neuroanatomical features and behavioral capacities in animals.
• Beyond Brain Size (p. 71) section: We argue that a more targeted bottom-up approach that measures brains and behaviors at the intraspecies level to investigate cognitive, neuroanatomical, and behavioral diversity is needed to fully understand how behavioral complexity emerges from neural systems, and how well, or poorly, brain size reflects this variation.

Limitations of Research on Brain Size and Cognition

Interpreting how variation in brain size might be related to variation in cognition involves a set of assumptions that are frequently made in comparative studies:

• Brain size can be measured with negligible error.
• Investing in a larger brain comes at a cost of investing in other tissues and/or life history traits.
• Scaling relationships between brain size and body size are conserved within and across species.
• Brain regions scale uniformly with total brain size.
• Brain size scales with neuron number.
• Cognitive abilities are discretely coded in the brain.
• Cognitive abilities can be unambiguously ascertained by measuring behavior.
• Brain size is directly and linearly associated with variation in cognition.
• Selection on cognitive abilities and brain measures acts uniformly across species.

These assumptions are applied uniformly both across and within species. The validity of these assumptions has previously been challenged by Snell (1892) and Healy and Rowe (2007), and we provide additional arguments in this section. First, the use of brain size as a trait makes implicit assumptions about how brains develop and evolve (see the Assumptions and Limitations of What Brain Size Measures section). Second, when correlating brain size and a measure of cognition we make assumptions about how selection acts on, or for, either trait (see the Does Selection Act on Brain Size? section). Finally, measuring cognition inevitably requires making some assumptions about the nature of behavioral complexity and what we view as a cognitive trait (see the Assumptions and Limitations About What Brains Mean for Cognition section). In each case, the lack of data supporting the validity of these assumptions directly limits our capacity to make reliable inferences on the link between brain size and cognition.

Assumptions and Limitations of What Brain Size Measures

Brain size may seem like an easy neuroanatomical trait to measure, and the ease of obtaining a data point for a species, using one to a few specimens, renders it a historically useful starting point for many studies (Healy & Rowe, 2007; Jerison, 1985). However, brain size has also become the end point for many studies, with the variability of this trait becoming a target for evolutionary explanation. Large databases are populated by both individual measures and species’ brain size averages, which are used to examine cross-species correlations between brain size and a number of other traits. Researchers look to these databases for answers to questions such as What is the significance of a large brain? What are the costs, and what are the benefits? (e.g., Aiello & Wheeler, 1995; Armstrong, 1983; Harvey & Bennett, 1983; Isler & van Schaik, 2009; Nyberg, 1971). Cross-species correlations reveal that relative brain size (brain size relative to body size) is putatively associated with a range of life history and ecological traits. For example, relative brain size may correlate positively with longevity (a benefit) and negatively with fecundity (a cost) in mammals (Allman, McLaughlin, & Hakeem, 1993; Deaner, Barton, & van Schaik, 2003; González-Lagos, Sol, & Reader, 2010; Isler, 2011; Isler & van Schaik, 2009; Sol, Székely, Liker, & Lefebvre, 2007). Crucially, however, these correlations are not necessarily independent or consistent across taxa; for example, relative brain size and longevity do not significantly correlate in strepsirrhine primates (lemurs and lorises; Allman et al., 1993). Other analyses suggest that the relationship may be a consequence of developmental costs rather than an adaptive relationship (Barton & Capellini, 2011). Such inconsistencies in applicability and explanation raise the question, Are we failing to accurately measure and explain brain size and associated traits?

Burgeoning research in artificial intelligence and machine learning suggests the correlation between raw computing power (“brain size”) and intelligence is unlikely to be straightforward. For example, a machine-learning algorithm designed to solve a specific task may indeed get a performance boost from a “bigger brain” (i.e., utilizing more hardware, for example, when playing Go; Silver et al., 2016). However, algorithmic improvements that create more efficient ways of forming “neuronal” connections based on input data may account for even greater performance or speed improvements given fixed hardware. The effective utilization of hardware resources is itself an active research field within machine learning (e.g., Nair et al., 2015), hinting that a bigger brain does not straightforwardly translate into greater speed or better performance.

Despite Healy and Rowe’s (2007) warning, studies reporting cross-species correlations between brain size measures and various behavioral and life history traits continue to accumulate. This is also in spite of recent evidence falsifying many of the assumptions listed in the Limitations of Research on Brain Size and Cognition section (see Montgomery, 2017, for a review). For example, brain size does not scale linearly with body size within (Rubinstein, 1936) or across (e.g., Fitzpatrick et al., 2012; Montgomery et al., 2013; Montgomery, Capellini, Barton, & Mundy, 2010) species, brain regions do not scale uniformly with total brain size across species (see Heterogeneity in Brain Composition Within Taxonomic Groups section; e.g., Barton & Harvey, 2000; Farris & Schulmeister, 2011; Gonzalez-Voyer, Winberg, & Kolm, 2009), brain size does not uniformly scale with neuron number across taxa (see Heterogeneity in Brain Composition Within Taxonomic Groups section; Herculano-Houzel, Catania, Manger, & Kaas, 2015; Olkowicz et al., 2016), brain size does not necessarily translate into cognitive ability (see the Assumptions and Limitations About What Brains Mean for Cognition section and the Measuring Cognition Through Behavior Is Noisy Because We Use Unvalidated Proxies section), and brain size is not consistently related to variables of interest even within species (see the Does Selection Act on Brain Size? section; e.g., there are sex differences with regard to brain size and its relationship with cognition [Kotrschal et al., 2014; Kotrschal et al., 2013] and fitness and longevity [Logan, Kruuk, Stanley, Thompson, & Clutton-Brock, 2016]). Therefore, a research program that relies on one or more of these assumptions is limited in its ability to make reliable inferences about what brain size measures and what it means when such measures correlate (or not) with other traits.

Does Selection Act on Brain Size?

Attempts to explain variation in brain size often implicitly assume that natural selection acts on it directly. In vertebrates this assumption has been given added traction from models exploring how brain development may shape patterns of evolution that place greater emphasis on the conservation of brain architecture (Montgomery, 2017). This renders brain size a potent target of selection, in contrast to selective adaptation of particular brain regions (see the Deep Convergence in Brain Architecture section). Artificial selection experiments further highlight the capacity for selection to directly act on brain size (e.g., Atchley, 1984; Kotrschal et al., 2013). For example, artificial selection for small and large brain size in guppies (Poecilia reticulata) produced a grade shift in the scaling relationship between brain and body size, resulting in an approximately 15% difference in relative brain size between selection lines (Kotrschal et al., 2013). Although the resulting large- and small-brained guppies differed in several traits, including performance in learning tasks (Kotrschal et al., 2014; Kotrschal et al., 2013) and survival (Kotrschal et al., 2015), almost all of these correlations between behavioral performance and brain size were either test context dependent or sex dependent (Kotrschal et al., 2015; Kotrschal et al., 2014; Kotrschal et al., 2013; van der Bijl, Thyselius, Kotrschal, & Kolm, 2015).

These various trade-offs and sex-specific effects suggest that the selection landscape in natural populations may routinely be more complex than under laboratory conditions. Several recent studies of variation in brain composition among closely related populations or species that are isolated by habitat reveal heritable divergence in particular brain components rather than overall size (Gonda, Herczeg, & Merilä, 2011; Montgomery & Merrill, 2017; Park & Bell, 2010). Indeed, a recent analysis of brain morphology in wild guppies suggests selection may frequently favor changes in the size of specific brain regions, although in this case a role for plasticity has not been ruled out (Kotrschal, Deacon, Magurran, & Kolm, 2017). Focusing solely on overall brain size, as in the artificial selection experiments, might mask the co-occurring changes within the brain that underlie the observed differences in behavior. Accordingly, adaptive responses to ecological change may involve alterations in specific components of neural systems, presumably in response to selection on particular behaviors. This latter distinction is important. It is unlikely that selection ever acts “on” any neuroanatomical trait because what selection “sees” is variation in the phenotypes produced by neural systems (i.e., behavior), and the energetic and physiological costs associated with their production.

Understanding how brain size relates to selection for behavioral complexity or cognition is therefore a two-step process. First, we must understand how behavioral variation emerges from variation in neural systems. Second, we must understand how this variation in neural systems relates to overall brain size. Currently, our ability to take these steps is limited by a paucity of well-understood examples of behavioral variation in natural populations. However, existing examples provide some insight into the limitations of total brain size as a unitary trait. Recent studies of the proximate basis of schooling behavior in fish (Greenwood, Wark, Yoshida, & Peichel, 2013; Kowalko et al., 2013), and burrowing (Weber, Peterson, & Hoekstra, 2013) and parental behaviors in Peromyscus mice (Bendesky et al., 2017) suggest that outwardly unitary “behaviors” may often be composites of genetically discrete behavioral phenotypes the variation of which is determined by independent neural mechanisms.

The role of FOXP2, a transcription factor, in language development and evolution provides another informative example. FOXP2 is generally highly conserved across mammals but it has two human-specific amino acid substitutions that were likely fixed by positive selection (Enard et al., 2002). Disruption of this gene in humans severely impacts language acquisition (Lai, Fisher, Hurst, Vargha-Khadem, & Monaco, 2001), suggesting that it plays a key role in vocal learning. Insertion of the human version of the protein into the mouse genome affects the development of particular cell types in the basal ganglia without gross effects on brain size or morphology (Enard et al., 2009) yet leads to improved performance on certain learning tasks and may have a broader role in motor learning (Schreiweis et al., 2014).

These examples illustrate how variation in behaviors that are considered by many comparative studies to be correlated with whole brain size may in fact arise from localized changes in brain development that do not affect total size. This may be the kind of incremental variation selection plays with over small evolutionary time scales, and it is reasonable to assume that the accumulation of this kind of change makes a significant contribution to species differences in total brain size. Although there is some evidence that genetic pleiotropy (i.e., genetic variation in loci that cause phenotypic variation in multiple traits) can drive shifts in multiple behaviors, in many cases selection may be able to shape specific behavioral traits independently of other behaviors. Global measures of brain size and cognition both suffer from a lack of support for the underlying assumption that the correlated variation in their component parts stems from a shared proximate basis.

Assumptions and Limitations About What Brains Mean for Cognition

A highly visible thread within the literature linking cognitive abilities and brain size is a sustained attempt to use brain size as a proxy for “intelligence” (e.g., Jerison, 1969; Table 1). Notably, Jerison (1973, 1985) hypothesized that species showing behaviors assumed to require increased neural processing required the evolution of a larger brain relative to their body size to create “extra neurons” for those seemingly complex behaviors.

In discussing indicators of cognition, we first need to know when a behavior is “cognitive” or indicative of “complex cognitive abilities” (sometimes referred to as “intelligence” and often invoking the term “behavioral flexibility”; Mikhalevich, Powell, & Logan, 2017; Table 2). This is problematic because these terms are not defined well enough to test empirically or even to properly operationalize, and therefore cannot be measured in a systematic way. Appeals to “neural processing” likewise suffer from ill-definition and are poorly suited for accurate quantification in most contexts. Researchers studying animal behavior tend to avoid using the term intelligence due to its anthropocentric connotations and instead often adopt Shettleworth’s (2010) definition of cognition as “the mechanisms by which animals acquire, process, store, and act on information from the environment. These include perception, learning, memory, and decision-making” (p. 4). However, this all-encompassing definition still does not allow us to answer basic questions about the proximate machinery underlying “cognitive” traits: Is a behavior more “cognitively complex” if it engages more neurons, or certain networks of neurons, or neurons only in particular brain regions that are responsible for learning and memory? Or should we think of neural processing in dynamic terms, such as the “flexibility” of neurons to abandon old connections and form new ones as task demands change? Is behavior considered to rely on complex cognition only if it is flexible? There are no clear answers to these questions because, without a clear articulation of the empirical target, data are greatly lacking.

Indeed, it is nearly impossible to determine which behaviors require increased neural processing when they are observed in isolation from real-time brain activity. Creative studies using imaging technology can now measure behavior and brain activity at the same time, but only in species that can be trained to remain stationary in an fMRI scanner (e.g., dogs: Andics et al., 2016; pigeons: De Groof et al., 2013; see also Mars et al., 2014). However, without a priori predictions about which neural measures indicate complex cognition, this will remain a process of post hoc explanations and goal-post moving based on anthropocentric biases about which species should be “intelligent” (see Mikhalevich et al., 2017).

Theoretical reflection within the field of artificial intelligence has provided alternative definitions of intelligence that highlight the difficulties faced by cognitive ethologists. For example, Legg and Hutter (2007) aimed to provide a universal definition that could apply to machine intelligence as well as human and nonhuman animal intelligence. Informally, their definition suggests, “Intelligence measures an agent’s ability to achieve goals in a wide range of environments.” Following Legg and Hutter’s definition (without committing to whether it is definitive) clarifies several difficulties with the current approach to evaluating intelligence in nonhuman animals, and subsequently our ability to relate it to brain size. More specifically:

1. Intelligence is goal dependent. A behavior, no matter how complex, cannot be counted as intelligent if it does not serve a clear goal. Yet, interpreting goal orientation in nonhumans is inherently difficult, even under strict experimental conditions.
2. Intelligence is environment dependent. Problematically, behavioral features often associated with complex cognition such as innovation, planning, and tool use may have varying degrees of availability or relevance in different environments, which may affect whether they are displayed, irrespective of the organism’s ability to display them.
3. Intelligence of an organism is displayed across a range of environments. The few experimental setups usually used to quantify “intelligence” in captive animals may therefore be minimally informative; instead, the ability of an organism to achieve its goals should be evaluated across the range of environments it is likely to encounter within its lifetime.

Regardless of the validity of the definition, these three features—goal orientedness, environment dependency, and utility across heterogeneous conditions—highlight the practical limitations of assessing cognition in animals. The focus on utility further illustrates why selection may favor “simple” behavioral solutions to a task, or why the expression of simple behavior does not preclude the ability of an organism to identify and carry out more complex solutions in alternative contexts. Research by Bird and Emery (2009) illustrates this point nicely: Wild rooks are not reported to make or use tools; however, when given the opportunity in the lab, they are highly proficient at it. If cognition is something akin to problem-solving capacity, then we should develop measures that pay careful attention to the range of problems animals face in their natural environments, rather than transferring proxies of intelligence in humans that are relevant to the problems humans face in human environments (see also, e.g., McAuliffe & Thornton, 2015; Pritchard, Hurly, Tello-Ramos, & Healy, 2016; Rowe & Healy, 2014; Thornton, Isden, & Madden, 2014). For example, a population of a spider species (Portia orientalis) that normally encounters a wide range of prey in its natural habitat is more proficient at solving tasks than another population of the same species that normally encounters a narrow range of prey (Jackson & Carter, 2001; Jackson, Cross, & Carter, 2006).

Nevertheless, many comparative studies do find associations between gross measures of brain size and broadly descriptive behaviors. In what follows, we focus on two factors that explain when and why the results of such comparative studies should be treated with caution: biological heterogeneity, and statistical noise and interference.

Why Do These Limitations of Brain–Behavior Comparative Studies Arise? Noise

The lack of consistency in results from comparative studies (see Healy & Rowe, 2007, for an overview) strongly suggests some underlying variability in the relationship between brain size and complex cognition. In attempting to understand the properties of a particular system, it is useful to distinguish between noise (exogenous) and interference (endogenous; Currie & Walsh, 2018) as distinct kinds of confounds in brain–behavior correlations. Noise limits our ability to accurately determine and measure coevolving brain structures and cognitive abilities (this section), whereas interference hampers our ability to make inferences about the relationships between interacting features of systems (see the Why Do Limitations in Brain–Behavior Comparative Studies Arise? Evidence of Interference section). Noise and interference are major factors that shape the limitations of comparative studies of brain size and cognition by virtue of the numerous covariates that influence the reliability and power of attempts to detect true associations. Noise results from exogeneous factors that undermine our capacity to extrapolate across data points. Measurement error is an inevitable source of noise in these studies because behavior is noisy (Measuring Behavior Is Noisy Because Behavior Is Noisy section); the behaviors observed might not directly reflect single specific cognitive abilities (Measuring Cognition Through Behavior Is Noisy Because We Use Unvalidated Proxies section); and the feasibility of obtaining brain measures differs across species, thus limiting comparison (Measuring Brain Size Is Noisy Because It Is More Difficult Than It Seems section). As we show next, reducing noise requires a range of different experimental and analytical approaches.

Measuring Behavior Is Noisy Because Behavior Is Noisy

Animal behavior depends on the integration of internal motivational states and external environmental cues. Although many behaviors are largely stereotyped, the kinds of behavioral traits routinely studied in comparative studies of cognition are not. The expression of behaviors that we might interpret as social cognition (such as theory of mind) or physical cognition (such as tool use) depend on an individual’s internal state and perception of the external environment, factors that are not readily assessed. This introduces a degree of stochasticity in an animal’s behavioral expression and noise in our behavioral measurements.

Even if there seems to be an intuitive way to identify and describe a behavior, generating clear definitions that tie the behavior to a cognitive ability is frequently elusive. For example, identifying that some species live in groups does not indicate anything about their cognitive abilities (L. E. Powell, Isler, & Barton, 2017). In another example, Japanese macaques (Macaca fuscata) provide a famous case of innovation. In this case, two novel behaviors involving washing sweet potatoes before eating them and separating grains from dirt by throwing them in water were innovated by a single female (called Imo) and spread through a wild population via social transmission (cf. Allritz, Tennie, & Call, 2013; Kawai, 1965). At a population level the high rate of social transmission may be impressive, but at an individual level does the innovativeness of Imo suggest some neuroanatomical variation that supports more complex cognition and increased innovation achieved through trial-and-error learning? Imo’s brain may be no more innovative than her peers: She may simply have been in the right place at the right time or more receptive to reward stimuli. Whether inferred behavioral categories such as innovation reflect population-level variation in cognition is therefore unclear. The assumption that innovation and behavioral flexibility reflect similar cognitive processes is extrapolated from anthropocentric concepts and experiences of innovation, but this clearly requires empirical validation.

Developments in artificial intelligence and machine learning, especially in the field of reinforcement learning, illustrate how easy it is to have misconceptions regarding the neural processes that underlie innovative behavior. One possible conclusion is that Imo’s innovative ability is due to neuroanatomical variation at the intraspecies level. However, high levels of task performance can be achieved by systems that combine trial-and-error learning with feedback on their performance in the form of a reward (similar to reward-based associative learning). As reported by Mnih and colleagues (2015), a system trained from raw pixel inputs and reinforced using an environment-provided performance metric (game score) was able to achieve human-level performance in Atari gameplay by iteratively searching for the patterns that maximize game score. Silver and colleagues (2016) reported another achievement of artificial intelligence, namely of human-level performance on the game of Go, with gameplay that has been described as “creative” and “innovative” by the artificial agent first learning to predict expert moves (supervised learning) and then by improving performance through self-play (reinforcement learning). These engineering achievements suggest that the combination of chance, feedback, and repeated iterations (possibly over generations) could yield the same behavioral performance by artificial intelligence, at least in narrow domains, as organisms associated with having “complex cognition.”

Measuring Cognition Through Behavior Is Noisy Because We Use Unvalidated Proxies

Cognition is unobservable and must be inferred from behavior. Many of the 50-plus traits that have been correlated with brain size across species (Healy & Rowe, 2007) are proxy measures of the actual trait of interest (e.g., the number of novel foraging innovations at the species level is a proxy for individual-level behavioral flexibility). This would not be a problem if proxies were validated by directly testing the link between the trait of interest and its correlational or causal relationship with its proxy within a population (see Rosati, 2017, for an example of how to validate a foraging cognition proxy). However, the proxies used are generally not validated, which contributes noise and uncertainty about what the correlations, or lack thereof, between these trait proxies and brain size actually mean. We illustrate why and how unvalidated proxies are an issue for this field using innovation frequency as an example.

The hypothetical link between innovation frequency per species and their relative brain size was originally proposed by Wyles and colleagues (1983). Lefebvre, Whittle, Lascaris, and Finkelstein (1997) operationalized the term innovation to make it measurable and comparable across bird species, defining it as the number of novel food items eaten and the number of novel foraging techniques used per species as anecdotally reported in the literature (see also Overington, Morand-Ferron, Boogert, & Lefebvre, 2009). Innovation is assumed to represent a species’ ability to modify its behavior in response to a change in its environment, and is therefore a trait proxy for behavioral flexibility (e.g., Overington et al., 2009; Reader & Laland, 2002; Sol & Lefebvre, 2000; Sol, Timmermans, & Lefebvre, 2002). Behavioral flexibility is defined here as modifying behavior in response to changes in the environment based on learning from previous experience (Mikhalevich et al., 2017; Swaddle, 2016). Two challenges emerge from this conceptualization.

First, it is unclear how to calculate innovation frequency per species, or its biological significance to the species in question. For example, when one of us (Logan, 2015) tried to follow standard methods (from Lefebvre et al., 1997; Overington et al., 2009) to quantify the number of innovations in New Caledonian crows (Corvus moneduloides), it was unclear how distinct each technical innovation was. Sometimes New Caledonian crows used tools in a similar way, but they were made from different materials. More important, many innovations were only novel or unusual to the humans who saw crows performing these behaviors; these behaviors are commonly performed by New Caledonian crows across their natural habitat (e.g., Hunt & Gray, 2002) and are certainly not novel to them, suggesting that innovation frequency databases (e.g., Overington et al., 2009) may contain many similar cases of species-typical behaviors that had gone unnoticed to the human observer. Therefore, it is also unclear what innovation frequency per species means to that species, which further confounds the significance of innovation frequency per species. At this stage, it is unclear what an appropriate measure of innovation frequency would be when comparing across species. However, it is clear that any measure needs to be grounded in direct observations at the within-species level.

Second, the small number of comparative studies that exist shows that innovation frequency per species does not correlate with measures of behavioral flexibility in individuals (Auersperg, Bayern, Gajdon, Huber, & Kacelnik, 2011; Bond, Kamil, & Balda, 2007; Ducatez, Clavel, & Lefebvre, 2015; Jelbert et al., 2015; Logan, 2016a, 2016b; Logan, Harvey, Schlinger, & Rensel, 2016; Logan, Jelbert, Breen, Gray, & Taylor, 2014; Manrique, Völter, & Call, 2013; Reader et al., 2011; Tebbich, Sterelny, & Teschke, 2010) or with species-level estimates of brain size (Cnotka, Güntürkün, Rehkämper, Gray, & Hunt, 2008; Ducatez et al., 2015; Emery & Clayton, 2004; Isler et al., 2008; Iwaniuk & Nelson, 2003; Pravosudov & de Kort, 2006) in predictable ways. More generally, despite a lack of validation that they accurately reflect the trait of interest, proxies of behavioral traits are pervasive in the comparative brain size literature and introduce unknown amounts of exogenous noise into cross-species analyses. This noise may generate spurious results, masking “true” patterns in the data and impeding their interpretation.

Measuring Brain Size Is Noisy Because It Is More Difficult Than It Seems

Most work on brain evolution has focused on overall brain size or changes in large regions of the brain, such as the forebrain and the cerebellum (see review in Healy & Rowe, 2007; see also Herculano-Houzel, 2012; Reader et al., 2011). However, volumetric measurements are particularly noisy. We use primate brain data to illustrate the difficulties involved in obtaining, preserving, and measuring brain volumes.

More is known about brain anatomy in primates than in other orders, yet volumetric measurements of specific brain regions in this group are available only for a few species, from only a few individuals per species (Reader & Laland, 2002); are limited to only a few brain collections (Zilles, Amunts, & Smaers, 2011); and often come from captive individuals. This introduces a large amount of noise because a species’ average brain, or brain region, volume might be biased due to sexual dimorphism or other variables that differ across individuals (Montgomery & Mundy, 2013).

Complications arise in determining whether it is appropriate to correlate behavioral data from wild individuals with morphological data (e.g., brain size) obtained from captive individuals. Studies comparing the morphology of wild and captive animals have shown that rearing conditions may influence body composition (e.g., skull shape, brain size, digestive tract) after only a few generations (O’Regan & Kitchener, 2005). In primates, brain mass is not generally affected by captivity (Isler et al., 2008), but body mass is: some species become heavier, while others become lighter due to inadequate diets (O’Regan & Kitchener, 2005).

Furthermore, although brain size might not be affected by captivity, primate populations of the same species that were reared under different captive conditions differ in cortical organization (Bogart, Bennett, Schapiro, Reamer, & Hopkins, 2014). In macaques and humans, there is evidence that individual differences in social network size correlate with amygdala volume and areas related to this structure (Bickart, Wright, Dautoff, Dickerson, & Barrett, 2011; Kanai, Bahrami, Roylance, & Rees, 2011; Sallet et al., 2011). Among individuals of the same species, brain anatomy changes significantly with age (Hopkins, Cantalupo, & Taglialatela, 2007). Choosing individuals with closely matched histories can reduce noise in brain measures that are introduced by individual differences in previous experience, but the noise involved in brain volume measurements is most effectively controlled and minimized by obtaining large sample sizes per species to acquire more reliable species averages. This problem is particularly vexing when combining behavioral data sets from observations in the wild with neuroanatomical data from captive populations.

Data collection methods can also compromise the quality of the data. Many reported brain weights and brain volumes are actually proxies of these measures, obtained instead by calculating endocranial volume from skulls, which are much easier data to collect (e.g., Isler et al., 2008; Iwaniuk & Nelson, 2002). Although endocranial volume has been shown to reliably approximate brain volume across species of primates (Isler et al., 2008) and birds (Iwaniuk & Nelson, 2002) and within species of birds (Iwaniuk & Nelson, 2002), this might not always be the case. For example, Ridgway, Carlin, Alstyne, Hanson, and Tarpley (2016) suggested that endocranial vascular networks and other peripheral appendages can account for 8% to 65% of endocranial volume in cetaceans, leading to a consistent overestimation of brain size that is more severe in some species than others. In addition, there is a risk of bias during the measurement or assessment if researchers might favor a particular hypothesis and if the identity of the skulls or brains is not blinded during the study (Lewis et al., 2011).

Because brains are valuable tissues, noninvasive methods such as magnetic resonance imaging (MRI) are preferred for obtaining data on brain anatomy and function. Yet high-resolution, high-quality MRIs from primate brains are difficult to obtain from live individuals. Images obtained using in vivo techniques, where the animal is sedated for a short period while scanning the brain, might be more accessible, but image quality and resolution are poorer than in images obtained postmortem (K. L. Miller et al., 2011). Postmortem MRIs can have a higher resolution and are therefore more suited to calculating volumes. However, even MRIs are problematic because of other sources of noise that arise from brain extraction methods, including the postmortem delay between death and extraction and preservation, and the “age” of the preserved brain (i.e., how long a brain has been stored; Grinberg et al., 2008; K. L. Miller et al., 2011). Although postmortem MRI is the best method available for calculating brain volumes, brain volume in itself is a noisy measure because of its unclear, and usually untested, relationship with other variables of interest (see the Measuring Cognition Through Behavior Is Noisy Because We Use Unvalidated Proxies section).

Why Do Limitations in Brain–Behavior Comparative Studies Arise? Evidence of Interference

Interference occurs when systems consist of multiple interacting parts whose interactions tend to be complex. A potentially useful way of understanding some critiques of brain size–cognition comparative studies is to consider the ramifications of heterogeneity within and across species in terms of their brain architectures and associated traits (e.g., behavior, cognition, life history; Figure 1). If parts of the brain evolve in concert due to developmental coupling, for instance, then interference from those components makes it difficult to isolate the evolutionary causes of changes in brain size, or any of its components, over time. Similarly, if many ecological and life history traits covary, identifying which factors drive changes in brain size is complicated by autocorrelation between independent variables. Philosophers distinguish heterogeneity within and between systems as a useful concept for framing the validity of comparisons (Elliott-Graves, 2016; Matthewson, 2011). Heterogeneity arises as a confounding factor in comparisons among individuals and/or species when the components of a system (e.g., brain structures) differ (see the Heterogeneity in Brain Composition Within Taxonomic Groups section), or when similar components exist but differ in scaling relationships or patterns of connectivity (e.g., neuron density, neural network; see the Deep Convergence in Brain Architecture section). Treating brain size as a unitary trait assumes either that the brain is a unitary trait or that any signal from a brain–behavior association is sufficient to overpower the influence of heterogeneity on either trait. Comparisons of taxonomically diverse neural systems can identify where similar brain architectures exist and where heterogeneity in brain composition is masked by comparisons of brain size (see the Effects of Size-Efficient Selection section). Interference in the form of heterogeneity between systems occurs because of the complex interactions among life history and ecological factors that shape the coevolution of cognitive abilities and particular brain measures (see the Correlations Suffer From Interference section).

Heterogeneity in Brain Composition Within Taxonomic Groups: Brains That Appear Similar According to Certain Measures May Actually Be Different

The brain architecture underlying ecologically relevant neural computation will depend on the behavioral requirements of a task; the evolutionary history of the machinery that selection is building on; and the strength of potentially opposing selective forces such as energetic, volumetric, and functional trade-offs and constraints. Even across more closely related species—for example, among mammals—heterogeneity between brain structures introduces noise and variation that can complicate brain–behavior relationships.

Although some authors argue that the major axis of variation in mammalian brains is overall size (e.g., Clancy, Darlington, & Finlay, 2001; Finlay, Darlington, & Nicastro, 2001), there is ample evidence for variation in brain structure across species (e.g., Kaas & Collins, 2001; Workman, Charvet, Clancy, Darlington, & Finlay, 2013) caused by brain region–specific selection pressures, so-called mosaic brain evolution (Barton & Harvey, 2000; Smaers & Soligo 2013). When a behavior generated by a specific brain structure is targeted by selection, the effect on total brain size will depend on the scaling relationship between that brain structure and total brain size. For example, one general trend across mammalian brain evolution is a correlated expansion of the neocortex and cerebellum, which occurs independently of total brain size (Barton, 2012; Whiting & Barton, 2003). These structures share extensive physical connections and are functionally interdependent (Ramnani, 2006), but, although they tend to coevolve, both have evolved independently in some evolutionary lineages (Barton & Venditti, 2014; Maseko, Spocter, Haagensen, & Manger, 2012). Independent selection pressure on individual brain components such as the neocortex and cerebellum do not have equal effects on overall brain size or measures of encephalization (Figure 2). Neocortex volume scales hyperallometrically with brain volume (i.e., as brain size increases, the proportion of neocortex tissue increases), whereas cerebellum volume, and several other major brain components, scale hypoallometrically with brain volume (Barton, 2012). As a result, increases in neocortex volume have a disproportionate effect on brain volume compared to similar proportionate increases in cerebellum size, largely due to differences in the scaling of neuron density and white matter in the two structures (Barton & Harvey, 2000; Herculano-Houzel, Collins, Wong, & Kaas, 2007). Variations in whole brain size, or measures of brain size relative to body size, such as the encephalization quotient (Jerison, 1973), therefore essentially correspond to variation in neocortex size and mask variation in other brain components, even though the latter may be of great functional significance. For example, the frequency of tool use in primates (Barton, 2012) and the complexity of nest structure in birds (Z. J. Hall, Street, & Healy, 2013) have been linked with variation in relative cerebellum volume; and hippocampal volume has been linked with performance on a variety of cognitive tasks in primates (Shultz & Dunbar, 2010), and with spatial memory in birds (e.g., Krebs et al., 1996).

When comparing brain size across species, further heterogeneity is apparent at the level of the cellular composition of brain structures. Recent data on neuron number in brain regions of birds and mammals have revealed extensive variation across taxonomic groups (Herculano-Houzel et al., 2015). For example, primates have significantly higher neuron densities in the neocortex and cerebellum than other closely related terrestrial mammals, whereas elephants have substantially higher neuron densities in the cerebellum than other Afrotheria (e.g., golden moles and sea cows; Herculano-Houzel et al., 2015), and the brains of some birds pack similar numbers of neurons as found in monkeys due to the relatively higher neuron densities in avian brains (Olkowicz et al., 2016). Because neurons and their synaptic connections are the basic computational units of any neural system, if variation in brain, or brain region, volume does not consistently reflect variation in neuron number, then any inference made about the cognitive significance of brain size is largely invalid. To illustrate this effect, averaging across brain regions, a 1g brain that follows primate neuron number brain-size scaling rules will contain approximately 26% more neurons than a brain that follows the glire scaling rules (the clade including rodents; Herculano-Houzel et al., 2015). A 1g brain that follows psittacine (parrots) scaling rules will contain about 100% more neurons than a brain that follows the glire scaling rules and about 58% more than a brain that follows the primate scaling rules (Olkowicz et al., 2016). Comparing brain size across taxa with different or unknown scaling rules thus erroneously assumes that the computational output (based on neuron number) of these hypothetical brains would be equal. At an even smaller scale, brains differ in traits such as neuronal connectivity, receptor density, or neurochemistry (Butler & Hodos, 2005), traits that are difficult to measure but could have important roles in the functioning of the brain (Mars et al., 2014).

The assumption that brain volume is comparable and meaningful across species is often explicitly made in broad phylogenetic studies of cognitive ability (e.g., MacLean et al., 2014). Variation in brain structure and cellular composition strongly questions this assumption. The effect of incorporating more fine-grained data, even if they are relatively crude, is apparent in existing studies. For example, in Benson-Amram, Dantzer, Stricker, Swanson, and Holekamp’s (2016) analysis of how performance on a puzzle-box test is associated with brain size across 39 species of mammalian carnivore, the addition of data on volumetric variation in brain structure significantly improved their predictive model compared to one containing only brain volume. In a recent opinion piece, Herculano-Houzel (2017) also argued that (cortical) neuron number outperforms total brain size as a predictor of behavioral performance in self-control tests reported by MacLean and colleagues (2014). The power of brain size as a causative predictor of cognitive performance is therefore apparently vulnerable to the addition of only narrowly more fine-grained data.

Deep Convergence in Brain Architecture: Brains That Appear Different According to Certain Measures May Actually Be Homologous

At the broadest taxonomic scale, brain composition is remarkably diverse. For example, comparative studies have traditionally focused on linking learning and memory with arachnid protocerebrums (e.g., Meyer & Idel, 1977; Punzo & Ludwig, 2002), insect mushroom bodies (e.g., Snell-Rood, Papaj, & Gronenberg, 2009), cephalopod vertical lobes (e.g., Grasso & Basil, 2009), the vertebrate pallium (e.g., Jarvis et al., 2005), and mammalian neocortices (e.g., Pawłowskil, Lowen, & Dunbar, 1998). Despite their independent evolution, some research points toward commonalities in the molecular and neural systems that function in heterogeneous brain organizations across animal phyla. A combinatory expression pattern of developmental control genes suggests the deep origin of key learning and memory centers, including in the complex sensory centers and cell types of the mushroom bodies of annelids and arthropods, and the pallium of vertebrates (Tomer, Denes, Tessmar-Raible, & Arendt, 2010). Similarly, Pfenning and colleagues (2014) proposed that vocal learning in birds and humans has evolved via convergent modification of brain pathways and molecular mechanisms. G. Roth (2013) proposed that the centers for learning and memory in insect, octopus, avian, and mammalian brains share a comparable associative network that “bring[s] the most diverse kinds of input into the same data format and [integrates] the respective kinds of information” (p. 292). These broad comparisons suggest that such brain structures in taxonomically and anatomically diverse animals may share a number of features, including high neuron density, and similar organizations with hierarchical connectivity (G. Roth, 2013). Similarly, the vertebrate basal ganglia and insect central complex have been shown to exhibit a deep homology, sharing similar network organizations, neuromodulators, and developmental expression machineries (Strausfeld & Hirth, 2013). Accordingly, divergent structures may have converged on similar architectures and computational solutions to analogous behavioral challenges (Shigeno, 2017). By simplifying brain measures by focusing only on size, we may miss out on opportunities to study how convergences in behavior and complex neural systems can inform how cognition evolves.

Nevertheless, the heterogeneity identified by these studies may also provide useful variation that can contribute to our understanding of brain and cognitive evolution. For example, if neuronal density can vary independently of volume, why? And how does this impact the functional properties of the pathways that produce complex behaviors associated with cognitive prowess?

Effects of Size-Efficient Selection

Although heterogeneity in brain systems limits the scope of comparative studies of brain size, it also provides an opportunity to understand how selection acts on neural systems and why selection favors particular solutions over others. One key factor may be the role of size-efficient selection and redundancy in nervous systems. Neurons are energetically expensive cells, and their total cost scales predictably with the size of the neural system (Laughlin, de Ruyter van Steveninck, & Anderson, 1998). Selection must therefore constantly trade off behavioral performance with energetic and computational efficiency. Exploring how these trade-offs are resolved in real and artificial systems has the capacity to greatly inform why some animals invest in larger brains and others do not (Burns, Foucaud, & Mery, 2010; Chittka & Niven, 2009; Chittka, Rossiter, Skorupski, & Fernando, 2012; Menzel & Giurfa, 2001).

Compared with vertebrates, arthropods have tiny brains and vastly fewer neurons in their nervous systems (Eberhard & Wcislo, 2011), yet many insects and spiders display highly sophisticated motor behaviors, social organizations, and cognitive abilities (Chittka & Niven, 2009). For example, insects and spiders exhibit numerical cognition (Cross & Jackson, 2017; Dacke & Srinivasan, 2008; Rodríguez, Briceño, Briceño-Aguilar, & Höbel, 2015), planning (Cross & Jackson, 2016; Tarsitano & Jackson, 1997), selective attention (Jackson & Li, 2004), working memory (Brown & Sayde, 2013; Cross & Jackson, 2014; Zhang, Bock, Si, Tautz, & Srinivasan, 2005), and they flexibly match behavior to changes in prey behavior (Wardill et al., 2017)—all typically studied in vertebrates and considered cognitively demanding (Chittka & Niven, 2009), illustrating that selection has favored highly efficient neuronal systems in these taxa.

Although an imperfect analogy, researchers’ experience with training artificial neural networks provides an insight into how efficient neural networks can be constructed. Indeed, researchers who aim to create an artificial network that serves as a pattern-learning machine have been largely inspired by the organization of the cerebral cortex in mammals (Mnih et al., 2015). This comparison between artificial networks and cerebral cortex organization was made more notable with recent advances in deep convolutional neural networks (an artificial neural network with a large number of intermediary layers, specialized in identifying patterns in perceptual inputs) such as the deep-Q network. Beyond mammals, this layerlike organization can also be identified in the brains of, for example, the common octopus and Drosophila, suggesting that a common functionality of information processing patterns may be represented in both artificial and biological neural networks (Shigeno, 2017).

One of the key messages from such research is that training large neural networks is still difficult (Bengio, Simard, & Frasconi, 1994; Glorot & Bengio, 2010; Pascanu, Mikolov, & Bengio, 2013). Even when training is successful, it requires a great deal of time and input data, but, more importantly, training too large a network without the right algorithm often simply fails. In artificial systems, this happens when feedback from the environment is used by the neural network to determine certain flexible values of the computational architecture (e.g., connections between artificial neurons). This problem scales up: Greater numbers of flexible values (i.e., network parameters, which grow in tandem with “brain size”) require greater amounts of input data and increasingly complex algorithms. Such trade-offs are likely also faced by biological organisms. Thus, in addition to the energetic costs of larger brains, there are also informational costs (i.e., a need for more, better, and/or faster inputs) and computational costs (i.e., efficient ways to use inputs, which may be architecturally difficult for natural selection to find) that limit brain size and may channel the response to selection away from simple increases in the total size of the system or brain.

The hand of size-efficient selection can also be seen in the network architecture of large brains that display a “small-world” topology (Ahn, Jeong, & Kim, 2006; Chen, Hall, & Chklovskii, 2006), which minimizes energetically costly long-range connections in favor of proportionally high local connectivity (Bullmore & Sporns, 2012; Buzsáki, Geisler, Henze, & Wang, 2004; Watts & Strogatz, 1998). Yet, if network architecture is constrained by energetic costs, then what does the evidence of variation in cellular scaling between brain components within and across species tell us about how brains evolve?

Variation in the scaling of neuron number with volume likely reflects differences in cell size and patterns of connectivity between neurons. The low neuron density in the neocortex in mammals, compared to that of the cerebellum, reflects the high proportion of the neocortex given over to white matter that consists of mid- to long-range fibers connecting neurons (Ringo, 1991). Variation in the pattern of neuronal connections, and integration between brain regions, may help explain variation in cellular scaling. Similar explanations may also apply to scaling differences across taxa, with the high neuronal density of primates being associated with relatively smaller volumes of white matter and connectivity (Ventura-Antunes, Mota, & Herculano-Houzel, 2013). However, these scaling differences could also be driven in part by external influences related to ecology, body size, and morphology. Body size affects many aspects of an animal’s ecology, diet and energy consumption, and physiology (LaBarbera, 1986). It should be no surprise that this may extend to brain composition. For example, the ancestor of extant primates, and most of its descendants, occupied arboreal niches (Cartmill, 1972) and had arboreal locomotor strategies that constrain body size and favor a low center of mass—a strategy that is likely inconsistent with volumetrically expensive modes of brain expansion. Selection pressures that favored the evolution of increased neuron number may therefore have been constrained by the physical demands of occupying an arboreal niche, resulting in changes in neural development that were associated with increased neuron density. Similar, but stronger, selection regimes may also explain the extremely high neuron densities in bird brains (Olkowicz et al., 2016). Conversely, the much lower neuron densities of cetaceans (Eriksen & Pakkenberg, 2007) would be consistent with the relaxed constraints on body size and locomotor evolution associated with the rapid diversification of this lineage (Slater, Price, Santini, & Alfaro, 2010).

The expectation that brain size should be a simple predictor of cognitive performance ignores the effect of size-related selection pressures (Chittka & Niven, 2009; Chittka et al., 2012). Size-efficiency is most obvious when considering brain function in small invertebrates, but mounting evidence suggests that the same principles may apply even among vertebrates occupying distinct ecological niches that define the range of permissible body sizes and architectures (Olkowicz et al., 2016). Body size is regularly used as a “size-correction factor” on the assumption that residual brain size is more cognitively relevant, but variation in body size itself reflects the presence of wider ecological and physical selection pressures that may render brain composition and function more divergent than size alone (Fitzpatrick et al., 2012; Montgomery et al., 2010; Montgomery et al., 2013).

Correlations Suffer From Interference

Problems of noise are compounded by interference from the complex relationships between many behavioral and anatomical traits. This interference influences our ability to determine not only whether a mechanistic link exists between specific brain measures and a certain behavior or cognitive ability, but also their functional link and their adaptive evolutionary history. The comparative study of different species can provide insights into how differences in behavior link with differences in brains (Harvey & Pagel, 1991), and phylogenetic comparisons have been the most widely used approach to test hypotheses about adaptation (see the section Does Selection Act on Brain Size?). However, in addition to relying on unvalidated proxies, adaptive stories are frequently based on correlations. It is therefore necessary to identify potential interference from unmeasured variables to gather evidence for causation before we can accept such adaptive accounts as accurate.

There are four main ways in which interference limits the potential to interpret whether correlations represent adaptations. First, any association between differences in brain measures and behavior might not be direct but be caused by interfering factors. For example, increases in brain size and group size both appear to occur in species that eat foods with high nutritional value; therefore, the correlation between brain size and group size might be the result of noise from dietary changes (Clutton-Brock & Harvey, 1980; DeCasien, Williams, & Higham, 2017; L. E. Powell et al., 2017). Second, even if population studies indicate that a measure of brain size and a behavior are directly linked, comparisons across species cannot immediately reveal the causal direction of the association. For example, an association between increased brain size and decreased risk of predation might result from large-brained species being better able to avoid predation (Kotrschal et al., 2015), or from species with low predation pressure having the opportunity to invest additional resources into brain growth (Walsh, Broyles, Beston, & Munch, 2016). Third, external factors frequently mediate the expression of any link across taxonomic groups. For example, switching to a frugivorous diet might lead to selection on olfactory ability in nocturnal species and visual abilities in diurnal species, resulting in independent episodes of brain expansion driven by selection on distinct sensory modalities and brain components (Barton, Purvis & Harvey, 1995). Fourth, any current link between brain size and behavior might be the product of co-option, after the initial evolution of that brain aspect, rather than the driving selection pressure itself. For example, abilities such as object permanence (i.e., the ability to recall the presence of an out-of-sight object) might have been selected because individuals need to remember the spatial position and temporal availability of food sources in their home range, but it could subsequently be used to distinguish neighbors from strangers (Barton, 1998). Similarly, selection for improved visual acuity in foraging primates may have later been co-opted to serve in individual recognition and social cognition (Barton, 1998). Although some attempts have been made to tease apart these relationships using path analysis (Dunbar & Shultz, 2007b), this approach still suffers from the effects of colinearity among variables and does not provide a mechanistic understanding of causative relationships (Petraitis, Dunham, & Niewiarowski, 1996). Recent advances provide some ways to overcome these limitations in the comparative approach (see Scaling Across Taxa to Integrate Evidence section), but as previous authors have pointed out (Garland, Bennett, & Rezende, 2005; Gonzalez-Voyer & Hardenberg, 2014; Harvey & Pagel, 1991), interference fundamentally limits our ability to determine past evolutionary processes based on simple observations of species alive today.

These effects are likely to be particularly influential in the small data sets that characterize many comparative analyses of cognition and brain measures, due to the difficulty in obtaining data. With small data sets, correlations are unlikely to be stable, unless the effect size is large, or noise and interference are low (Schönbrodt & Perugini, 2013). In the vast majority of studies, accuracy and sample size are directly traded off against each other due to logistical and cost constraints. Although this is inevitable, studies aiming for broad phylogenetic comparisons by relying on crude proxies of cognition supposedly measurable across very divergent taxonomic groups may be futile. Any trade-off that reduces accuracy to increase taxonomic breadth risks relying on invalid measures, resulting in unstable and potentially meaningless correlations. Comparisons across large, diverse taxonomic groups can be helpful to identify and describe patterns of variation; however, key insights into the evolutionary history of traits and their associations will be gained by incorporating detailed population studies (see the section Bottom-Up versus Top-Down). As neuronanatomical, behavioral, and statistical tools become increasingly comprehensive and sophisticated, the solutions to these issues will be reachable in the near future.

Beyond Brain Size

Matching the Right Tool With the Right Question

In the previous two sections we emphasized how heterogeneity in brain composition and behavior/cognition, and the subsequent noise this generates, can influence our attempts to measure the relationships between these variables. We think these issues motivate turning from coarse-grained, “taxon-neutral” (or hominid-inspired) measures to more local, taxon-specific studies. This is not to say that heterogeneity on its own undermines existing “monolithic” narratives of brain size and behavioral complexity. Rather, these narratives ignore the complexity of links between brain morphology, body morphology, and behavior, and often abstract away from the important ecological and evolutionary drivers of complex behavior that we are trying to understand. We therefore argue against privileging anthropocentric measures or criteria. Instead, we urge a recognition of the multidimensional and multileveled structure of brains, as well as the disparate and varied ways that brains evolve—in conjunction with bodies, and in response to specific environments—to produce complex behavior. Understanding how brains evolve in response to selection on behavioral complexity or cognition is therefore a two-step process. First, we must understand how behavioral variation emerges from variation in neural systems. Second, we must understand how brains change across species and how this might relate to differences in adaptive regimes.

Discovering and probing correlations between properties of brains and behavioral features can be part of a powerful comparative approach, but we should be wary of reification: mistaking an operationalized target of measurement with a “real” object (Whitehead, 1925). There is a difference between something being measurable and it being causally meaningful. We, and others (e.g., Chittka et al., 2012; Healy & Rowe, 2007), have questioned whether coarse-grained, cross-taxa measurements, such as the encephalization quotient, pick out relations that are in fact explanatory of the evolutionary and developmental relationships between brain, cognition, and behavior across lineages. In fact, similar arguments have been made since scientists first started comparing brain measures across species (Snell, 1892). Instead, we argue for an increased focus on a bottom-up approach that begins with (a) measurements of features that can be validated within particular taxa in ecologically relevant experimental contexts, before (b) testing the evolutionary variability in the relationships between brains and behavior across related species. This will help avoid reification by starting with intraspecific, experimentally verifiable causal connections. The first task involves probing how various taxa respond behaviorally to their environments and other stimuli and determining whether those properties correlate with brain measures in revealing ways. These brain measures will frequently be more fine-grained than brain size, concerning particular neuroanatomical and/or neurophysiological features. The second task involves the construction and testing of hypotheses about the ancestral and evolutionary relationships between those taxa, enabling us to expand to broader categories and correlations in a careful, piecemeal fashion. We expect the results of these two tasks to relate in dynamic ways: Considerations of evolutionary scenarios are likely to highlight new kinds of experimental tests and hypotheses in local contexts, and these scenarios will depend crucially on information about local taxa.

Bottom-Up versus Top-Down

The top-down approach uses cross-species correlations between brain measures and a trait of interest and can be useful for generating hypotheses. However, although these are important for motivating research into the links between brains and behavior, we argue that specific hypotheses should then be tested at the within-species level: from the bottom up. The bottom-up approach involves directly testing behavior and cognition in individuals to determine how they relate to brain measures in these particular individuals of a particular species (ideally measured at the same time as behavior/cognition) to build validated, causative correlations (Chittka et al., 2012). When sufficient data on individuals from a variety of species have accumulated, phylogenetic meta-analyses can be conducted to test whether consistent patterns emerge and hold within and across species (see the section Scaling Across Taxa to Integrate Evidence; Table 3). Correlations within contemporary populations can tell us whether processes are homologous or analogous across species and show the limits of which processes are likely to occur.

The contrast between top-down and bottom-up approaches is often presented as a difference in terms of investigating the ultimate (top-down looking at adaptations and fitness) versus proximate (bottom-up looking at mechanisms and development) reasons for the evolution of a trait (Laland, Sterelny, Odling-Smee, Hoppitt, & Uller, 2011; Scott-Phillips, Dickins, & West, 2011). However, the approach we suggest does not necessarily make this potentially problematic distinction (Beatty, 1994; Calcott, 2013; Cauchoix & Chaine, 2016; Laland, Odling-Smee, Hoppitt, & Uller, 2013). Our main argument for a bottom-up approach is to encourage researchers to have a clear understanding of what they are investigating rather than to rely on proxies. Detailed individual-based studies can not only reveal which brain measures are involved in a particular cognitive ability or behavior but also provide important insights into the ecological correlates and fitness consequences of variation in particular brain measures (Table 3). Further, starting from behaviors in particular species makes ensuring ecological, developmental, and evolutionary relevance significantly more straightforward: It is a strategy for both avoiding reification and being sensitive to the heterogeneity of both brains and behavior. Embracing neural diversity provides an opportunity to take a step below volumetric variation to try to understand what larger brains can do that smaller ones cannot. This research does not have to be designed from a blank slate: Although many of the previous comparative studies that link brain measures to behavior/cognition do not help us understand how brains produce behavioral variation, they can help direct researchers in choosing which questions to address.

For example, spatial navigation behavior has been directly linked to the hippocampus using the bottom-up approach. Supporting evidence comes from intraspecies behavioral studies in birds with hippocampal lesions, which indicates the causal relationship between location memory and the hippocampus (Hampton & Shettleworth, 1996; Patel, Clayton, & Krebs, 1997). In addition, ecological correlates were found in black-capped chickadees where individuals living in harsher environments (higher latitudes) were more efficient at recovering caches (spatial memory) and had larger hippocampal volumes with higher neuron densities and more neurogenesis than individuals at lower latitudes (Chancellor, Roth, LaDage, & Pravosudov, 2011; Pravosudov & Clayton, 2002; T. C. Roth & Pravosudov, 2009). Further, real-time brain activity has been paired with navigational behavior in rats: When navigating through a maze, particular neurons (place cells) fired at particular locations in the hippocampus (Gupta, van der Meer, Touretzky, & Redish, 2010). Later, when the rats were not in the maze, rats mentally “ran” through the maze and even invented novel routes as evidenced by the sequences of the firing of their place cells (Gupta et al., 2010). Place cell research and experimental designs that behaviorally test episodic-like memory (e.g., Clayton & Dickinson, 1998) provide evidence for brain–behavior causations from the bottom-up.

Where functional assays are either unfeasible or unethical, causality can be determined using a quantitative genetics approach to model how multiple measured traits are related to one another. Analyzing brain and behavioral data in pedigrees or full-sibling/half-sibling families allows the estimation of genetic correlations between traits (i.e., demonstrating variation in two traits that share a common genetic basis). If variation in brain size or composition causatively produces variation in behavior we should expect strong genetic correlations between these traits. This approach can be used not only to test brain–behavior relationships (e.g., Kotrschal et al., 2014), but also to help resolve debates about, for example, the relative roles of domain-general and domain-specific cognition (e.g., Pedersen, Plomin, Nesselroade, & McClearn, 1992), and developmental models of brain evolution (e.g., Hager, Lu, Rosen, & Williams, 2012; Noreikiene et al., 2015).

The Comparative Approach as a Tool for Generating Hypotheses and Testing Generality

Although we argue for increased emphasis on intraspecific studies to validate causative relationships, the comparative approach will remain an integral part of investigations of the evolution of brains and cognitive abilities, though their scope and design might change. Phylogenetic studies extend and inform detailed intraspecific studies, ideally leading to constant feedback that can enhance both (Figure 3). Continuously developing comparative approaches have the potential to reduce noise from small sample sizes, reveal relationships among multiple interfering traits, and indicate the directionality of a causal association—though not all at once. Combining findings from multiple populations can inform mechanistic studies by illustrating the range of possible solutions that might exist, indicating where natural experiments might have shaped evolution in similar ways, revealing potential mediators by indicating in which taxonomic groups established relationships break down, and showing which species to target for further study. In particular, the systematic combination of effect sizes from population studies in phylogenetic meta-analyses reduces noise and can test the robustness of an association between brain measures and behavior while revealing potential mediators that systematically change the form of the association in some populations or species (Nakagawa & Santos, 2012). For example, they might reveal whether the heritability of brain measures might depend on environmental variability.

In turn, the historical component of phylogenetic reconstructions extends population-level studies by revealing whether detected patterns are evolutionarily stable or lineage specific, and they can contribute to determining causal or adaptive relationships between traits by revealing temporal contingencies (Beaulieu, Jhwueng, Boettiger, & O’Meara, 2012; Pagel, 1999; Pagel & Meade, 2006) in whether a behavior consistently changed prior to or after associated changes in brain measures. The historical component provided by phylogenetic comparisons is necessary to determine whether traits not only occur together but also evolved together. For example, although the enlarged brains (compared to most other reptiles) among birds appear to be linked to cognitive capacities required for flight (Balanoff, Bever, Rowe, & Norell, 2013), evolutionary origins of flying behavior are not associated with particular increases in endocranial volume (Balanoff, Smaers, & Turner, 2016).

Our discussion of the power of the comparative approach in elucidating the adaptive history of traits indicates the inherent limits in fully explaining traits that supposedly make any species unique. The evolutionary processes themselves are not unique, but the particular combination of processes at play are. As such, understanding how such processes come together in a particular instance is problematic due to a lack of evidence required to confirm these hypotheses (Tucker, 1998). In addition, studies that focus on extraordinary traits in a single species (such as humans) frequently risk misrepresenting evolutionary processes by fixating on the endpoint as an optimal solution, whereas evolution typically progresses by responding to stochastic variation in selection regimes, incrementally adapting to the environment.

Scaling Across Taxa to Integrate Evidence

The bottom-up approach we suggest means that scaling across taxa will initially be more difficult to achieve because studies will have to be designed to take into account the characteristics of the particular species, as well as its phylogenetic and ecological context. Questions, approaches, and methods might need time to converge or to be repeated across a relevant sample of different taxa (Figure 3). However, over an intermediate time frame, the bottom-up approach will be invaluable for comparing and elucidating brain and cognitive evolution across taxa. Although the bottom-up approach initially makes scaling look difficult, we think it has two advantages. First, rather than positing or assuming a coarse-grained, cross-taxa category and applying it across a range of cases (thus losing ecological relevance and increasing the potential for post hoc explanations and reification), the bottom-up approach makes scaling a much more piecemeal, empirically tractable matter. Second, it more easily allows scaling to take place in an evolutionary context. Understanding whether the same genes, genetic pathways, neural regions, neural physiology, and/or neural networks are involved in generating cognitive abilities across taxa will provide us with an understanding of how evolution has shaped the diversity of brains and the behavior they produce. In this sense, phenotypic heterogeneity and taxonomic diversity become a tool for discovery rather than a source of statistical noise.

It is not straightforward to bring together the disparate evidence involved in shifting from local experimental contexts to cross-taxa, evolutionary hypotheses. However, a detailed understanding of the mechanisms underlying brain measures and behavior is crucial to clarify whether traits are homologous, analogous, or completely independent solutions to ecological challenges. To give a sense of the possibilities for integration, we sketch three kinds of approaches to shifting from local (bottom-up) to general (top-down) scales (from Currie, 2013; see also Mikhalevich et al., 2017):

1. Detecting homologous relationships, where the same brain measures and behavior are related to the same environment across species descended from a recent common ancestor (Currie, 2012), offers opportunities to combine independent findings into one mechanistic pathway. In these instances, inheritance and stabilizing selection have maintained a stable trait, such that findings from one species can be accurately inferred for another. Such investigations will rely on integrated models that bring together disparate evidence to support hypotheses about the evolutionary, developmental, and ecological features of a particular lineage.
2. Determining whether the same behavior occurs in similar environments across distantly related species can indicate environments most likely relevant for the emergence of the behavior. A bottom-up approach can reveal whether the observed behavior represents analogous reemergence of a behavior within the same adaptative environment (e.g., repeated evolution of feathers across dinosaurs; B. K. Hall, 2003; McGhee, 2011). This approach will rely on parallel models that identify brain–behavior correlations within related taxa for which the main principles of brain evolution are known to be similar. As closely related taxa will likely share meaningful brain–behavior correlations, such models are likely to be well validated, stable, and causally meaningful.
3. Observing a similar behavior in similar adaptive environments can reveal whether the behavior represents a convergent solution to the same environment (e.g., feathered wings for flight vs. bat wings) or whether the relationship is more complex (e.g., wings to escape into the air vs. jumping legs; Currie, 2014; Pearce, 2012; R. Powell, 2007). This type of convergent model is similar to the top-down approach; however, convergent modeling avoids many of the cross-taxa comparison problems by (a) being placed in an explicitly ecological and phylogenetic context, (b) being carried out alongside parallel and integrated modeling, and (c) avoiding overinterpretation that arises from defining categories based on superficial similarities because convergent models are inherently sensitive to the explanatory limits of analogous categories (see Griffiths, 1994, for a discussion of these limits).

We may need to infer evolutionary relationships between brain measures, behavior, and environments across taxa in a wide variety of different scales, and the ecological and evolutionary relevance granted by starting in local contexts is crucial for doing this.

Conclusion

We support a two-pronged strategy for understanding cross-taxa relationships between brain size, brain composition, behavior, and cognition that focuses on ecologically relevant contexts rather than attempting broad scale comparisons at gross phenotypic levels. The first prong is an experimental program examining correlations between behavior/cognition and brain structures at the smallest level at which variation can be detected and can be studied both from a mechanistic perspective (linking particular structures in the brain to behavior) and from a functional perspective (the ecological relevance of the behavior): Comparing individuals within species or from closely related species. The second prong involves the piecemeal identification of correlations at broader taxonomic scales. We have contrasted our approach with one that has become dominant in recent years. The alternative approach relies on coarse-grained phenotypes and proxy measures, typically in anthropocentric contexts, and attempts to apply these to cross-taxa, correlative contexts.

We have highlighted a number of limitations to this approach. First, applying anthropocentric conceptions of brain correlates with behavior to disparate taxa comes at the crucial cost of ecological and evolutionary coherence. Second, the heterogeneity of brain composition and behavior makes coarse-grained conceptions problematic because cross-taxa comparisons inevitably discount variation that matters for particular lineages. This variation creates noise in statistical comparisons. Heterogeneity can also be a source of interference because various interdependencies both between brain structures (e.g., in development or function) and between multiple behavioral and ecological traits undermine our capacity to identify selection pressures shaping individual traits or systems. Third, beginning with “general” measures of intelligence potentially leads to reification and the establishment of misguided or causally meaningless properties. The top-down approach has not necessarily been misguided itself: Scientific progress is often facilitated by applying relatively crude measures, highlighting the value of using many investigative techniques. Indeed, the heterogeneity of these traits have become known to us because the top-down approach has exposed inconsistencies through cross-species correlations. However, it is time to take the cognitive, behavioral, and brain features of particular lineages seriously, rather than demand that they be shoehorned into anthropocentric notions or judged against some general metric. In doing so, a more general understanding of the nature of cognition and behavior, and their relationship with brain measures, will be built from the bottom up.

References

1. Ahn, Y.-Y., Jeong, H., & Kim, B. J. (2006). Wiring cost in the organization of a biological neuronal network. Physica A: Statistical Mechanics and Its Applications, 367, 531–537. doi:10.1016/j.physa.2005.12.013

2. Aiello, L. C., & Wheeler, P. (1995). The expensive-tissue hypothesis: The brain and the digestive system in human and primate evolution. Current Anthropology, 36, 199–221. doi:10.1086/204350

3. Allman, J., McLaughlin, T., & Hakeem, A. (1993). Brain weight and life-span in primate species. Proceedings of the National Academy of Sciences, 90, 118–122. doi:10.1073/pnas.90.1.118

4. Allritz, M., Tennie, C., & Call, J. (2013). Food washing and placer mining in captive great apes. Primates, 54, 361–370. doi:10.1007/s10329-013-0355-5

5. Andics, A., Gábor, A., Gácsi, M., Faragó, T., Szabó, D., & Miklósi, Á. (2016). Neural mechanisms for lexical processing in dogs. Science, 353(6303), 1030–1032. doi:10.1126/science.aaf3777

6. Armstrong, E. (1983). Relative brain size and metabolism in mammals. Science, 220(4603), 1302–1304. doi:10.1126/science.6407108

7. Atchley, W. R. (1984). Ontogeny, timing of development, and genetic variance-covariances structure. The American Naturalist, 123, 519–540. doi:10.1086/284220

8. Auersperg, A. M. I., Bayern, A. M. P. von, Gajdon, G. K., Huber, L., & Kacelnik, A. (2011). Flexibility in problem solving and tool use of kea and New Caledonian crows in a multi access box paradigm. PLOS One, 6(6), e20231. doi:10.1371/journal.pone.0020231

9. Balanoff, A. M., Bever, G. S., Rowe, T. B., & Norell, M. A. (2013). Evolutionary origins of the avian brain. Nature, 501(7465), 93–96. doi:10.1038/nature12424

10. Balanoff, A. M., Smaers, J. B., & Turner, A. H. (2016). Brain modularity across the theropod–bird transition: Testing the influence of flight on neuroanatomical variation. Journal of Anatomy, 229, 204–214. doi:10.1111/joa.12403

11. Barton, R. A. (1998). Visual specialization and brain evolution in primates. Proceedings of the Royal Society of London B: Biological Sciences, 265(1409), 1933–1937. doi:10.1098/rspb.1998.0523

12. Barton, R. A. (2012). Embodied cognitive evolution and the cerebellum. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 367(1599), 2097–2107. doi:10.1098/rstb.2012.0112

13. Barton, R. A., & Capellini, I. (2011). Maternal investment, life histories, and the costs of brain growth in mammals. Proceedings of the National Academy of Sciences, 108, 6169–6174. doi:10.1073/pnas.1019140108

14. Barton, R. A., & Harvey, P. H. (2000). Mosaic evolution of brain structure in mammals. Nature, 405(6790), 1055–1058. doi:10.1038/35016580

15. Barton, R. A., Purvis, A., & Harvey, P. H. (1995). Evolutionary radiation of visual and olfactory brain systems in primates, bats and insectivores. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 348(1326), 381–392. doi:10.1098/rstb.1995.0076

16. Barton, R. A., & Venditti, C. (2014). Rapid evolution of the cerebellum in humans and other great apes. Current Biology, 24, 2440–2444. doi:10.1016/j.cub.2014.08.056

17. Beatty, J. (1994). The proximate/ultimate distinction in the multiple careers of Ernst Mayr. Biology and Philosophy, 9, 333–356. doi:10.1007/BF00857940

18. Beaulieu, J. M., Jhwueng, D.-C., Boettiger, C., & O’Meara, B. C. (2012). Modeling stabilizing selection: Expanding the Ornstein–Uhlenbeck model of adaptive evolution. Evolution, 66, 2369–2383. doi:10.1111/j.1558-5646.2012.01619.x

19. Bendesky, A., Kwon, Y.-M., Lassance, J.-M., Lewarch, C. L., Yao, S., Peterson, B. K., . . . Hoekstra, H. E. (2017). The genetic basis of parental care evolution in monogamous mice. Nature, 544(7651), 434–439. doi:10.1038/nature22074

20. Bengio, Y., Simard, P., & Frasconi, P. (1994). Learning long-term dependencies with gradient descent is difficult. IEEE Transactions on Neural Networks, 5, 157–166. doi:10.1109/72.279181

21. Benson-Amram, S., Dantzer, B., Stricker, G., Swanson, E. M., & Holekamp, K. E. (2016). Brain size predicts problem-solving ability in mammalian carnivores. Proceedings of the National Academy of Sciences, 113, 2532–2537. doi:10.1073/pnas.1505913113

22. Benson-Amram, S., & Holekamp, K. E. (2012). Innovative problem solving by wild spotted hyenas. Philosophical Transactions of the Royal Society of London B, 279(1744), 4087–4095. doi:10.1098/rspb.2012.1450

23. Bickart, K. C., Wright, C. I., Dautoff, R. J., Dickerson, B. C., & Barrett, L. F. (2011). Amygdala volume and social network size in humans. Nature Neuroscience, 14, 163–164. doi:10.1038/nn.2724

24. Bird, C. D., & Emery, N. J. (2009). Insightful problem solving and creative tool modification by captive nontool-using rooks. Proceedings of the National Academy of Sciences, 106(25), 10370–10375. doi:10.1073/pnas.0901008106

25. Bogart, S. L., Bennett, A. J., Schapiro, S. J., Reamer, L. A., & Hopkins, W. D. (2014). Different early rearing experiences have long-term effects on cortical organization in captive chimpanzees (Pan troglodytes). Developmental Science, 17, 161–174. doi:10.1111/desc.12106

26. Bond, A. B., Kamil, A. C., & Balda, R. P. (2007). Serial reversal learning and the evolution of behavioral flexibility in three species of North American corvids (Gymnorhinus cyanocephalus, Nucifraga columbiana, Aphelocoma californica). Journal of Comparative Psychology, 121, 372–379. doi:10.1037/0735-7036.121.4.372

27. Boogert, N. J., Anderson, R. C., Peters, S., Searcy, W. A., & Nowicki, S. (2011). Song repertoire size in male song sparrows correlates with detour reaching, but not with other cognitive measures. Animal Behaviour, 81, 1209–1216. doi:10.1016/j.anbehav.2011.03.004

28. Boogert, N. J., Monceau, K., & Lefebvre, L. (2010). A field test of behavioural flexibility in Zenaida doves (Zenaida aurita). Behavioural Processes, 85, 135–141. doi:10.1016/j.beproc.2010.06.020

29. Boogert, N. J., Reader, S. M., Hoppitt, W., & Laland, K. N. (2008). The origin and spread of innovations in starlings. Animal Behaviour, 75, 1509–1518. doi:10.1016/j.anbehav.2007.09.033

30. Brown, M. F., & Sayde, J. M. (2013). Same/different discrimination by bumblebee colonies. Animal Cognition, 16, 117–125. doi:10.1007/s10071-012-0557-z

31. Bullmore, E., & Sporns, O. (2012). The economy of brain network organization. Nature Reviews Neuroscience, 13, 336–349. doi:10.1038/nrn3214

32. Burkart, J. M., Schubiger, M. N., & van Schaik, C. P. (2016). The evolution of general intelligence. Behavioral and Brain Sciences, 40, e195. doi:10.1017/S0140525X16000959

33. Burns, J. G., Foucaud, J., & Mery, F. (2010). Costs of memory: Lessons from ‘mini’ brains. Proceedings of the Royal Society of London B: Biological Sciences, 278(1707). doi:10.1098/rspb.2010.2488

34. Butler, A. B., & Hodos, W. (2005). Comparative vertebrate neuroanatomy: Evolution and adaptation. Hoboken, NJ: Wiley & Sons.

35. Buzsáki, G., Geisler, C., Henze, D. A., & Wang, X.-J. (2004). Interneuron Diversity series: Circuit complexity and axon wiring economy of cortical interneurons. Trends in Neurosciences, 27, 186–193. doi:10.1016/j.tins.2004.02.007

36. Calcott, B. (2013). Why the proximate-ultimate distinction is misleading, and why it matters for understanding the evolution of cooperation. In K. Sterelny, R. Joyce, B. Calcott, & B. Fraser (Eds.), Cooperation and its evolution (pp. 249–264). Cambridge, MA: MIT Press.

37. Cartmill, M. (1972). Arboreal adaptations and the origin of the order primates. The Function and Evolutionary Biology of Primates, 97–122.

38. Cauchoix, M., & Chaine, A. S. (2016). How can we study the evolution of animal minds? Frontiers in Psychology, 7. doi:10.3389/fpsyg.2016.00358

39. Chancellor, L. V., Roth, T. C., LaDage, L. D., & Pravosudov, V. V. (2011). The effect of environmental harshness on neurogenesis: A large-scale comparison. Developmental Neurobiology, 71, 246–252. doi:10.1002/dneu.20847

40. Chen, B. L., Hall, D. H., & Chklovskii, D. B. (2006). Wiring optimization can relate neuronal structure and function. Proceedings of the National Academy of Sciences of the United States of America, 103, 4723–4728. doi:10.1073/pnas.0506806103

41. Chittka, L., & Niven, J. (2009). Are bigger brains better? Current Biology, 19(21), R995–R1008. doi:10.1016/j.cub.2009.08.023

42. Chittka, L., Rossiter, S. J., Skorupski, P., & Fernando, C. (2012). What is comparable in comparative cognition? Philosophical Transactions of the Royal Society of London B: Biological Sciences, 367(1603), 2677–2685. doi:10.1098/rstb.2012.0215

43. Clancy, B., Darlington, R. B., & Finlay, B. L. (2001). Translating developmental time across mammalian species. Neuroscience, 105, 7–17. doi:10.1016/S0306-4522(01)00171-3

44. Clayton, N. S., & Dickinson, A. (1998). Episodic-like memory during cache recovery by scrub jays. Nature, 395(6699), 272–274. doi:10.1038/26216

45. Clutton-Brock, T. H., & Harvey, P. H. (1980). Primates, brains and ecology. Journal of Zoology, 190, 309–323. doi:10.1111/j.1469-7998.1980.tb01430.x

46. Cnotka, J., Güntürkün, O., Rehkämper, G., Gray, R. D., & Hunt, G. R. (2008). Extraordinary large brains in tool-using New Caledonian crows (Corvus moneduloides). Neuroscience Letters, 433, 241–245. doi:10.1016/j.neulet.2008.01.026

47. Cross, F. R., & Jackson, R. R. (2014). Specialised use of working memory by Portia africana, a spider-eating salticid. Animal Cognition, 17, 435–444. doi:10.1007/s10071-013-0675-2

48. Cross, F. R., & Jackson, R. R. (2016). The execution of planned detours by spider-eating predators. Journal of the Experimental Analysis of Behavior, 105, 194–210. doi:10.1002/jeab.189

49. Cross, F. R., & Jackson, R. R. (2017). Representation of different exact numbers of prey by a spider-eating predator. Interface Focus, 7(3), 20160035. doi:10.1098/rsfs.2016.0035

50. Currie, A. (2012). Convergence, contingency & morphospace. Biology & Philosophy, 27, 583–593. doi:10.1007/s10539-012-9319-1

51. Currie, A. (2013). Convergence as evidence. The British Journal for the Philosophy of Science, 64, 763–786. doi:10.1093/bjps/axs027

52. Currie, A. (2014). Venomous dinosaurs and rear-fanged snakes: Homology and homoplasy characterized. Erkenntnis, 79(3), 701–727. doi:10.1007/s10670-013-9533-5

53. Currie, A., & Walsh, K. (2018). Newton on Islandworld: Ontic-Driven Explanations of Scientific Method. Perspectives on Science, 26(1), 119–156. doi:10.1162/POSC_a_00270

54. Dacke, M., & Srinivasan, M. V. (2008). Evidence for counting in insects. Animal Cognition, 11, 683–689. doi:10.1007/s10071-008-0159-y

55. Deaner, R. O., Barton, R. A., & van Schaik, C. P. (2003). Primate brains and life histories: Renewing the connection. In P. Kappeler & M. Pereira (Eds.), Primate life histories and socioecology (pp. 233–265). Chicago, IL: University of Chicago Press.

56. Deaner, R. O., Isler, K., Burkart, J., & van Schaik, C. (2007). Overall brain size, and not encephalization quotient, best predicts cognitive ability across non-human primates. Brain, Behavior and Evolution, 70, 115–124. doi:10.1159/000102973

57. Deaner, R. O., van Schaik, C. P., & Johnson, V. (2006). Do some taxa have better domain-general cognition than others? A meta-analysis of nonhuman primate studies. Evolutionary Psychology, 4(1). doi:10.1177/147470490600400114

58. DeCasien, A. R., Williams, S. A., & Higham, J. P. (2017). Primate brain size is predicted by diet but not sociality. Nature Ecology & Evolution, 1, 0112. doi:10.1038/s41559-017-0112

59. De Groof, G., Jonckers, E., Güntürkün, O., Denolf, P., Van Auderkerke, J., & Van der Linden, A. (2013). Functional MRI and functional connectivity of the visual system of awake pigeons. Behavioural Brain Research, 239, 43–50. doi:10.1016/j.bbr.2012.10.044

60. Ducatez, S., Clavel, J., & Lefebvre, L. (2015). Ecological generalism and behavioural innovation in birds: Technical intelligence or the simple incorporation of new foods? Journal of Animal Ecology, 84, 79–89. doi:10.1111/1365-2656.12255

61. Dunbar, R. I. M., & Shultz, S. (2007a). Evolution in the social brain. Science, 317(5843), 1344–1347. doi:10.1126/science.1145463

62. Dunbar, R. I. M., & Shultz, S. (2007b). Understanding primate brain evolution. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 362(1480), 649–658. doi:10.1098/rstb.2006.2001

63. Eberhard, W. G., & Wcislo, W. T. (2011). Grade changes in brain–body allometry: Morphological and behavioural correlates of brain size in miniature spiders, insects and other invertebrates. In J. Casas (Ed.), Advances in insect physiology (Vol. 40, pp. 155–214). London, UK: Academic Press. doi:10.1016/B978-0-12-387668-3.00004-0

64. Elliott-Graves, A. (2016). The problem of prediction in invasion biology. Biology & Philosophy, 31(3), 373–393. doi:10.1007/s10539-015-9504-0

65. Emery, N. J., & Clayton, N. S. (2004). Comparing the complex cognition of birds and primates. In L. J. Rogers & G. Kaplan (Eds.), Comparative vertebrate cognition (pp. 3–55). New York, NY: Springer. doi:10.1007/978-1-4419-8913-0_1

66. Enard, W., Gehre, S., Hammerschmidt, K., Hölter, S. M., Blass, T., Somel, M., . . . Pääbo, S. (2009). A humanized version of Foxp2 affects cortico-basal ganglia circuits in mice. Cell, 137, 961–971. doi:10.1016/j.cell.2009.03.041

67. Enard, W., Przeworski, M., Fisher, S. E., Lai, C. S. L., Wiebe, V., Kitano, T., . . . Pääbo, S. (2002). Molecular evolution of FOXP2, a gene involved in speech and language. Nature, 418(6900), 869–872. doi:10.1038/nature01025

68. Eriksen, N., & Pakkenberg, B. (2007). Total neocortical cell number in the mysticete brain. The Anatomical Record: Advances in Integrative Anatomy and Evolutionary Biology, 290(1), 83–95. doi:10.1002/ar.20404

69. Farris, S. M., & Schulmeister, S. (2011). Parasitoidism, not sociality, is associated with the evolution of elaborate mushroom bodies in the brains of hymenopteran insects. Proceedings of the Royal Society of London B: Biological Sciences, 278(1707), 940–951. doi:10.1098/rspb.2010.2161

70. Finlay, B. L., Darlington, R. B., & Nicastro, N. (2001). Developmental structure in brain evolution. Behavioral and Brain Sciences, 24, 263–278. doi:10.1017/S0140525X01003958

71. Fitzpatrick, J. L., Almbro, M., Gonzalez-Voyer, A., Hamada, S., Pennington, C., Scanlan, J., & Kolm, N. (2012). Sexual selection uncouples the evolution of brain and body size in pinnipeds. Journal of Evolutionary Biology, 25, 1321–1330. doi:10.1111/j.1420-9101.2012.02520.x

72. Freas, C. A., LaDage, L. D., Roth, T. C., II, & Pravosudov, V. V. (2012). Elevation-related differences in memory and the hippocampus in mountain chickadees, Poecile gambeli. Animal Behaviour, 84, 121–127. doi:10.1016/j.anbehav.2012.04.018

73. Garland, T., Bennett, A. F., & Rezende, E. L. (2005). Phylogenetic approaches in comparative physiology. Journal of Experimental Biology, 208, 3015–3035. doi:10.1242/jeb.01745

74. Glorot, X., & Bengio, Y. (2010). Understanding the difficulty of training deep feedforward neural networks. In Y. W. Teh & M. Titterington (Eds.), Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics (Vol. 9, pp. 249–256). Sardinia, Italy: PMLR. Retrieved from http://proceedings.mlr.press/v9/glorot10a.html

75. Gonda, A., Herczeg, G., & Merilä, J. (2011). Population variation in brain size of nine-spined sticklebacks (Pungitius pungitius)—Local adaptation or environmentally induced variation? BMC Evolutionary Biology, 11, 75. doi:10.1186/1471-2148-11-75

76. González-Lagos, C., Sol, D., & Reader, S. M. (2010). Large-brained mammals live longer. Journal of Evolutionary Biology, 23, 1064–1074. doi:10.1111/j.1420-9101.2010.01976.x

77. Gonzalez-Voyer, A., & Hardenberg, A. von. (2014). An Introduction to Phylogenetic Path Analysis. In L. Z. Garamszegi (Ed.), Modern phylogenetic comparative methods and their application in evolutionary biology (pp. 201–229). Berlin, Germany: Springer. doi:10.1007/978-3-662-43550-2_8

78. Gonzalez-Voyer, A., Winberg, S., & Kolm, N. (2009). Distinct evolutionary patterns of brain and body size during adaptive radiation. Evolution, 63, 2266–2274. doi:10.1111/j.1558-5646.2009.00705.x

79. Grasso, F. W., & Basil, J. A. (2009). The evolution of flexible behavioral repertoires in cephalopod molluscs. Brain, Behavior and Evolution, 74, 231–245. doi:10.1159/000258669

80. Greenwood, A. K., Wark, A. R., Yoshida, K., & Peichel, C. L. (2013). Genetic and neural modularity underlie the evolution of schooling behavior in threespine sticklebacks. Current Biology, 23, 1884–1888. doi:10.1016/j.cub.2013.07.058

81. Griffiths, P. E. (1994). Cladistic classification and functional explanation. Philosophy of Science, 61, 206–227. doi:10.1086/289796

82. Grinberg, L. T., Amaro, E., Teipel, S., Santos, D. D. dos, Pasqualucci, C. A., Leite, R. E. P., . . . Group, B. A. B. S. (2008). Assessment of factors that confound MRI and neuropathological correlation of human postmortem brain tissue. Cell and Tissue Banking, 9, 195. doi:10.1007/s10561-008-9080-5

83. Gronenberg, W., & Couvillon, M. J. (2010). Brain composition and olfactory learning in honey bees. Neurobiology of Learning and Memory, 93, 435–443. doi:10.1016/j.nlm.2010.01.001

84. Gupta, A. S., van der Meer, M. A. A., Touretzky, D. S., & Redish, A. D. (2010). Hippocampal replay is not a simple function of experience. Neuron, 65, 695–705. doi:10.1016/j.neuron.2010.01.034

85. Hager, R., Lu, L., Rosen, G. D., & Williams, R. W. (2012). Genetic architecture supports mosaic brain evolution and independent brain–body size regulation. Nature Communications, 3, 1079. doi:10.1038/ncomms2086

86. Hall, B. K. (2003). Descent with modification: The unity underlying homology and homoplasy as seen through an analysis of development and evolution. Biological Reviews, 78, 409–433. doi:10.1017/S1464793102006097

87. Hall, Z. J., Street, S. E., & Healy, S. D. (2013). The evolution of cerebellum structure correlates with nest complexity. Biology Letters, 9, 20130687. doi:10.1098/rsbl.2013.0687

88. Hampton, R. R., & Shettleworth, S. J. (1996). Hippocampal lesions impair memory for location but not color in passerine birds. Behavioral Neuroscience, 110, 831–835. doi:10.1037/0735-7044.110.4.831

89. Harvey, P. H., & Bennett, P. M. (1983). Evolutionary biology: Brain size, energetics, ecology and life history patterns. Nature, 306(5941), 314–315. doi:10.1038/306314a0

90. Harvey, P. H., & Pagel, M. D. (1991). The comparative method in evolutionary biology (Vol. 1). Oxford, UK: Oxford University Press.

91. Healy, S. D., & Rowe, C. (2007). A critique of comparative studies of brain size. Proceedings of the Royal Society of London B: Biological Sciences, 274(1609), 453–464. doi:10.1098/rspb.2006.3748

92. Herculano-Houzel, S. (2012). The remarkable, yet not extraordinary, human brain as a scaled-up primate brain and its associated cost. Proceedings of the National Academy of Sciences, 109(Supplement 1), 10661–10668. doi:10.1073/pnas.1201895109

93. Herculano-Houzel, S. (2017). Numbers of neurons as biological correlates of cognitive capability. Current Opinion in Behavioral Sciences, 16, 1–7. doi:10.1016/j.cobeha.2017.02.004

94. Herculano-Houzel, S., Catania, K., Manger, P. R., & Kaas, J. H. (2015). Mammalian brains are made of these: A dataset of the numbers and densities of neuronal and nonneuronal cells in the brain of glires, primates, scandentia, eulipotyphlans, afrotherians and artiodactyls, and their relationship with body mass. Brain, Behavior and Evolution, 86, 145–163. doi:10.1159/000437413

95. Herculano-Houzel, S., Collins, C. E., Wong, P., & Kaas, J. H. (2007). Cellular scaling rules for primate brains. Proceedings of the National Academy of Sciences, 104, 3562–3567. doi:10.1073/pnas.0611396104

96. Hochner, B., Shomrat, T., & Fiorito, G. (2006). The octopus: A model for a comparative analysis of the evolution of learning and memory mechanisms. The Biological Bulletin, 210, 308–317. doi:10.2307/4134567

97. Hopkins, W. D., Cantalupo, C., & Taglialatela, J. (2007). Handedness is associated with asymmetries in gyrification of the cerebral cortex of chimpanzees. Cerebral Cortex, 17, 1750–1756. doi:10.1093/cercor/bhl085

98. Hunt, G. R., & Gray, R. D. (2002). Species-wide manufacture of stick-type tools by New Caledonian crows. Emu-Austral Ornithology, 102, 349–353. doi:10.1071/mu01056

99. Isler, K. (2011). Energetic trade-offs between brain size and offspring production: Marsupials confirm a general mammalian pattern. BioEssays, 33, 173–179. doi:10.1002/bies.201000123

100. Isler, K., Christopher Kirk, E., Miller, J. M. A., Albrecht, G. A., Gelvin, B. R., & Martin, R. D. (2008). Endocranial volumes of primate species: Scaling analyses using a comprehensive and reliable data set. Journal of Human Evolution, 55, 967–978. doi:10.1016/j.jhevol.2008.08.004

101. Isler, K., & van Schaik, C. P. (2009). The expensive brain: A framework for explaining evolutionary changes in brain size. Journal of Human Evolution, 57, 392–400. doi:10.1016/j.jhevol.2009.04.009

102. Iwaniuk, A. N., & Nelson, J. E. (2002). Can endocranial volume be used as an estimate of brain size in birds? Canadian Journal of Zoology, 80, 16–23. doi:10.1139/z01-204

103. Iwaniuk, A. N., & Nelson, J. E. (2003). Developmental differences are correlated with relative brain size in birds: A comparative analysis. Canadian Journal of Zoology, 81, 1913–1928. doi:10.1139/z03-190

104. Jackson, R. R., & Carter, C. M. (2001). Geographic variation in reliance on trial-and-error signal derivation by Portia labiata, an araneophagic jumping spider from the Philippines. Journal of Insect Behavior, 14, 799–827. doi:10.1023/A:1013041501584

105. Jackson, R. R., Cross, F. R., & Carter, C. M. (2006). Geographic variation in a spider’s ability to solve a confinement problem by trial and error. International Journal of Comparative Psychology, 19, 282–296.

106. Jackson, R. R., & Li, D. (2004). One-encounter search-image formation by araneophagic spiders. Animal Cognition, 7, 247–254. doi:10.1007/s10071-004-0219-x

107. Jacobs, I. F., & Osvath, M. (2015). The string-pulling paradigm in comparative psychology. Journal of Comparative Psychology, 129, 89–120. doi:10.1037/a0038746

108. Jarvis, E. D., Güntürkün, O., Bruce, L., Csillag, A., Karten, H., Kuenzel, W., . . . Butler, A. B. (2005). Avian brains and a new understanding of vertebrate brain evolution. Nature Reviews Neuroscience, 6, 151–159. doi:10.1038/nrn1606

109. Jelbert, S. A., Taylor, A. H., Cheke, L. G., Clayton, N. S., & Gray, R. D. (2014). Using the Aesop’s Fable paradigm to investigate causal understanding of water displacement by New Caledonian crows. PLOS ONE, 9(3), e92895. doi:10.1371/journal.pone.0092895

110. Jelbert, S. A., Taylor, A. H., & Gray, R. D. (2015). Investigating animal cognition with the Aesop’s Fable paradigm: Current understanding and future directions. Communicative & Integrative Biology, 8, e1035846. doi:10.1080/19420889.2015.1035846

111. Jelbert, S. A., Taylor, A. H., & Gray, R. D. (2016). Does absolute brain size really predict self-control? Hand-tracking training improves performance on the A-not-B task. Biology Letters, 12, 20150871. doi:10.1098/rsbl.2015.0871

112. Jerison, H. J. (1969). Brain evolution and dinosaur brains. The American Naturalist, 103(934), 575–588. doi:10.1086/282627

113. Jerison, H. J. (1973). Evolution of the brain and intelligence. London, UK: Academic Press.

114. Jerison, H. J. (1985). Animal intelligence as encephalization [and discussion]. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 308(1135), 21–35. doi:10.1098/rstb.1985.0007

115. Kaas, J. H., & Collins, C. E. (2001). Variability in the sizes of brain parts. Behavioral and Brain Sciences, 24, 288–290. doi:10.1017/S0140525X01333952

116. Kabadayi, C., Taylor, L. A., Bayern, A. M. P. von, & Osvath, M. (2016). Ravens, New Caledonian crows and jackdaws parallel great apes in motor self-regulation despite smaller brains. Royal Society Open Science, 3, 160104. doi:10.1098/rsos.160104

117. Kanai, R., Bahrami, B., Roylance, R., & Rees, G. (2011). Online social network size is reflected in human brain structure. Proceedings of the Royal Society of London B: Biological Sciences, 279, 1327–1334. doi:10.1098/rspb.2011.1959

118. Kawai, M. (1965). Newly-acquired pre-cultural behavior of the natural troop of Japanese monkeys on Koshima islet. Primates, 6, 1–30. doi:10.1007/BF01794457

119. Kotrschal, A., Buechel, S. D., Zala, S. M., Corral-Lopez, A., Penn, D. J., & Kolm, N. (2015). Brain size affects female but not male survival under predation threat. Ecology Letters, 18, 646–652. doi:10.1111/ele.12441

120. Kotrschal, A., Deacon, A. E., Magurran, A. E., & Kolm, N. (2017). Predation pressure shapes brain anatomy in the wild. Evolutionary Ecology, 1–15. doi:10.1007/s10682-017-9901-8

121. Kotrschal, A., Lievens, E. J., Dahlbom, J., Bundsen, A., Semenova, S., Sundvik, M., . . . Kolm, N. (2014). Artificial selection on relative brain size reveals a positive genetic correlation between brain size and proactive personality in the guppy. Evolution, 68, 1139–1149. doi:10.1111/evo.12341

122. Kotrschal, A., Rogell, B., Bundsen, A., Svensson, B., Zajitschek, S., Brännström, I., . . . Kolm, N. (2013). Artificial selection on relative brain size in the guppy reveals costs and benefits of evolving a larger brain. Current Biology, 23, 168–171. doi:10.1016/j.cub.2012.11.058

123. Kowalko, J. E., Rohner, N., Rompani, S. B., Peterson, B. K., Linden, T. A., Yoshizawa, M., . . . Tabin, C. J. (2013). Loss of schooling behavior in cavefish through sight-dependent and sight-independent mechanisms. Current Biology, 23, 1874–1883. doi:10.1016/j.cub.2013.07.056

124. Krebs, J. R., Clayton, N. S., Healy, S. D., Cristol, D. A., Patel, S. N., & Jolliffe, A. R. (1996). The ecology of the avian brain: Food-storing memory and the hippocampus. Ibis, 138, 34–46. doi:10.1111/j.1474-919X.1996.tb04311.x

125. LaBarbera, M. (1986). The evolution and ecology of body size. In D. Raup & D. Jablonski (Eds.), Patterns and processes in the history of life (Vol. 36, pp. 69–98). Berlin, Germany: Springer. doi:10.1007/978-3-642-70831-2_5

126. Lai, C. S. L., Fisher, S. E., Hurst, J. A., Vargha-Khadem, F., & Monaco, A. P. (2001). A forkhead-domain gene is mutated in a severe speech and language disorder. Nature, 413(6855), 519–523. doi:10.1038/35097076

127. Laland, K. N., Odling-Smee, J., Hoppitt, W., & Uller, T. (2013). More on how and why: Cause and effect in biology revisited. Biology & Philosophy, 28, 719–745. doi:10.1007/s10539-012-9335-1

128. Laland, K. N., Sterelny, K., Odling-Smee, J., Hoppitt, W., & Uller, T. (2011). Cause and effect in biology revisited: Is Mayr’s proximate-ultimate dichotomy still useful? Science, 334(6062), 1512–1516. doi:10.1126/science.1210879

129. Larson, T. A., Thatra, N. M., Lee, B. H., & Brenowitz, E. A. (2014). Reactive neurogenesis in response to naturally occurring apoptosis in an adult brain. Journal of Neuroscience, 34, 13066–13076. doi:10.1523/JNEUROSCI.3316-13.2014

130. Larson, T. A., Wang, T.-W., Gale, S. D., Miller, K. E., Thatra, N. M., Caras, M. L., . . . Brenowitz, E. A. (2013). Postsynaptic neural activity regulates neuronal addition in the adult avian song control system. Proceedings of the National Academy of Sciences, 110, 16640–16644. doi:10.1073/pnas.1310237110

131. Laughlin, S. B., de Ruyter van Steveninck, R. R., & Anderson, J. C. (1998). The metabolic cost of neural information. Nature Neuroscience, 1, 36–41. doi:10.1038/236

132. Lefebvre, L., Whittle, P., Lascaris, E., & Finkelstein, A. (1997). Feeding innovations and forebrain size in birds. Animal Behaviour, 53, 549–560. doi:10.1006/anbe.1996.0330

133. Legg, S., & Hutter, M. (2007). Universal intelligence: A definition of machine intelligence. Minds and Machines, 17, 391–444. doi:10.1007/s11023-007-9079-x

134. Lewis, J. E., DeGusta, D., Meyer, M. R., Monge, J. M., Mann, A. E., & Holloway, R. L. (2011). The mismeasure of science: Stephen Jay Gould versus Samuel George Morton on skulls and bias. PLoS Biology, 9(6), e1001071. doi:10.1371/journal.pbio.1001071

135. Logan, C. J. (2016a). Behavioral flexibility and problem solving in an invasive bird. PeerJ, 4, e1975. doi:10.7717/peerj.1975

136. Logan, C. J. (2015, unpublished). Innovation does not indicate behavioral flexibility in great-tailed grackles. bioRxiv, 027706.

137. Logan, C. J. (2016b). How far will a behaviourally flexible invasive bird go to innovate? Royal Society Open Science, 3, 160247. doi:10.1098/rsos.160247

138. Logan, C. J., Harvey, B. D., Schlinger, B. A., & Rensel, M. (2016). Western scrub-jays do not appear to attend to functionality in Aesop’s Fable experiments. PeerJ, 4, e1707. doi:10.7717/peerj.1707

139. Logan, C. J., Jelbert, S. A., Breen, A. J., Gray, R. D., & Taylor, A. H. (2014). Modifications to the Aesop’s Fable paradigm change New Caledonian crow performances. PLOS One, 9(7), e103049. doi:10.1371/journal.pone.0103049

140. Logan, C. J., Kruuk, L. E. B., Stanley, R., Thompson, A. M., & Clutton-Brock, T. H. (2016). Endocranial volume is heritable and is associated with longevity and fitness in a wild mammal. Royal Society Open Science, 3(12), 160622. doi:10.1098/rsos.160622

141. MacLean, E. L., Hare, B., Nunn, C. L., Addessi, E., Amici, F., Anderson, R. C., . . . Zhao, Y. (2014). The evolution of self-control. Proceedings of the National Academy of Sciences, 111(20), E2140–E2148. doi:10.1073/pnas.1323533111

142. Macphail, E. M. (1982). Brain and intelligence in vertebrates. Oxford, UK: Clarendon Press.

143. Manrique, H. M., Völter, C. J., & Call, J. (2013). Repeated innovation in great apes. Animal Behaviour, 85, 195–202. doi:10.1016/j.anbehav.2012.10.026

144. Mars, R. B., Neubert, F.-X., Verhagen, L., Sallet, J., Miller, K. L., Dunbar, R. I. M., & Barton, R. A. (2014). Primate comparative neuroscience using magnetic resonance imaging: Promises and challenges. Frontiers in Neuroscience, 8. doi:10.3389/fnins.2014.00298

145. Maseko, B. C., Spocter, M. A., Haagensen, M., & Manger, P. R. (2012). Elephants have relatively the largest cerebellum size of mammals. The Anatomical Record: Advances in Integrative Anatomy and Evolutionary Biology, 295, 661–672. doi:10.1002/ar.22425

146. Matthewson, J. (2011). Trade-offs in model-building: A more target-oriented approach. Studies in History and Philosophy of Science Part A, 42, 324–333. doi:10.1016/j.shpsa.2010.11.040

147. McAuliffe, K., & Thornton, A. (2015). The psychology of cooperation in animals: an ecological approach. Journal of Zoology, 295, 23–35. doi:10.1111/jzo.12204

148. McGhee, G. R. (2011). Convergent evolution: Limited forms most beautiful. Cambridge, MA: MIT Press. doi:10.7551/mitpress/9780262016421.001.0001

149. Meitzen, J., Thompson, C. K., Choi, H., Perkel, D. J., & Brenowitz, E. A. (2009). Time course of changes in Gambel’s white-crowned sparrow song behavior following transitions in breeding condition. Hormones and Behavior, 55, 217–227. doi:10.1016/j.yhbeh.2008.10.006

150. Menzel, R., & Giurfa, M. (2001). Cognitive architecture of a mini-brain: The honeybee. Trends in Cognitive Sciences, 5, 62–71. doi:10.1016/S1364-6613(00)01601-6

151. Meyer, W., & Idel, K. (1977). The distribution of acetylcholinesterase in the central nervous system of jumping spiders and wolf spiders (Arachnida, Araneida: Salticidae et Lycosidae). The Journal of Comparative Neurology, 173, 717–743. doi:10.1002/cne.901730406

152. Mikhalevich, I., Powell, R., & Logan, C. (2017). Is behavioural flexibility evidence of cognitive complexity? How evolution can inform comparative cognition. Interface Focus, 7(3), 20160121. doi:10.1098/rsfs.2016.0121

153. Miller, K. L., Stagg, C. J., Douaud, G., Jbabdi, S., Smith, S. M., Behrens, T. E. J., . . . McNab, J. A. (2011). Diffusion imaging of whole, post-mortem human brains on a clinical MRI scanner. NeuroImage, 57, 167–181. doi:10.1016/j.neuroimage.2011.03.070

154. Miller, R., Jelbert, S. A., Taylor, A. H., Cheke, L. G., Gray, R. D., Loissel, E., & Clayton, N. S. (2016). Performance in object-choice Aesop’s Fable tasks are influenced by object biases in New Caledonian crows but not in human children. PLOS ONE, 11(12), e0168056. doi:10.1371/journal.pone.0168056

155. Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., . . . Hassabis, D. (2015). Human-level control through deep reinforcement learning. Nature, 518(7540), 529–533. doi:10.1038/nature14236

156. Montgomery, S. H. (2017). Evolution of large brain and body size in mammals. In J. H. Kaas (Ed.), Evolution of nervous systems (2nd ed., pp. 103–136). Oxford, UK: Academic Press. doi:10.1016/B978-0-12-804042-3.00034-8

157. Montgomery, S. H., Capellini, I., Barton, R. A., & Mundy, N. I. (2010). Reconstructing the ups and downs of primate brain evolution: Implications for adaptive hypotheses and Homo floresiensis. BMC Biology, 8, 9. doi:10.1186/1741-7007-8-9

158. Montgomery, S. H., Geisler, J. H., McGowen, M. R., Fox, C., Marino, L., & Gatesy, J. (2013). The evolutionary history of cetacean brain and body size. Evolution, 67, 3339–3353. doi:10.1111/evo.12197

159. Montgomery, S. H., & Merrill, R. M. (2017). Divergence in brain composition during the early stages of ecological specialization in Heliconius butterflies. Journal of Evolutionary Biology, 30, 571–582. doi:10.1111/jeb.13027

160. Montgomery, S. H., & Mundy, N. I. (2013). Microcephaly genes and the evolution of sexual dimorphism in primate brain size. Journal of Evolutionary Biology, 26, 906–911. doi:10.1111/jeb.12091

161. Mulcahy, N. J., & Call, J. (2006). How great apes perform on a modified trap-tube task. Animal Cognition, 9, 193–199. doi:10.1007/s10071-006-0019-6

162. Nair, A., Srinivasan, P., Blackwell, S., Alcicek, C., Fearon, R., De Maria, A., . . . Silver, D. (2015, July). Massively parallel methods for deep reinforcement learning. Paper presented at the Deep Learning Workshop, International Conference on Machine Learning, Lille, France. Retrieved from http://arxiv.org/abs/1507.04296

163. Nakagawa, S., & Santos, E. S. A. (2012). Methodological issues and advances in biological meta-analysis. Evolutionary Ecology, 26, 1253–1274. doi:10.1007/s10682-012-9555-5

164. Navarrete, A. F., Reader, S. M., Street, S. E., Whalen, A., & Laland, K. N. (2016). The coevolution of innovation and technical intelligence in primates. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 371(1690), 20150186. doi:10.1098/rstb.2015.0186

165. Noreikiene, K., Herczeg, G., Gonda, A., Balázs, G., Husby, A., & Merilä, J. (2015). Quantitative genetic analysis of brain size variation in sticklebacks: Support for the mosaic model of brain evolution. Proceedings of the Royal Society of London B: Biological Sciences, 282(1810), 20151008. doi:10.1098/rspb.2015.1008

166. Nyberg, D. (1971). A hypothesis concerning the larger brains of homoiotherms. The American Naturalist, 105(942), 183–185. doi:10.1086/282713

167. O’Hara, M., Huber, L., & Gajdon, G. K. (2015). The advantage of objects over images in discrimination and reversal learning by kea, Nestor notabilis. Animal Behaviour, 101, 51–60. doi:10.1016/j.anbehav.2014.12.022

168. Olkowicz, S., Kocourek, M., Lučan, R. K., Porteš, M., Fitch, W. T., Herculano-Houzel, S., & Němec, P. (2016). Birds have primate-like numbers of neurons in the forebrain. Proceedings of the National Academy of Sciences, 113(26), 7255–7260. doi:10.1073/pnas.1517131113

169. O’Regan, H. J., & Kitchener, A. C. (2005). The effects of captivity on the morphology of captive, domesticated and feral mammals. Mammal Review, 35, 215–230. doi:10.1111/j.1365-2907.2005.00070.x

170. Overington, S. E., Morand-Ferron, J., Boogert, N. J., & Lefebvre, L. (2009). Technical innovations drive the relationship between innovativeness and residual brain size in birds. Animal Behaviour, 78, 1001–1010. doi:10.1016/j.anbehav.2009.06.033

171. Packard, A. (1972). Cephalopods and fish: The limits of convergence. Biological Reviews, 47, 241–307. doi:10.1111/j.1469-185X.1972.tb00975.x

172. Pagel, M. (1999). Inferring the historical patterns of biological evolution. Nature, 401(6756), 877–884. doi:10.1038/44766

173. Pagel, M., & Meade, A. (2006). Bayesian analysis of correlated evolution of discrete characters by reversible-jump Markov chain Monte Carlo. The American Naturalist, 167, 808–825. doi:10.1086/503444

174. Park, P. J., & Bell, M. A. (2010). Variation of telencephalon morphology of the threespine stickleback (Gasterosteus aculeatus) in relation to inferred ecology. Journal of Evolutionary Biology, 23, 1261–1277. doi:10.1111/j.1420-9101.2010.01987.x

175. Pascanu, R., Mikolov, T., & Bengio, Y. (2013). On the difficulty of training recurrent neural networks. In S. Dasgupta & D. McAllester (Eds.), Proceedings of the 30th International Conference on Machine Learning (Vol. 28, pp. 1310–1318). Atlanta, GA: PMLR. Retrieved from http://proceedings.mlr.press/v28/pascanu13.html

176. Patel, S. N., Clayton, N. S., & Krebs, J. R. (1997). Hippocampal tissue transplants reverse lesion-induced spatial memory deficits in zebra finches (Taeniopygia guttata). Journal of Neuroscience, 17, 3861–3869.

177. Pawłowskil, B., Lowen, C. B., & Dunbar, R. I. M. (1998). Neocortex size, social skills and mating success in primates. Behaviour, 135(3), 357–368. doi:10.1163/156853998793066285

178. Pearce, T. (2012). Convergence and parallelism in evolution: A neo-Gouldian account. The British Journal for the Philosophy of Science, 63, 429–448. doi:10.1093/bjps/axr046

179. Pedersen, N. L., Plomin, R., Nesselroade, J. R., & McClearn, G. E. (1992). A quantitative genetic analysis of cognitive abilities during the second half of the life span. Psychological Science, 3, 346–353. doi:10.1111/j.1467-9280.1992.tb00045.x

180. Petraitis, P. S., Dunham, A. E., & Niewiarowski, P. H. (1996). Inferring multiple causality: The limitations of path analysis. Functional Ecology, 10, 421–431. doi:10.2307/2389934

181. Pfenning, A. R., Hara, E., Whitney, O., Rivas, M. V., Wang, R., Roulhac, P. L., . . . Jarvis, E. D. (2014). Convergent transcriptional specializations in the brains of humans and song-learning birds. Science, 346(6215), 1256846. doi:10.1126/science.1256846

182. Powell, L. E., Isler, K., & Barton, R. A. (2017). Re-evaluating the link between brain size and behavioural ecology in primates. Proceedings of the Royal Society of London B: Biological Sciences, 284(1865), 20171765. doi:10.1098/rspb.2017.1765

183. Powell, R. (2007). Is convergence more than an analogy? Homoplasy and its implications for macroevolutionary predictability. Biology & Philosophy, 22, 565–578. doi:10.1007/s10539-006-9057-3

184. Pravosudov, V. V., & Clayton, N. S. (2002). A test of the adaptive specialization hypothesis: Population differences in caching, memory, and the hippocampus in black-capped chickadees (Poecile atricapilla). Behavioral Neuroscience, 116, 515–522. doi:10.1037/0735-7044.116.4.515

185. Pravosudov, V. V., & de Kort, S. R. (2006). Is the western scrub-jay (Aphelocoma californica) really an underdog among food-caching corvids when it comes to hippocampal volume and food caching propensity? Brain, Behavior and Evolution, 67, 1–9. doi:10.1159/000088855

186. Pritchard, D. J., Hurly, T. A., Tello–Ramos, M. C., & Healy, S. D. (2016). Why study cognition in the wild (and how to test it)? Journal of the Experimental Analysis of Behavior, 105, 41–55. doi:10.1002/jeab.195

187. Punzo, F., & Ludwig, L. (2002). Contact with maternal parent and siblings affects hunting behavior, learning, and central nervous system development in spiderlings of Hogna carolinensis (Araeneae: Lycosidae). Animal Cognition, 5, 63–70. doi:10.1007/s10071-002-0128-9

188. Ramnani, N. (2006). The primate cortico-cerebellar system: Anatomy and function. Nature Reviews Neuroscience, 7, 511–522. doi:10.1038/nrn1953

189. Reader, S. M., Hager, Y., & Laland, K. N. (2011). The evolution of primate general and cultural intelligence. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 366(1567), 1017–1027. doi:10.1098/rstb.2010.0342

190. Reader, S. M., & Laland, K. N. (2002). Social intelligence, innovation, and enhanced brain size in primates. Proceedings of the National Academy of Sciences, 99, 4436–4441. doi:10.1073/pnas.062041299

191. Ridgway, S. H., Carlin, K. P., Alstyne, K. R. V., Hanson, A. C., & Tarpley, R. J. (2016). Comparison of dolphins’ body and brain measurements with four other groups of cetaceans reveals great diversity. Brain, Behavior and Evolution, 88, 235–257. doi:10.1159/000454797

192. Ringo, J. L. (1991). Neuronal interconnection as a function of brain size. Brain, Behavior and Evolution, 38, 1–6. doi:10.1159/000114375

193. Rodríguez, R. L., Briceño, R. D., Briceño-Aguilar, E., & Höbel, G. (2015). Nephila clavipes spiders (Araneae: Nephilidae) keep track of captured prey counts: testing for a sense of numerosity in an orb-weaver. Animal Cognition, 18, 307–314. doi:10.1007/s10071-014-0801-9

194. Rosati, A. G. (2017). Foraging cognition: Reviving the ecological intelligence hypothesis. Trends in Cognitive Sciences, 21, 691–702. doi:10.1016/j.tics.2017.05.011

195. Roth, G. (2013). The long evolution of brains and minds. Heidelberg, Germany: Springer. doi:10.1007/978-94-007-6259-6

196. Roth, T. C., & Pravosudov, V. V. (2009). Hippocampal volumes and neuron numbers increase along a gradient of environmental harshness: A large-scale comparison. Proceedings of the Royal Society of London B: Biological Sciences, 276(1656), 401–405. doi:10.1098/rspb.2008.1184

197. Rowe, C., & Healy, S. D. (2014). Measuring variation in cognition. Behavioral Ecology, 25, 1287–1292. doi:10.1093/beheco/aru090

198. Rubinstein, H. S. (1936). The effect of the growth hormone upon the brain and brain weight-Body weight relations. The Journal of Comparative Neurology, 64, 469–496. doi:10.1002/cne.900640308

199. Sallet, J., Mars, R. B., Noonan, M. P., Andersson, J. L., O’Reilly, J. X., Jbabdi, S., . . . Rushworth, M. F. S. (2011). Social network size affects neural circuits in macaques. Science, 334(6056), 697–700. doi:10.1126/science.1210027

200. Schönbrodt, F. D., & Perugini, M. (2013). At what sample size do correlations stabilize? Journal of Research in Personality, 47, 609–612. doi:10.1016/j.jrp.2013.05.009

201. Schreiweis, C., Bornschein, U., Burguière, E., Kerimoglu, C., Schreiter, S., Dannemann, M., . . . Graybiel, A. M. (2014). Humanized Foxp2 accelerates learning by enhancing transitions from declarative to procedural performance. Proceedings of the National Academy of Sciences, 111, 14253–14258. doi:10.1073/pnas.1414542111

202. Scott-Phillips, T. C., Dickins, T. E., & West, S. A. (2011). Evolutionary theory and the ultimate–proximate distinction in the human behavioral sciences. Perspectives on Psychological Science, 6, 38–47. doi:10.1177/1745691610393528

203. Seed, A. M., Tebbich, S., Emery, N. J., & Clayton, N. S. (2006). Investigating physical cognition in rooks, Corvus frugilegus. Current Biology, 16, 697–701. doi:10.1016/j.cub.2006.02.066

204. Shettleworth, S. J. (2010). Cognition, evolution, and behavior (2nd ed., Vol. 13). New York, NY: Oxford University Press.

205. Shigeno, S. (2017). Brain evolution as an information flow designer: The ground architecture for biological and artificial general intelligence. In S. Shigeno, Y. Murakami, & T. Nomura (Eds.), Brain evolution by design (pp. 415–438). Tokyo, Japan: Springer. doi:10.1007/978-4-431-56469-0_19

206. Shultz, S., & Dunbar, R. I. M. (2010). Species differences in executive function correlate with hippocampus volume and neocortex ratio across nonhuman primates. Journal of Comparative Psychology, 124, 252–260. doi:10.1037/a0018894

207. Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., van den Driessche, G., . . . Hassabis, D. (2016). Mastering the game of Go with deep neural networks and tree search. Nature, 529(7587), 484–489. doi:10.1038/nature16961

208. Slater, G. J., Price, S. A., Santini, F., & Alfaro, M. E. (2010). Diversity versus disparity and the radiation of modern cetaceans. Proceedings of the Royal Society B, 277(1697), 3097–3104. doi:10.1098/rspb.2010.0408

209. Smaers, J. B., & Soligo, C. (2013). Brain reorganization, not relative brain size, primarily characterizes anthropoid brain evolution. Proc. Roy. Soc. B, 280(1759), 20130269. doi:10.1098/rspb.2013.0269

210. Snell, O. (1892). Die Abhängigkeit des Hirngewichtes von dem Körpergewicht und den geistigen Fähigkeiten (German, Dependence of brain weight on body weight and the intellectual capacity). European Archives of Psychiatry and Clinical Neuroscience, 23, 436–446. doi:10.1007/BF01843462

211. Snell-Rood, E. C., Papaj, D. R., & Gronenberg, W. (2009). Brain size: A global or induced cost of learning? Brain, Behavior and Evolution, 73, 111–128. doi:10.1159/000213647

212. Sol, D., & Lefebvre, L. (2000). Behavioural flexibility predicts invasion success in birds introduced to New Zealand. Oikos, 90, 599–605. doi:10.1034/j.1600-0706.2000.900317.x

213. Sol, D., Székely, T., Liker, A., & Lefebvre, L. (2007). Big-brained birds survive better in nature. Proceedings of the Royal Society of London B: Biological Sciences, 274(1611), 763–769. doi:10.1098/rspb.2006.3765

214. Sol, D., Timmermans, S., & Lefebvre, L. (2002). Behavioural flexibility and invasion success in birds. Animal Behaviour, 63, 495–502. doi:10.1006/anbe.2001.1953

215. Strausfeld, N. J., & Hirth, F. (2013). Deep homology of arthropod central complex and vertebrate basal ganglia. Science, 340(6129), 157–161. doi:10.1126/science.1231828

216. Swaddle, J. P. (2016). Evolution and Conservation Behavior. In O. Berger-Tal & D. Saltz (Eds.), Conservation behavior: Applying behavioral ecology to wildlife conservation and management (Vol. 21, pp. 36–65). Cambridge, UK: Cambridge University Press. doi:10.1017/CBO9781139627078.004

217. Tarsitano, M. S., & Jackson, R. R. (1997). Araneophagic jumping spiders discriminate between detour routes that do and do not lead to prey. Animal Behaviour, 53, 257–266. doi:10.1006/anbe.1996.0372

218. Tebbich, S., Sterelny, K., & Teschke, I. (2010). The tale of the finch: Adaptive radiation and behavioural flexibility. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 365(1543), 1099–1109. doi:10.1098/rstb.2009.0291

219. Thornton, A., Isden, J., & Madden, J. R. (2014). Toward wild psychometrics: Linking individual cognitive differences to fitness. Behavioral Ecology, 25, 1299–1301. doi:10.1093/beheco/aru095

220. Thornton, A., & Lukas, D. (2012). Individual variation in cognitive performance: Developmental and evolutionary perspectives. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 367(1603), 2773–2783. doi:10.1098/rstb.2012.0214

221. Thornton, A., & Samson, J. (2012). Innovative problem solving in wild meerkats. Animal Behaviour, 83, 1459–1468. doi:10.1016/j.anbehav.2012.03.018

222. Tomasello, M. (1999). The cultural origins of human cognition. Cambridge, MA: Harvard University Press.

223. Tomer, R., Denes, A. S., Tessmar-Raible, K., & Arendt, D. (2010). Profiling by image registration reveals common origin of annelid mushroom bodies and vertebrate pallium. Cell, 142, 800–809. doi:10.1016/j.cell.2010.07.043

224. Tramontin, A. D., & Brenowitz, E. A. (2000). Seasonal plasticity in the adult brain. Trends in Neurosciences, 23(6), 251–258. doi:10.1016/S0166-2236(00)01558-7

225. Tucker, A. (1998). Unique events: The underdetermination of explanation. Erkenntnis, 48, 59–80. doi:10.1023/A:1005315532171

226. van der Bijl, W., Thyselius, M., Kotrschal, A., & Kolm, N. (2015). Brain size affects the behavioural response to predators in female guppies (Poecilia reticulata). Proc. R. Soc. B, 282(1812), 20151132. doi:10.1098/rspb.2015.1132

227. van Schaik, C. P., & Burkart, J. M. (2011). Social learning and evolution: The cultural intelligence hypothesis. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 366(1567), 1008–1016. doi:10.1098/rstb.2010.0304

228. van Schaik, C. P., Isler, K., & Burkart, J. M. (2012). Explaining brain size variation: From social to cultural brain. Trends in Cognitive Sciences, 16(5), 277–284. doi:10.1016/j.tics.2012.04.004

229. Ventura-Antunes, L., Mota, B., & Herculano-Houzel, S. (2013). Different scaling of white matter volume, cortical connectivity, and gyrification across rodent and primate brains. Frontiers in Neuroanatomy, 7. doi:10.3389/fnana.2013.00003

230. Walsh, M. R., Broyles, W., Beston, S. M., & Munch, S. B. (2016). Predator-driven brain size evolution in natural populations of Trinidadian killifish (Rivulus hartii). Proceedings of the Royal Society of London B: Biological Sciences, 283(1834), 20161075. doi:10.1098/rspb.2016.1075

231. Wardill, T. J., Fabian, S. T., Pettigrew, A. C., Stavenga, D. G., Nordström, K., & Gonzalez-Bellido, P. T. (2017). A novel interception strategy in a miniature robber fly with extreme visual acuity. Current Biology, 27, 854–859. doi:10.1016/j.cub.2017.01.050

232. Watts, D. J., & Strogatz, S. H. (1998). Collective dynamics of “small-world” networks. Nature, 393(6684), 440–442. doi:10.1038/30918

233. Weber, J. N., Peterson, B. K., & Hoekstra, H. E. (2013). Discrete genetic modules are responsible for complex burrow evolution in Peromyscus mice. Nature, 493(7432), 402–405. doi:10.1038/nature11816

234. Whitehead, A. N. (1925). The origins of modern science. In A. I. Tauber (Ed.), Science and the quest for reality (pp. 53–69). New York, NY: Palgrave Macmillan. doi:10.1007/978-1-349-25249-7_2

235. Whiting, B. A., & Barton, R. A. (2003). The evolution of the cortico-cerebellar complex in primates: Anatomical connections predict patterns of correlated evolution. Journal of Human Evolution, 44, 3–10. doi:10.1016/S0047-2484(02)00162-8

236. Workman, A. D., Charvet, C. J., Clancy, B., Darlington, R. B., & Finlay, B. L. (2013). Modeling transformations of neurodevelopmental sequences across mammalian species. Journal of Neuroscience, 33, 7368–7383. doi:10.1523/JNEUROSCI.5746-12.2013

237. Wyles, J. S., Kunkel, J. G., & Wilson, A. C. (1983). Birds, behavior, and anatomical evolution. Proceedings of the National Academy of Sciences, 80, 4394–4397. doi:10.1073/pnas.80.14.4394

238. Zhang, S., Bock, F., Si, A., Tautz, J., & Srinivasan, M. V. (2005). Visual working memory in decision making by honey bees. Proceedings of the National Academy of Sciences of the United States of America, 102, 5250–5255. doi:10.1073/pnas.0501440102

239. Zilles, K., Amunts, K., & Smaers, J. B. (2011). Three brain collections for comparative neuroanatomy and neuroimaging. Annals of the New York Academy of Sciences, 1225(S1), E94–E104. doi:10.1111/j.1749-6632.2011.05978.x

More Situated Cognition in Animals: Reply to Commentators

Ken Cheng

Department of Biological Sciences
Macquarie University

Reading Options

---

Abstract

The commentators added a number of strands of discussion that expanded on embodied, extended, enactive, and distributive cognition across the animal kingdom and indeed beyond. I thank all the authors and continue the discourse in this reply. Action routines in the form of movements might form a common part of information delivery in perceptual systems; this means that sensory organs do not have to be highly acute in their entirety. Sensory systems and matched filters, outside of the central brain, seem to carry on some computations, exemplifying embodied cognition including some morphological computation. In addition to ratbots, animals operating neuroprosthetic devices make another kind of cyborg exhibiting extended and distributed cognition. Further examples of distributed cognition in the form of collective intelligence are presented, in humans and other animals, including a looming brand of trans-kingdom distributed cognition revolving around the gut bacteria of animals, now known to affect cognition even though the cognitive mechanisms remain unclear. All in all, the topic of situated cognition in animals looks even richer, and further dialogue is welcome.

Keywords: action, neuroprosthetic, morphological computation, distributed cognition

Author Note: Ken Cheng, Department of Biological Sciences, Macquarie University, Sydney, NSW 2109 Australia.

Correspondence concerning this article should be addressed to Ken Cheng at ken.cheng@mq.edu.au

Acknowledgments: I would like to thank the commentators for their interesting and thought provoking comments.

---

I would like to thank the commentators for their insights on situated cognition in animals, including humans. Without exception, they have continued and expanded the discussion, in a number of directions including comparative cognition, artificial intelligence, cognitive science, and philosophy. In reply, I am doing more of the same, continuing the fibers of discourse with ideas that came to me as I read the astute commentaries. And again, as with the target article from which the comments were generated (Cheng, 2018, this issue), this reply amounts to less than a full-fledged position and more a continuing dialogue.

Focus on Action and Behavior

I applaud and thank Pritchard (2018) for stressing the importance of observing and detailing behaviors and actions beyond those that end up in spreadsheets as dependent measures. Pritchard presents various scanning and looking routines of birds as well as insects. Such action routines might well form enactive cognition of the form that I described (Cheng, 2018), in which action supplements computations in central cognition. Indeed, in perceptual systems, actions often serve in information delivery. Primate vision and the tactile sense in the star-nosed mole furnish excellent additional cases.

It is textbook knowledge that the primate retina consists of a central part, the fovea, packed densely with receptors, including many cones that provide the sensory basis for color vision (Johnson, 2012). The periphery around the fovea has a lower density of receptors and less acute visual resolution. If something catches a primate’s attention in peripheral vision, the primate’s strategy is to move the eyes or make saccades to put the fovea on the object of interest. Action is crucial for conveying important visual information.

The star-nosed mole (Figure

 1) exhibits a parallel principle in its tactile sense. The mole sports a pinkish fleshy protuberance on its face called the star nose. But the star nose is not a nose, not containing any chemoreceptors (Catania, 1999). Catania’s (1999) title conveys the story: a nose that looks like a hand and acts like an eye. The star nose is a mechanoreceptor par excellence, brandishing 11 rays on each side, fleshy bits looking like fingers. One of those rays is especially sensitive and is called the tactile fovea, in deliberate analogy to the primate visual system (Catania, 1999; Catania & Kaas, 1997; Catania & Remple, 2004). When a nonfoveal ray brushes against something of interest, the mole moves its star nose to foveate, placing the tactile fovea on the object. The fast-moving star nose buys its owner efficiency in foraging. Along with primate vision, action routines save the animals from having to stack high-density neural representation in the entire eye or for the entire star nose. The word “for” is necessary in the latter case because the foveal ray does not contain a higher density or number of receptors, in contrast to the primate’s visual fovea, but still hogs a disproportionate chunk of neural space in cortical representation. Active perception may harbor a large dose of enactive cognition of the form that I described (Cheng, 2018).

Cyborgs and the Brave New World of Neuroprosthetics

Brown (2018) describes animal–machine hybrid intelligence as a new—at least evolutionarily new—form of situated cognition (Wu et al., 2016; Yu et al., 2016; comment: Brown & Brown, 2017). In these studies, rats’ learning of a navigational task was supplemented by an artificial intelligence system feeding rats’ brains, via implants, information about what to do. Cyborg intelligence now also flows in the other direction, with brains directly telling machines what to do, again via implants, often in the motor cortex of primates (Capogrosso et al., 2016; Velliste, Perel, Spalding, Whitford, & Schwartz, 2008), including humans (Collinger et al., 2012; Hochberg et al., 2006; Wodlinger et al., 2014). Signals from the motor cortex move a cursor on a monitor (Hochberg et al., 2006) or operate robotic arms (Collinger et al., 2012; Velliste et al., 2008; Wodlinger et al., 2014). In Capogrosso et al. (2016), signals from monkeys’ motor cortex command the monkeys’ own limb after a spinal lesion via a remote computer receiving and giving signals via Bluetooth. The monkeys needed to carry only a small receiver and pulse generator that fit in a coat pocket. Besides raising questions in the philosophy of mind concerning extended and distributed cognition of a brave new kind, work on neuroprosthetic devices also raises a host of ethical concerns (Clausen et al., 2017).

Situated Cognition in Humans

I thank Lavoie et al. (2018) for enlightening us on situated cognition in humans. Besides the more typical form of human embodied cognition, in which aspects of action or thinking about action participate in cognition, Lavoie et al. present interesting cases of cognition in the peripheral nervous system as well, the sense attributed to some arm movements in the octopus (Cheng, 2018). And I thank Hewitson, Kaplan, and Sutton (2018) for distinguishing between morphology that cuts down on computation, which could be called decomputation, and morphology that actually contributes computation, for which the term morphological computation is truly apt. It seems to me that cases that mix embodied cognition with morphological computation are found in certain matched filters (in Wehner’s, 1987, sense). Japyassú and Laland (2017) explicitly differentiated matched filters from their cases of extended cognition because central cognition as a rule does not influence the operation of matched filters, but matched filters can serve as cases of embodied cognition. Some of them rely on morphology to help with computations. It is again textbook knowledge that in the primate eye, horizontal cells carry out lateral inhibition to sharpen up contrast at visual edges (Johnson, 2012). I would count lateral inhibition as a bit of computing, and morphology, the arrangement of cells in this case, plays a role in the computation. The horizontal cell spreads over a number of bipolar cells, and the anatomical arrangement means that horizontal cells inhibit bipolar cells nearby; the anatomy defines the range of inhibition. In all kinds of foveas as well, anatomical factors, whether in the density of receptors in primate eyes or density of fibers to the cortex in the star nose of the star-nosed mole (Catania, 1999), contribute to cognition.

In perusing literature on counting in trying to come up with an example of morphological computation on a larger anatomical scale, I came across authors claiming that even numerical cognition, seemingly an abstract mathematical realm, is embodied (R. A. Carlson, Avraamides, Cary, & Strasberg, 2007; Fischer, 2018). Thus, bodily actions such as pointing, or even nodding, improve counting in humans (R. A. Carlson et al., 2007). Such findings also constitute enactive cognition in that seemingly unrelated actions are helping out cognitive processes. With regard to morphological computation, perhaps the human use of fingers in counting might serve as an example. In counting in thought—imagine, for instance, tallying the number of U.S. presidents in the 20th century—one might raise a finger for each person thought up in order to relieve working memory of having to hold the current count.

A form of distributed human cognition that comes to mind from reading Lavoie et al. (2018) is transactive memory (Wegner, 1987). In transactive memory, the memory load is shared among a team of multiple humans, such as intimate couples, so that the total knowledge store of the team far exceeds what each member of the team knows. Besides intimate couples, Wegner also discussed departments of organizations, patient–doctor pairs, and teacher–student pairs as examples of transactive memories at work. Distributed cognition is widespread in current human cultures, and may well have helped to shape the course of human evolution in contributing to cumulative culture (Richerson & Boyd, 2005), on which a few more comments follow in the next section.

Musings on Other Philosophical Issues

I thank Hewitson et al. (2018) for elucidating and straightening out some philosophical loose ends, for broadening the topics of morphological computation and the mutual manipulability criterion, and for their interesting historical account. Especially illuminating are their further developments of the concept of mutual manipulability. These points are helpful in guiding empirical research. I thank Theiner (2018) for expertly explicating a variety of distributed cognition using interesting examples. This topic was perhaps the most underdone in the target article (Cheng, 2018), and a few more illustrations are worth adding.

Leafcutter ants not only are astounding in having invented agriculture tens of millions of years ago and for building “civilization by instinct” (Hölldobler & Wilson, 2011) but also exhibit an interesting form of teamwork in carrying leaves back. The leaves are used to feed fungus in their farms. The worker ants come in multiple castes and greatly vary in size. Big leafcutting workers possess strong muscles and mandibles for cutting foliage. In hauling back a piece of leaf, the carrier also totes a small minor along. One service that the hitchhiker provides is fly swatting, keeping pesky and destructive parasitic flies at bay. This is impossible to accomplish when a leafcutter is hoisting her hefty luggage. Cognitive and muscular work are thus both distributed across two animals, and such distributed cognition works to maintain the enormous ecological impact that this clade wields.

Distributed cognition may also have cumulative effects over a stretch of time, leading to cumulative culture. In cumulative culture, practices improve over generations, with new generations advancing what older generations achieved. Cumulative culture is well known in humans (Richerson & Boyd, 2005), but its occurrence in other species is uncertain. Recently, Sasaki and Biro (2017; see also Biro, Sasaki, & Portugal, 2016) showed what they called cumulative culture in the homing performance of pairs of homing pigeons. Sasaki and Biro measured homing efficiency in single pigeons or pairs of homing pigeons repeatedly released from an initially unfamiliar site. In the interesting experimental group that showed cumulative improvement, bird A homed singly for 12 trips; then A was paired with a new bird, B, for 12 trips; then A was retired and B was paired with a new bird, C, for 12 trips; then C and D flew 12 trips; then D and E flew 12 trips. The two control groups consisted of single pigeons homing repeatedly for 60 trips, or pairs of pigeons performing the trip 60 times. The exciting result was that the experimental group with a substitution at each generation (12 trips) improved over generations, whereas the control groups reached a plateau and stayed there. The improvements were found at the end of each generation for the substitute (experimental) group; at the start of each generation, this group got worse. This last finding is not surprising considering that one of two team members was naïve at the start of each generation. Nevertheless, in each generation the substitute group ended up improving to a higher level than the control groups’ performance level. It would have been good to carry the experiment on for longer, as it appeared from the data that the experimental group had not yet reached asymptote. If this phenomenon shows cumulative culture, as the authors maintained, it adds a new perspective to the evolution of behavior and cognition.

My final flight of philosophical musing foreshadows a form of distributed cognition looming on the horizon. Gut bacteria are now known to affect cognition via multiple routes, and thought to influence syndromes such as anxiety, autism spectrum disorders, depression (de la Fuente-Nunez, Meneguetti, Franco, & Lu, 2017; Mason, 2017), and, most recently, even cognitive development (A. L. Carlson et al., 2018). This topic is still looming in that we are still unsure of how gut bacteria are affecting cognitive processes, although a number of physiological routes have been suggested. We do have evidence that human microbiota has changed in history, including large changes in modern times (Gillings, Paulsen, & Tetu, 2015; Smits et al., 2017), so that investigation of this issue holds much practical and philosophical importance. I thus echo Theiner (2018) in alerting us to cross-kingdom distributed cognition.

In conclusion, distributed cognition of the collective-intelligence variety should be added to the list, in addition to distributed cognition of the reduced-brain variety and to the three e’s of cognition: embodied, extended, and enactive cognition. The comments show that this entire topic of situated cognition is richer than what was painted in my target article (Cheng, 2018) and promises even more ideas and research leads. Let the dialogue continue.

References

1. Biro, D., Sasaki, T., & Portugal, S. J. (2016). Bringing a time–depth perspective to collective animal behaviour. Trends in Ecology & Evolution, 31, 550–562. doi:10.1016/j.tree.2016.03.018

2. Brown, M. F. (2018). Musings on comparative directions for situated cognition. Comparative Cognition & Behavior Reviews, 13, 21–24. doi:10.3819/CCBR.2018.130002

3. Brown, M. F., & Brown, A. A. (2017). The promise of cyborg intelligence. Learning & Behavior, 45, 5-6.

4. Capogrosso, M., Milekovic, T., Borton, D., Wagner, F., Moraud, E. M., Mignardot, J.-B., … Courtine, G. (2016). A brain–spine interface alleviating gait deficits after spinal cord injury in primates. Nature, 539, 284–288. doi:10.1038/nature20118

5. Carlson, A. L., Xia, K., Azcarate-Peril, M. A., Goldman, B. D., Ahn, M., Styner, M. A., … Knickmeyer, R. C. (2018). Infant gut microbiome associated with cognitive development. Biological Psychiatry, 83, 148–159. doi:10.1016/j.biopsych.2017.06.021

6. Carlson, R. A., Avraamides, M. N., Cary, M., & Strasberg, S. (2007). What do the hands externalize in simple arithmetic? Journal of Experimental Psychology: Learning, Memory, and Cognition, 33, 747–756. doi:10.1037/0278-7393.33.4.747

7. Catania, K. C. (1999). A nose that looks like a hand and acts like an eye: The unusual mechanosensory system of the star-nosed mole. Journal of Comparative Physiology A, 185, 367–372. doi:10.1007/s003590050396

8. Catania, K. C., & Kaas, J. H. (1997). Somatosensory fovea in the star-nosed mole: Behavioral use of the star in relation to innervation patterns and cortical representation. Journal of Comparative Neurology, 387, 213–233. doi:10.1002/(SICI)1096-9861(19971020)387:2<215::AID-CNE4>3.0.CO;2-3

9. Catania, K. C., & Remple, F. E. (2004). Tactile foveation in the star-nosed mole. Brain, Behavior and Evolution, 63, 1–12. doi:10.1159/000073755

10. Cheng, K. (2018). Cognition beyond representation: Varieties of situated cognition in animals. Comparative Cognition & Behavior Reviews, 13, 1–20. doi:10.3819/CCBR.2018.130001

11. Clausen, J., Fetz, E., Donoghue, J., Ushiba, J., Spörhase, U., Chandler, J., … Soekadar, S. R. (2017). Help, hope, and hype: Ethical dimensions of neuroprosthetics. Science, 356, 1338–1339. doi:10.1126/science.aam7731

12. Collinger, J. L., Wodlinger, B., Downey, J. E., Wang, W., Tyler-Kabara, E. C., Weber, D. J., … Schwartz, A. B. (2012). High-performance neuroprosthetic control by an individual with tetraplegia. Lancet, 381, 557–564. doi:10.1016/s0140-6736(12)61816-9

13. de la Fuente-Nunez, C., Meneguetti, B. T., Franco, O. L., & Lu, T. K. (2017). Neuromicrobiology: How microbes influence the brain. ACS Chemical Neuroscience, 9, 141–150. doi:10.1021/acschemneuro.7b00373

14. Fischer, M. H. (2018). Why numbers are embodied concepts. Frontiers in Psychology, 8, 2347. doi:10.3389/fpsyg.2017.02347

15. Gillings, M. R., Paulsen, I. T., & Tetu, S. G. (2015). Ecology and evolution of the human microbiota: Fire, farming and antibiotics. Genes, 6, 841–857. doi:10.3390/genes6030841

16. Hewitson, C., Kaplan, D. M., & Sutton, J. (2018). Yesterday the earwig, today man, tomorrow the earwig? Comparative Cognition & Behavior Reviews, 13, 25–30. doi:10.3819/CCBR.2018.130003

17. Hochberg, L. R., Serruya, M. D., Friehs, G. M., Mukland, J. A., Saleh, M., Caplan, A. H., … Donoghue, J. P. (2006). Neuronal ensemble control of prosthetic devices by a human with tetraplegia. Nature, 442, 164–171. doi:10.1038/nature04970

18. Hölldobler, B., & Wilson, E. O. (2011). The leafcutter ants. New York, London: W. W. Norton.

19. Japyassú, H. F., & Laland, K. N. (2017). Extended spider cognition. Animal Cognition, 20, 375–395. doi:10.1007/s10071-017-1069-7

20. Johnson, G. B. (2012). The living world (7th ed.). New York, NY: McGraw-Hill.

21. Lavoie, E. B., Bertrand, J. K., Stone, S. A., Wispinski, N. J., Sawalha, J., & Chapman, C. S. (2018). Examining the “species” of situated cognition in humans. Comparative Cognition & Behavior Reviews, 13, 31–34. doi:10.3819/CCBR.2018.130004

22. Mason, B. L. (2017). Feeding systems and the gut microbiome: Gut-brain interactions with relevance to psychiatric conditions. Psychosomatics, 58, 574–580. doi:10.1016/j.psym.2017.06.002

23. Pritchard, D. J. (2018). Situated cognition and the function of behaviour: a commentary on Cheng (2018). Comparative Cognition & Behavior Reviews, 13, 35–40. doi:10.3819/CCBR.2018.130005

24. Richerson, P. J., & Boyd, R. (2005). Not by genes alone. Chicago, IL: University of Chicago Press.

25. Sasaki, T., & Biro, D. (2017). Cumulative culture can emerge from collective intelligence in animal groups. Nature Communications, 8, 15049. doi:10.1038/ncomms15049

26. Smits, S. A., Leach, J., Sonnenburg, E. D., Gonzalez, C. G., Lichtman, J. S., Reid, G., … Elias, J. E. (2017). Seasonal cycling in the gut microbiome of the Hadza hunter-gatherers of Tanzania. Science, 357, 802–806. doi:10.1126/science.aan4834

27. Theiner, G. (2018). Collaboration, exploitation, and distributed animal cognition. Comparative Cognition & Behavior Reviews, 13, 41–48. doi:10.3819/CCBR.2018.130006

28. Velliste, M., Perel, S., Spalding, M. C., Whitford, A. S., & Schwartz, A. B. (2008). Cortical control of a prosthetic arm for self-feeding. Nature, 453, 1098–1101. doi:10.1038/nature06996

29. Wegner, D. M. (1987). Transactive memory: A contemporary analysis of the group mind. In B. Mullen & G. R. Goethals (Eds.), Theories of group behavior (pp. 185–208). New York, NY: Springer. doi:10.1007/978-1-4612-4634-3_9

30. Wehner, R. (1987). “Matched filters”—Neural models of the external world. Journal of Comparative Physiology A, 161, 511–531. doi:10.1007/BF00603659

31. Wodlinger, B., Downey, J. E., Tyler-Kabara, E. C., Schwartz, A. B., Boninger, M. L., & Collinger, J. L. (2014). Ten-dimensional anthropomorphic arm control in a human brain−machine interface: Difficulties, solutions, and limitations. Journal of Neural Engineering, 12, 016011. doi:10.1088/1741-2560/12/1/016011

32. Wu, Z., Zheng, N., Zhang, S., Zheng, X., Gao, L., & Su, L. (2016). Maze learning by a hybrid brain–computer system. Scientific Reports, 6, 31746. doi:10.1038/srep31746

33. Yu, Y., Pan, G., Gong, Y., Xu, K., Zheng, N., Hua, W., … Wu, Z. (2016). Intelligence-augmented rat cyborgs in maze solving. PLoS One, 11, e0147754. doi:10.1371/journal.pone.0147754

Collaboration, Exploitation, and Distributed Animal Cognition

Georg Theiner

Villanova University

Reading Options

---

Abstract

In this commentary, I explore the space of “distributed cognition” across human and nonhuman animal cognition. First, I distinguish between three varieties in which cognition can be socially distributed and consider their respective implications for the conjectured relationship between group size (social complexity) and individual brain size (cognitive complexity). Second, I probe the relationship between distributed (collaborative) and extended (exploitative) cognition in contexts where our anthropomorphic understanding of this distinction begins to fade.

Keywords: distributed cognition, extended cognition, collective intelligence, collective intentionality, anthropomorphism

Author Note: Georg Theiner, Department of Philosophy, St. Augustine Center of the Liberal Arts 172, Villanova University, 800 E. Lancaster Ave., Villanova, PA 19085, USA.

Correspondence concerning this article should be addressed to Georg Theiner at georg.theiner@villanova.edu

---

As a card-carrying proponent of “4E cognition” approaches to cognitive science, which strive to understand mind and cognition as embodied, embedded, extended, and enactive phenomena, I am grateful to Ken Cheng’s erudite and insightful undertaking to bring animal cognition under the ambit of 4E cognition. In his own taxonomy, he ditches the “embedded” strand but adds the important facet of distributed cognition (DC), for which again I am thankful, because a concern with DC as a central feat of human cognition has been at the center of my own philosophical work. With my commentary, I seek to explore the space of DC across human and nonhuman animal cognition and probe the relationship between distributed and extended cognition in contexts where our anthropomorphic understanding of this distinction begins to fade.

My first observation concerns the conjectured relationship between group size (social complexity) and individual brain size (cognitive complexity). Assuming a rather straightforward correlation between the two, we are told that the DC hypothesis (at least as applied to eusocial species; cf. O’Donnell et al., 2015) and the social brain (SB) hypothesis (Dunbar, 1998) yield conflicting predictions about the direction of this correlation (p. 3). As Cheng recounts the debate, DC generally predicts that brain size is expected to decrease with group size, because larger groups afford a greater differentiation of cognitive labor. Because this is assumed to reduce the cognitive load on any given member of the group, they can get by with smaller nervous systems (thereby cutting down metabolic costs). This prediction is said to stand in contrast to SB, which asserts a positive correlation between group size and brain size, exacted by the greater cognitive demands of sociality in primate modes of group living, for example, reciprocal altruism and coalition formation, but also resource competition and social deception.

I have my doubts that “Cheng’s conjecture” about DC (as I refer to it) holds at the suggested level of generality if we consider the full gamut of “socially distributed cognition.” Without a careful analysis of cognitive task requirements, the assertion of any straightforward correlation between DC and brain size (or, more generously, individual cognitive complexity) must be viewed with suspicion. To begin with, whenever we analyze socially distributed cognition, we have to take into account two kinds of cognitive labor: first, the cognition required of individuals to perform the task at hand, and second, the cognition required to coordinate the task-specific cognitions that are distributed across individuals (Hutchins, 1995). Clearly, whatever organizational regimes are installed to regulate the division of labor greatly impact the kinds of cognitive tasks that individuals have to perform in particular group settings, which in turn affect the coordination requirements for putting them back together (cf. Davies & Michaelian, 2016; Goldstone & Gureckis, 2009; Theiner, 2017). I now describe three types of socially distributed cognition—each of which I illustrate with a concrete example—that differ qualitatively among each other with respect to the relationship between social and individual cognitive complexity, and thus bear directly on the plausibility of Cheng’s conjecture.

My example of DC1 is the ability of schooling fish—golden shiners, in this case—to track surrounding patterns of light by virtue of swarm intelligence (Berdahl, Torney, Ioannou, Faria, & Couzin, 2013). Being able to sense environmental gradients such as light, temperature, or salinity is a vital necessity for migrating animals, generally taken to require significant amounts of cognitive sampling and comparison, integrated over time. Single shiners, it turns out, are underachievers in this area. When exposed to shifting patterns of light, single fish perform only marginally better than random at staying in their preferred habitat, which is shaded (dark) waters. Incapable of tracking gradients, they follow a rudimentary, nondirectional rule to “swim slower when it’s dark here.” But in addition, shiners have a strong social instinct to stay close to their neighbors. As a result, if a few of them hit a darker patch and thus slow down, the rest of the shoal swivel into the shade; once inside, they all slow down, bunching up within the darkest region.

Important to note, a shiner’s decision to move is influenced far more by social than environmental cues. Although single shiners perform only local, scalar measurements, the collective gradient-tracking ability of the shoal emerges from the social dynamics of local interactions, the accuracy of which substantially increases as a function of group size. Applied to behaviorally equivalent instances of DC1, Cheng’s conjecture has a lot of purchase. Conceived as an emergent group-only feat, collective sensing has the advantage of being a fault-tolerant, cost-effective strategy that poses only minimal cognitive demands on the individual. From an evolutionary perspective, viewing the behavior of the collective as an adaptation to compensate for individual cognitive limitations, we should expect that the benefits reaped from higher level information processing relax the selection pressures on individuals’ cognitive abilities.

Let me contrast the preceding case, then, with a distinct kind of DC2 that we find exemplified in the house-hunting abilities of honeybees (cf. Seeley, 1995, 2010, for reviews). In particular, I am interested here in the mechanisms by which the hive negotiates the speed–accuracy trade-off to maximize its chances of choosing the best available nest site. In this process, a search committee of several hundreds of “scout bees” is sent out, roaming the surrounding area in search of potential targets. Upon return to the hive, they draw others’ attention to good sites they have discovered by performing a “waggle dance”; the dance orientation indicates the site location, and the duration indicates the site quality. Initially, the search is random, but as a result of observing other bees’ dancing, scouts are more likely to investigate attractive sites advertised by others. If, after inspection of the site, they agree with the assessment, a scout will join the dance, which further increases support for popular sites. Once a certain threshold (“consensus”) is reached, the hive as a whole decides to move there. Does the observed distribution (DC2) of cognitive labor lend support to Cheng’s conjecture?

The answer is not as straightforward as in the first case if we break down the cognitive task requirements of nest choice as a collective decision-making problem. Following List and Vermeule (2014), we ought to distinguish between the “epistemic agenda-setting” stage, during which a collective settles on a range of options among which it will eventually decide, and the “stage of choice” in which this decision is made. For the hive to succeed, each individual agent (scout bee) must be able to (a) roam the space of possible options and identify noteworthy candidates, (b) make comparative assessments and rank these options, and (c) communicate to one another which options are worth considering. As List and Vermeule showed, a speedy consensus for high-quality sites can be reliably reached only with the right admixture between interdependence at the agenda-setting stage and independence at the voting stage. The independence condition implies that each scouting bee must be able to assess the quality of a site it comes across and share that information in a way that positively (albeit imperfectly) correlates with that quality. The strength of this correlation can be taken to represent the individual bee’s cognitive competence.

Unlike the collective sensing of golden shiners, where the shoal solves a problem that no fish is individually capable of cognizing, the cognitive competence displayed by individual scout bees lies in the same task domain as the decision at which the hive collectively arrives. The “wisdom of the hive” here stems mainly from pooling imperfect individual estimates of nest site quality in ways that effectively cancel out one another’s errors (cf. Simons, 2004). This outcome is formally related to Condorcet’s jury theorem, which states that majority rule will lead a group to choose the best option in such cases provided that individual judgments are positively correlated with the objectively best choice, and mutually independent (e.g., Grofman, Owen, & Feld, 1983). With the right blend of interdependence and independence (see earlier), moderate levels of individual cognitive competence (e.g., performing the waggle dance) are sufficient to secure the desired outcome (List & Vermeule, 2014). Still, the competence condition puts a lower bound on the cognitive complexity of individual bees that is necessary for the hive to succeed. This makes nest choice in honeybees, as a type of DC2, a scenario for which the validity of Cheng’s conjecture is not evidently true.

My third and final example are collaborative interactions that go beyond basic cooperation insofar as they require distinctive forms of “we-thinking” or joint intentionality over and above individual intentionality (cf. Jankovic & Ludwig, 2017; Tomasello, 2014). Individual intentionality, as I use the term here, refers to the suite of cognitive competencies for engaging in flexible, goal-directed, individually self-regulated behavior. Consider tool use. Compared to other primates, great apes are especially skillful in making and using an open-ended variety of tools in insightful, often creative ways. For example, chimpanzees have been observed using both stone (or wooden) cleavers and stone anvils to fracture large fruit into smaller, bite-sized portions (Koops & McGrew, 2010). This implies an instrumental understanding of how cleavers have to be wielded to pound the fruit and how fruit has to be pounded on the rocky outcrops that serve as anvils. More generally, individual intentionality involves the ability to represent causally and intentionally relevant features of a situation, choose actions that lead to fulfillment of one’s goals, and self-monitor and evaluate specific behavioral outcomes vis-à-vis those goals.

Joint intentionality, as understood here, is an “upgraded” form of individual intentionality, repurposed for the collaborative pursuit of shared goals with a division of labor, often involving role specialization. A canonical model of collaboration in this sense is the “stag hunt” scenario known from game theory (Skyrms, 2004). Famously discussed by Rousseau, it describes a group of two (or more) individuals going out on a hunt. Each can get a hare by herself, but a hare is worth less than a stag, which the hunters can get only by joining forces, thereby incurring the risk that others might defect. Shared collaborative activities are cognitively more demanding than basic forms of cooperation or altruistic behavior. They require of each participant the ability to represent a goal that “we” aim to fulfill working together (rather than in parallel), with “you” and “me” simultaneously playing different but complementary parts, in mutually responsive ways, and an at least implicit understanding that “our” roles are (in principle) interchangeable. As Tomasello (2014) and others have argued, the coordinative and communicative requirements for joint intentionality depend on advanced psychological abilities and motivational propensities that are different in kind from those associated with individual intentionality. For example, to jointly engage in collaborative activities, participants need to attend to and conceptualize one and the same situation under different, perhaps even conflicting, perspectives (“This is how it must look from your point”) and draw specific kinds of socially recursive inferences (“Given our shared goals, you’d expect me to think this is what you’ll do”). Collaborators must also monitor and regulate their behavior with respect to the normative standards of the group, with a shared commitment to uphold their parts in the process, such as helping one’s partner if necessary, sharing the spoils, and so forth.

Somewhat controversially, Tomasello has argued that joint intentionality is a species-specific human cognitive trait that is not found among nonhuman primates. I won’t weigh in on this ongoing debate here, although it seems a safe bet that neither golden shiners nor honeybees have the cognitive prerequisites for joint intentionality. For present purposes, suffice it to note that the ability to engage in shared collaborative tasks requires, but at the same time enables, varieties of socially distributed cognition (DC3) that go against Cheng’s conjecture. In this respect, DC3 concurs with SB that we should expect a positive correlation between increases in social and cognitive complexity. However, the difference is that Tomasello and colleagues have placed a unique emphasis on collaboration and cooperative communication (“Vygotskian intelligence”; cf. Moll & Tomasello, 2007) as evolutionary drivers of joint intentionality, as displayed by humans, as opposed to the evolutionary focus on social competition and instrumentalist manipulation within the “Machiavellian intelligence” tradition (Byrne & Whiten, 1988; Humphrey, 1976).

Thus far, in my attempt to disambiguate the multifaceted notion of DC, I have taken for granted that all of the preceding varieties of DC are deployed in the service of group-level tasks. Indeed, it is hardly controversial to subsume the collective sensing of a shoal, the foraging and relocation patterns of honeybees, and the shared collaborative practices of humans under the banner of collectively intelligent behavior, in the fairly modest sense that intelligent groups are responsive to environmental contingencies; exhibit a division of labor; and can adapt to novel situations in flexible, goal-directed manners. As Cheng is well aware (p. 3), this somewhat conservative notion of collective intelligence does not exactly upset the Cartesian apple cart unless it is further argued that certain collectives form cognitive systems in their own right, with emergent cognitive states and processes that are distinct from those of their members.

Although there is a long tradition of comparisons between social insects and human beings, referring to not only their social but also cognitive organization (cf. Hofstadter, 1979), there has been a flurry of recent work on “colony-level cognition” (Marshall & Franks, 2009) that has brought to light deep structural correspondences between the ways in which brains and insect colonies gather, integrate, and process information (cf. Couzin, 2009; Seeley, 2010). For example, there are common mathematical models of the physically diverse mechanisms that underlie, for example, both migrating choices of ants and honeybees and motion discrimination of the primate visual cortex. Adopting this more liberal, functionally oriented perspective on DC holds great promises for unifying the study of individual and collective cognitive systems in revealing ways (Huebner, 2013; Theiner, 2017; Theiner, Allen, & Goldstone, 2010). From a network-theoretic perspective, Goldstone and Theiner (2017) reviewed a number of cognitive mechanisms involved in perception, attention, memory, and problem solving that have been attested at both individual and group levels, to argue for a “nonzero sum” perspective relating multiple, interacting levels of cognition.

The second main point on which I wish to reflect further is the relationship between (individual) tool use and (social) cooperation as potential vehicles of extended cognition. In his discussion of extended animal cognition, Cheng describes the cooperative efforts of weaver ants in constructing their nests from leaves, such as pulling them into shape, bending the foliage, and drawing their edges toward one another. The collective nest-building activities of weaver ants are another instance of DC, of course, but my focus here lies on Cheng’s treatment of the ants’ recruitment of silk-dispensing larvae (of their conspecifics) as a putative example of “extended cognition” (pp. 6–8). These larvae, which workers retrieve from nearby nests, secrete sticky substances that they normally use to spin their cocoons but are reused by the workers in this context to glue together the edges of leaves. As Cheng points out, the selection and handling of the larvae requires great care: First, the larvae to be chosen must be neither too young nor too old, their heads must be tapped in a special way to secrete the silk, and the workers need to maneuver diligently from edges to edges to make sure the silk-glue is applied in just the right way.

Appealing to Kaplan’s (2012) “mutual manipulability” criterion, Cheng proposes that we view the workers’ glueing behavior as a manifestation of extended cognition, thereby likening the workers’ use of other live organisms, the larvae, to an environmental resource or tool. To show that Kaplan’s criterion is satisfied, Cheng argues that a two-way communication (“signaling”) is taking place between the worker and the larva. In one direction, a worker’s cognitive state affects her handling of the glue stick (e.g., placement and timing); in the other direction, the silk-secreting (or lack thereof) behavior of the larva is taken as a signal modulating the worker’s tapping. This raises the question, On what grounds exactly do we consider this mutual arrangement as an instance of environmentally (or technologically) extended worker cognition, rather than as socially distributed worker–larva cognition?

In the case at hand, this may be a moot issue because the contribution of the larva is behaviorally (let alone cognitively) minimal to an extent that it essentially functions as nothing but a glue stick—a mere tool. But if we take human intuitions as our guide, shared collaborative interactions among people are categorically distinct from a single person’s exploitative incorporation of an external artifact (cf. Huebner, 2016; for an interesting discussion of borderline cases, see Blomberg, 2011). Must we leave such anthropomorphic sentiments behind in the analysis of situated animal cognition? If not, then what might distinguish genuine social cooperation (or collaboration) from what should be more aptly described (as Cheng does) as socially extended tool use? Kaplan’s mutual manipulability criterion, I surmise, is not fine-grained enough to differentiate between the causal couplings that underlie exploitative tool use from social collaboration. To show this, let us contemplate another evocative scenario in some detail.

My example draws on Turner’s (2002) groundbreaking analysis of the collective intelligence of mound-building termites (of the Macrotermes genus). Scattered across the savannas of Southern Africa, termite mounds function as impressively engineered respiratory devices (“external lungs”) for the colony, built to capture wind energy that ventilates the subterranean nest and to facilitate gas exchange. The structure and function of the mound is exquisitely adapted to serve the ventilation needs of a termite colony, which contains not only millions of workers but a considerably larger biomass of fungi, which they cultivate. Collectively, it is estimated that these organisms consume oxygen at the rate of a goat or cow. Thus, the maintenance of a viable nest atmosphere in which ventilation rate matches the respiratory demands of the colony (which can vary considerably) presents a formidable challenge. As Turner described in great detail, the termites succeed in this task by turning the mound into a “smart” organ of homeostasis.

Using Kaplan’s criterion, we can show that the dynamic architecture of the mound—an abiotic structure—is part of the extended physiology of the colony (see Turner, 2002, Chapter 11). As an ongoing source of disturbance, soil is continually eroded from the mound, which termites replace by transporting soil from inside to the corresponding surface location. For the mound to serve as an organ of homeostasis, the respective rates and patterns of external erosion from, and internal deposition to, the mound must be finely tuned, and dynamically coupled to the desired composition of the nest atmosphere. That is, if the mound is hyperventilating, this will trigger patterns of soil movement that reduce the mound’s capture of wind energy; conversely, a detected lack of oxygen will trigger patterns of soil transport that enhance the air flow into the mound. Hence, the termites and the mound stand in a relationship of “mutual manipulability.”

As a potential contrast, consider the relationship between the termites and fungi (Termitomyces), which termites cultivate in large structures (“fungus combs”) that take up most of their nest interior. This interspecies relationship is commonly described as a “mutualistic cooperation” or “symbiosis.” But by the preceding token, one could argue that the fungi are really extracorporeal parts of the termites’ digestive systems. They are digestive aids, “mere (biotic) tools” recruited by insects to convert otherwise indigestible cellulose into more nutritious compost that the termites can consume. Invoking Kaplan’s criterion, once again, reveals that the behavioral patterns of the two species are dynamically coupled, mediated by the structure of the combs. The combs as such are composed of macerated woody materials, collected and chewed up by foragers. When they return to the nest, after a quick pass through their guts, the foragers excrete the material, which immediately gets picked up by nest workers who add it on the top of the combs. Somewhere along the way, the woody materials are inoculated with fungal spores, which—once deposited in the comb—grow and begin to spread. The fungi, in turn, play an indispensable metabolic role: They break down cellulose into simpler sugars and nitrogen on which the termites feed, from the bottom of the comb. Again, the two parties stand in a relationship of mutual manipulability, but we should take heed that their causal entanglement is symmetrical. That is, might we not equally conclude—as Turner is fond of quipping—that the Macrotermes mound should perhaps be viewed as a fungus-built structure, where fungi cultivate termite populations to act as homeostatic regulators?

My bigger question, expressed in Cheng’s terminology, is thus whether there is a principled distinction between socially distributed and socially (or technologically?) extended animal cognition that does not simply turn—in a potentially question-begging manner—on intuitive considerations as to what constitutes a living organism and what doesn’t. As I have argued, Kaplan’s purely causal criterion seems ill-equipped to discriminate between these two types of situated cognition. Are there any distinctively cognitive or social task requirements that might underpin such a distinction, similar to how shared intentionality differs from cognitively less demanding modes of cooperation? Or must we conclude that our intuition to delineate sharply between DC and extended cognition is an anthropomorphic (or perhaps animal-centric!) reflex, charged with moral overtones that, for the most part, have no purchase in comparisons with other phyla?

References

1. Berdahl, A., Torney, C. J., Ioannou, C. C., Faria, J. J., & Couzin, I. D. (2013). Emergent sensing of complex environments by mobile animal groups. Science, 339(6119), 574–576. doi:10.1126/science.1225883

2. Blomberg, O. (2011). Socially extended intentions-in-action. Review of Philosophy and Psychology, 2, 335–353. doi:10.1007/s13164-011-0054-3

3. Byrne, R. W., & Whiten, A. (Eds.). (1988). Machiavellian Intelligence: Social expertise and the evolution of intellect in monkeys, apes, and humans. Oxford, England: Clarendon Press.

4. Couzin, I. D. (2009). Collective cognition in animal groups. Trends in Cognitive Sciences, 13, 36–43. doi:10.1016/j.tics.2008.10.002

5. Davies, J., & Michaelian, K. (2016). Identifying and individuating cognitive systems: A task-based distributed cognition alternative to agent-based extended cognition. Cognitive Processing, 17, 307–319. doi:10.1007/s10339-016-0759-4

6. Dunbar, R. I. M. (1998). The social brain hypothesis. Evolutionary Anthropology, 6, 178–190. doi:10.1002/(SICI)1520-6505(1998)6:5<178::AID-EVAN5>3.0.CO;2-8

7. Goldstone, R. L., & Gureckis, T. M. (2009). Collective behavior. Topics in Cognitive Science, 1, 412–438. doi:10.1111/j.1756-8765.2009.01038.x

8. Goldstone, R., & Theiner, G. (2017). The multiple, interacting levels of cognitive systems (MILCS) perspective on group cognition. Philosophical Psychology, 30, 334–368. doi:10.1080/09515089.2017.1295635

9. Grofman, B., Owen, G., & Feld, S. L. (1983). Thirteen theorems in search of the truth. Theory and Decision, 15, 261–278. doi:10.1007/BF00125672

10. Hofstadter, D. R. (1979). Gödel, Escher, Bach: An eternal golden braid. New York, NY: Basic Books.

11. Huebner, B. (2013). Macrocognition. New York, NY: Oxford University Press. doi:10.1093/acprof:oso/9780199926275.001.0001

12. Huebner, B. (2016). Transactive memory reconstructed: Rethinking Wegner’s research program. The Southern Journal of Philosophy, 54, 48–69. doi:10.1111/sjp.12160

13. Humphrey, N. K. (1976). The social function of intellect. In B. B. G. Bateson & R. A. Hinde (Eds.), Growing points in ethology (pp. 303–317). Cambridge, UK: Cambridge University Press.

14. Hutchins, E. (1995). Cognition in the wild. Cambridge, MA: MIT Press.

15. Jankovic, M., & Ludwig, K. (Eds.). (2017). The Routledge handbook of collective intentionality. New York, NY: Routledge.

16. Kaplan, D. M. (2012). How to demarcate the boundaries of cognition. Biology & Philosophy, 27, 545–570. doi:10.1007/s10539-012-9308-4

17. Koops, K., McGrew, W. C., & Matsuzawa, T. (2010). Do chimpanzees (Pan troglodytes) use cleavers and anvils to fracture Treculia africana fruits? Preliminary data on a new form of percussive technology. Primates, 51, 175-178. doi:10.1007/s10329-009-0178-6

18. List, C., & Vermeule, A. (2014). Independence and interdependence: Lessons from the hive. Rationality and Society, 26, 170–207. doi:10.1177/1043463114523713

19. Marshall, J. A. R., & Franks, N. R. (2009). Colony-level cognition. Current Biology, 19, R395–R396. doi:10.1016/j.cub.2009.03.011

20. Moll, H., & Tomasello, M. (2007). Cooperation and human cognition: The Vygotskian intelligence hypothesis. Philosophical Transactions of the Royal Society B: Biological Sciences, 362, 639–648. doi:10.1098/rstb.2006.2000

21. O’Donnell, S., Bulova, S. J., DeLeon, S., Khodak, P., Miller, S., & Sulger, E. (2015). Distributed cognition and social brains: Reductions in mushroom body investment accompanied the origins of sociality in wasps (Hymenoptera: Vespidae). Proceedings. Biological Sciences, 282(1810). doi:10.1098/rspb.2015.0791

22. Seeley, T. D. (1995). The wisdom of the hive: The social physiology of honey bee colonies. Cambridge, MA: Harvard University Press.

23. Seeley, T. D. (2010). Honeybee democracy. Princeton, NJ: Princeton University Press.

24. Simons, A. M. (2004). Many wrongs: The advantage of group navigation. Trends in Ecology & Evolution, 19, 453–455. doi:10.1016/j.tree.2004.07.001

25. Skyrms, B. (2004). The stag hunt and the evolution of social structure. New York, NY: Cambridge University Press.

26. Theiner, G. (2017). Groups as distributed cognitive systems. In M. Jankovic & K. Ludwig (Eds.), The Routledge handbook of collective intentionality (pp. 233–248). New York, NY: Routledge.

27. Theiner, G., Allen, C., & Goldstone, R. (2010). Recognizing group cognition. Cognitive Systems Research, 11, 378–395. doi:10.1016/j.cogsys.2010.07.002

28. Tomasello, M. (2014). A natural history of human thinking. Cambridge, MA: Harvard University Press. doi:10.4159/9780674726369

29. Turner, J. S. (2002). The extended organism: The physiology of animal-built structures. Cambridge, MA: Harvard University Press.

Situated Cognition and the Function of Behavior

David J. Pritchard

School of Biology
University of St Andrews

Reading Options

---

Abstract

In his review of “situated cognition” Cheng reminds us that the properties of cognition can be influenced by much more than what is going on in the brain. In this commentary, I focus on the lessons that this situated approach can teach those of us using behavior as a tool for investigating animal cognition. Rather than just a measure telling us about hidden cognitive processes, the details of behavior can provide important clues about how animals are solving a task. By looking in more detail at the behavior of our animals, and the possible sensory consequences of these behaviors, we can not only learn more about how animals do what they need to do but also explore how situated cognition shapes the structure of behavior.

Keywords: active vision, active sensing, ethology, computational biology, spatial cognition.

Author Note: David J. Pritchard, Centre for Biological Diversity, University of St Andrews, Greenside Place, St. Andrews, Fife, UK, KY16 9TH.

Correspondence concerning this article should be addressed to David J. Pritchard at djp4@st-andrews.ac.uk.

Acknowledgments: I would like to thank Shoko Sugasawa for her useful comments on a previous version of this manuscript, and Marcia Spetch and Lauren Guillette for giving the opportunity to write this commentary.

---

Introduction: What’s the Point of Situated Cognition?

As someone who has debated the value of embodied and extended cognition with my friends and colleagues, I found Cheng’s review of “situated cognition” incredibly helpful. For most biologists and psychologists interested in understanding natural behavior, the usefulness of situated cognition depends on what it adds that current perspectives don’t. Does situated cognition get us closer to understanding how animals do what they need to do in nature?

Based on Cheng’s overview in his target article, I would say the answer to that question is yes. Situated cognition not only explicitly links animal cognition with ecology and morphology but also raises intriguing evolutionary questions. For example, are there limits to where and when distributed cognition can evolve, similar to those discussed in studies of altruism and eusociality (Gardner & Grafen, 2009)? In this commentary, however, I focus on what I think situated cognition has to offer those of us who study animal cognition through behavioral experiments. In my opinion, situated cognition does more than just challenge our ideas about what counts as “cognitive”: It tells us that we could benefit from looking much more closely at behavior.

Don’t We Already Look at Behavior?

Although traditional approaches to animal cognition have always measured behavior, I would argue that these measurements have been shaped by the assumed relationship between cognition and behavior. Behavior is considered either an opportunity to learn, for example, in exploration or information seeking, or an output of cognition. As a result, most experiments on animal cognition focus on the end points of behavior, such as where a bird digs (Kelly, Kamil, & Cheng, 2010) or which arm a rat runs down (M. F. Brown, 1992). These behaviors are thought to reflect “decisions,” and our experiments test the effect of manipulations on these decisions. In my own research, for example, I have studied where a hummingbird probes for sucrose (Pritchard, Scott, Healy, & Hurly, 2016) and where a hummingbird stops when searching for a removed flower (Pritchard, Hurly, & Healy, 2015). In both cases, I have assumed that a decision to probe or hover is the result of the spatial cognition of the hummingbirds, and by measuring where these events take place, I can understand how that cognition operates.

Situated cognition challenges this assumption and suggests that by focusing on the end points of behavior, we might be missing out on clues to how animals are actually solving these tasks. When choosing between two stimuli, for example, chickens follow repeated paths to the chosen object. Along this path, chickens make an idiosyncratic sequence of head movements, viewing the object with different parts of the eye at different points along their path. When the chickens are forced to take a different path, their ability to discriminate between the objects decreases (Dawkins & Woodington, 2000). By focusing only on the end points of the discrimination task, whether a chicken chooses the “correct” object or not, we would be missing an important part of how these birds were solving this task.

Detailed examinations of behavior are much more common in studies of navigating Hymenoptera. Since the 1970s, studies of navigating bees, wasps, and ants have analyzed not only where an insect goes but also how she moves during navigation (e.g., T. S. Collett & Land, 1975). As a result, we now know that insects use specialized behaviors, such as learning flights and scanning head movements, to acquire and use visual information to navigate (M. Collett, Chittka, & Collett, 2013; Zeil, 2012). There are signs that a similar, descriptive approach to spatial cognition could be used to study “enactive” cognition in vertebrates. Birds, for example, also show prominent head movements: Pigeons bob their heads (Green, Davies, & Thorpe, 1994; Troje & Frost, 2000), owls peer side to side (Ohayon, Van Der Willigen, Wagner, Katsman, & Rivlin, 2006; Van Der Willigen, Frost, & Wagner, 2002), and terns scan scenes with different parts of their retina (Land, 1999). You only need to watch birds at a feeder in your garden to see that birds are constantly using behavior to shape what they see. Although insect navigation strategies have been presented as efficient solutions for animals with poor resolution vision and small brains (e.g., Chittka & Skorupski, 2017), I don’t see any reason why similar “enactive” strategies might not be used by birds, despite their larger brains and higher resolution vision. Currently, however, we haven’t looked to see if this is the case. Studies of spatial cognition in birds have focused mostly on the end points of behavior, such as the location of digging or pecking, and much less on how birds reach these locations in the first place. If enactive and extended cognition encourages more people to look more closely at the behavior of their animals, then these approaches are already adding something important to our traditional approaches.

Situated Cognition and Active Sensing

The “active” visual strategies of navigating insects and discriminating chickens highlight the role that “active sensing” could play in situated cognition. Active sensing involves animals using energy to sense their environment, either in terms of behavior (in the case of active vision or whisker movements) or by producing signals (e.g., echolocation or electroreception; Nelson & MacIver, 2006). By definition, active sensing informs an animal’s cognitive state (it is sensing, after all), but many examples of active sensing also show signs of being influenced by an animal’s cognition. Rats will modify their whisker movements in anticipation of objects (Grant, Mitchinson, Fox, & Prescott, 2009), bats adjust the direction and structure of their echolocation calls based on their experience (Moss & Surlykke, 2010), and electric fish actively scan areas where they have previously found food (Jun, Longtin, & Maler, 2016). All of these examples potentially pass the mutual manipulability criterion (MMC) as discussed by Cheng. As in the spiders of Japyassú and Laland (2017), these animals are using behavior not only to sense their environment but also to direct and focus their attention.

The line between active sensing and situated cognition can, however, be quite slippery. Barn owls, for example, use “peering” head movements to assess distances before attacking prey (Ohayon et al., 2006). These movements generate motion parallax and so provide owls with information about depth and distances, a form of active sensing. I suspect peering would also pass the MMC, although I don’t know if this has been tested directly. Barn owls can, however, also perceive depth using stereo vision and can use stereo vision to recognize distances they previously learned via head movements (Van Der Willigen et al., 2002). This would suggest that, once acquired, depth information is represented independently of how it was perceived, which would seem to support the traditional Cartesian view. Even if peering was found to pass the MMC, it might therefore be difficult to classify peering as truly part of an enactive cognitive system. But does this matter? Under natural conditions, owls automatically make peering movements when inspecting visual scenes and, indeed, had to be actively trained not to make peering movements in Van Der Willigen et al.’s (2002) experiments. Peering seems to have evolved as part of the package of mechanisms that owls use to inspect the world. Rather than drawing a hard line separating behaviors such as peering from “true” examples of situated cognition, it might be more productive to embrace the fact that cognition (like all biology) is a bit messy around the edges. Indeed, it might be in the gray areas between cognition, behavior, perception, and morphology in which we discover the most interesting comparative data.

Cognition and the Organization of Behavior

If an animal’s behavior is involved in processing information, then what consequences does this have for the evolution of behavior? Many unusual behaviors in animals function as a way to simplify sensory processing. Head bobbing in birds, for example, reduces visual blur by restricting head movements to rapid thrusts forward (Necker, 2007), whereas flies restrict head rotations to short saccades to better separate translational optic flow (which contains useful depth information) from rotational optic flow (which doesn’t; Hateren & Schilstra, 1999). If the need to process other information influences how an animal behaves, then we might expect to see a similar adaptation in the structure of behaviors used in enactive or extended cognition.

How would we identify the influence of cognition on the structure of behavior? In the case of vision, the sensory “consequences” of behavior can be worked out based on the function of the eye (Zeil, Boeddeker, & Hemmi, 2008). For cognition, however, the consequences of behavior are likely to depend on the task that the animal faces. The “best” behavior for learning about space, for example, might look very different from the “best” behavior for learning about material properties, or for inspecting a conspecific. Instead, perhaps we could start by looking for any patterns in behavior at all. The computational analysis of behavior has recently become a hot topic in neuroscience and has resulted in a suite of computational methods to identify patterns in behavior (Anderson & Perona, 2014; A. E. Brown & de Bivort, 2017; Egnor & Branson, 2016). Although designed for computational neuroscience, these methods could provide a valuable new tool for comparative cognition researchers looking to broaden their measures of behavior. By identifying patterns in how animals behave during, for example, a spatial memory task, these computational methods could highlight candidate behaviors that could then be examined in more detail. In this manner, descriptive analyses of the structure of behavior could complement traditional experiments in comparative cognition and lead to a more integrative study of animal cognition.

Conclusion: Description and Diversity in Animal Cognition

One of my favorite aspects of the animal kingdom is its diversity, what Darwin (1859) referred to as its “endless forms most beautiful” (p. 490). By emphasizing the role that behavior, environment, and morphology can play in cognitive processing, Cheng presents a version of animal cognition that embraces this diversity. Situated cognition does not just evolve through changes in the brain or via tweaks in the accuracy, capacity, or duration of general processes. Situated cognition could evolve and adapt through changes in bodies, or in behavior, or even in social structure. At the beginning of one of the key early texts on embodied cognition, James J. Gibson (1979) wrote, “We are told that vision depends on the eye, which is connected to the brain. I shall suggest that natural vision depends on the eyes in the head on a body supported by the ground” (p. 1). Given the diversity in senses, anatomy, ecology, and behavior seen in the animal kingdom, we might therefore expect that diversity in cognition might be the norm.

A necessary step for those of us wishing to investigate this situated cognition will be to look much closer at how animals are behaving, and how this changes during our experiments. This is not a new suggestion. One of the first sections of Tinbergen’s (1963) seminal “On Aims and Methods in Ethology” was on the need for “Observation and Description.” But although Tinbergen’s four questions have been embraced by ecologically minded researchers in comparative cognition (e.g., B. Gibson & Kamil, 2009; Kamil, 1998), his warnings about the need for description and observation seem to have had less impact. The triumph of Tinbergen and the early ethologists was to expand animal behavior beyond the handful of model species studied in psychology laboratories in order to include the diverse range of species and behaviors seen in the wild. Although comparative psychology today is very different to that faced by the early ethologists, Tinbergen’s (1963) warning about the danger of recording only behavior we consider relevant and overlooking “trivial” behavior still rings true: “We might forget that naïve, unsophisticated, or intuitively guided observation may open our eyes to new problems. Contempt for simple observation is a lethal trait in any science” (p. 412).

We are now living in a time in which computational approaches are revolutionizing how we can study behavior, providing tools that can capture, quantify, and analyze behavior like never before. Observation and description can now be carried out automatically and in incredible detail. But technology itself is not necessary for looking closer at behavior. Studies of navigating insects have been using film and video to measure the details of behavior since the 1970s (T. S. Collett & Land, 1975). What is needed is a reason to look closer, and I think that situated cognition could provide one.

References

1. Anderson, D. J., & Perona, P. (2014). Toward a science of computational ethology. Neuron, 84, 18–31. doi:10.1016/j.neuron.2014.09.005

2. Brown, A. E., & de Bivort, B. (2017). The study of animal behaviour as a physical science. BioRxiv, 220855. doi:10.1101/220855

3. Brown, M. F. (1992). Does a cognitive map guide choices in the radial-arm maze? Journal of Experimental Psychology. Animal Behavior Processes, 18, 56–66. doi:10.1037/0097-7403.18.1.56

4. Chittka, L., & Skorupski, P. (2017). Active vision: A broader comparative perspective is needed. Constructivist Foundations, 13, 128–129.

5. Collett, M., Chittka, L., & Collett, T. S. (2013). Spatial memory in insect navigation. Current Biology, 23(17), R789–R800. doi:10.1016/j.cub.2013.07.020

6. Collett, T. S., & Land, M. F. (1975). Visual spatial memory in a hoverfly. Journal of Comparative Physiology A, 100, 59–84. doi:10.1007/BF00623930

7. Darwin, C. (1859). On the origin of species by means of natural selection, or, the preservation of favoured races in the struggle for life. London, England: J. Murray.

8. Dawkins, M. S., & Woodington, A. (2000). Pattern recognition and active vision in chickens. Nature, 403(6770), 652–655. doi:10.1038/35001064

9. Egnor, S. E. R., & Branson, K. (2016). Computational analysis of behavior. Annual Review of Neuroscience, 39, 217–236. doi:10.1146/annurev-neuro-070815-013845

10. Gardner, A., & Grafen, A. (2009). Capturing the superorganism: a formal theory of group adaptation. Journal of Evolutionary Biology, 22, 659–671. doi:10.1111/j.1420-9101.2008.01681.x

11. Gibson, B., & Kamil, A. (2009). The synthetic approach to the study of spatial memory: Have we properly addressed Tinbergen’s ‘four questions’? Behavioural Processes, 80, 278–287. doi:10.1016/j.beproc.2008.11.013

12. Gibson, J. J. (1979). The ecological approach to visual perception. Boston, MA: Houghton Mifflin.

13. Grant, R. A., Mitchinson, B., Fox, C. W., & Prescott, T. J. (2009). Active touch sensing in the rat: Anticipatory and regulatory control of whisker movements during surface exploration. Journal of Neurophysiology, 101, 862–874. doi:10.1152/jn.90783.2008

14. Green, P. R., Davies, M. N. O., & Thorpe, P. H. (1994). Head-bobbing and head orientation during landing flights of pigeons. Journal of Comparative Physiology A, 174, 249–256. doi:10.1007/BF00193791

15. Hateren, J. H., & Schilstra, C. (1999). Blowfly flight and optic flow. II. Head movements during flight. Journal of Experimental Biology, 202, 1491–1500.

16. Japyassú, H. F., & Laland, K. N. (2017). Extended spider cognition. Animal Cognition, 20, 375–395. doi:10.1007/s10071-017-1069-7

17. Jun, J. J., Longtin, A., & Maler, L. (2016). Active sensing associated with spatial learning reveals memory-based attention in an electric fish. Journal of Neurophysiology, 115, 2577–2592. doi:10.1152/jn.00979.2015

18. Kamil, A. C. (1998). On the proper definition of cognitive ethology. In R. Balda, I. Pepperberg, & A. C. Kamil (Eds.), Animal cognition in nature. The convergence of psychology and biology in laboratory and field (pp. 1–28). New York, NY: Academic Press.

19. Kelly, D. M., Kamil, A. C., & Cheng, K. (2010). Landmark use by Clark’s nutcrackers (Nucifraga columbiana): Influence of disorientation and cue rotation on distance and direction estimates. Animal Cognition, 13, 175–188. doi:10.1007/s10071-009-0256-6

20. Land, M. F. (1999). The roles of head movements in the search and capture strategy of a tern (Aves, Laridae). Journal of Comparative Physiology A, 184, 265–272.

21. Moss, C. F., & Surlykke, A. (2010). Probing the natural scene by echolocation in bats. Frontiers in Behavioral Neuroscience, 4, 33. doi:10.3389/fnbeh.2010.00033

22. Necker, R. (2007). Head-bobbing of walking birds. Journal of Comparative Physiology A, 193, 1177–1183. doi:10.1007/s00359-007-0281-3

23. Nelson, M. E., & MacIver, M. A. (2006). Sensory acquisition in active sensing systems. Journal of Comparative Physiology A, 192, 573–586. doi:10.1007/s00359-006-0099-4

24. Ohayon, S., Van Der Willigen, R. F., Wagner, H., Katsman, I., & Rivlin, E. (2006). On the barn owl’s visual pre-attack behavior: I. Structure of head movements and motion patterns. Journal of Comparative Physiology A, 192, 927–940. doi:10.1007/s00359-006-0130-9

25. Pritchard, D. J., Hurly, T. A., & Healy, S. D. (2015). Effects of landmark distance and stability on accuracy of reward relocation. Animal Cognition, 18, 1285–1297. doi:10.1007/s10071-015-0896-7

26. Pritchard, D. J., Scott, R. D., Healy, S. D., & Hurly, A. T. (2016). Wild rufous hummingbirds use local landmarks to return to rewarded locations. Behavioural Processes, 122, 59–66. doi:10.1016/j.beproc.2015.11.004

27. Tinbergen, N. (1963). On aims and methods of ethology. Ethology, 20, 410–433. doi:10.1111/j.1439-0310.1963.tb01161.x

28. Troje, N. F., & Frost, B. J. (2000). Head-bobbing in pigeons: How stable is the hold phase? The Journal of Experimental Biology, 203(Pt. 5), 935–940.

29. Van Der Willigen, R. F., Frost, B. J., & Wagner, H. (2002). Depth generalization from stereo to motion parallax in the owl. Journal of Comparative Physiology A, 187, 997–1007. doi:10.1007/s00359-001-0271-9

30. Zeil, J. (2012). Visual homing: An insect perspective. Current Opinion in Neurobiology, 22, 285–293. doi:10.1016/j.conb.2011.12.008

31. Zeil, J., Boeddeker, N., & Hemmi, J. M. (2008). Vision and the organization of behaviour. Current Biology, 18(8), 320–323. doi:10.1016/j.cub.2008.02.017

Examining the “Species” of Situated Cognition in Humans

Ewen B. Lavoie and Jennifer K. Bertrand

Faculty of Kinesiology, Sport, and Recreation
University of Alberta

Jeffrey Sawalha

Faculty of Kinesiology, Sport, and Recreation
University of Alberta

Scott A. Stone and Nathan J. Wispinski

Department of Psychology
University of Alberta

Craig S. Chapman

Faculty of Kinesiology, Sport, and Recreation and Neuroscience and Mental Health Institute
University of Alberta

Reading Options

---

Abstract

In the target article “Cognition Beyond Representation: Varieties of Situated Cognition in Animals,” Ken Cheng describes situated cognition as a “genus” of ideas and effects whereby cognition extends beyond the central nervous system of an organism to include its peripheral nervous system and/or the environment. Although Cheng’s article focuses specifically on nonhuman animals, here we apply his definitions of four “species” of situated cognition to find examples in humans. We highlight the ways in which each of distributed (e.g., a crew flying an airplane), embodied (e.g., computation in peripheral sense organs), extended (e.g., extensions of peripersonal space), and enactive (e.g., decision making reflected in movement) cognition are seen in humans. In doing so, we provide evidence for Cheng’s major hypothesis that cognition is not confined solely to the central nervous system and that this may be a fundamental principle of cognition across animal organisms.

Keywords: situated cognition, human behavior, cognition

Author Note: Ewen B. Lavoie, 4-222 Van Vliet Complex East, University of Alberta, Edmonton, Alberta, T6G 2H9.

Correspondence concerning this article should be addressed to Ewen B. Lavoie at elavoie@ualberta.ca.

Acknowledgments: All authors contributed equally.

---

Introduction

In the target article, “Cognition Beyond Representation: Varieties of Situated Cognition in Animals,” Ken Cheng defines a “genus” of notions that he claims have appeared in human cognitive neuroscience and philosophy whereby cognition extends beyond the central nervous system of the agent to include the peripheral nervous system and the environment. Cheng highlights four specific “species” of so-called situated cognition: distributed, enactive, embodied, and extended. Although Cheng’s clear aim is to discuss evidence for these “species” in nonhuman animals, situated cognition in humans is still a controversial proposal. Thus, to both support Cheng’s overall species classifications and provide some context in which to interpret the nonhuman animal work, here we outline examples of situated cognition in humans.

Distributed Cognition

According to Cheng, distributed cognition refers to the reduction of individual cognitive capacities among many to complete tasks that are otherwise too taxing to be completed alone. He presents ants cooperating to function seemingly as a single organism (i.e., as a “hive mind”) as a canonical example from the animal kingdom. Here, we point to research of aircraft cockpit crews as exhibiting the same core features (Hutchins, 1995; Hutchins & Klausen, 1996). That is, these crews are able to safely take off, fly, and land a plane, even though no one crew member is responsible for “flying the plane.” Like ants weaving a nest, information is disseminated among the crew in an organized fashion, where each individual’s contribution is seemingly small but decidedly important for task completion. For example, the captain is responsible for tasks such as contacting the airline traffic controller and relaying that information to the first officer, who must perform the translation of information into physical motor actions such as heading modification or thrust adjustments (Hutchins & Klausen, 1996).

Cheng’s idea of the “entire hive as a cognizing unit” comes with hypotheses about the purposes served when cognition is distributed. A distributed cognition hypothesis suggests that “hive-minded” animals may reduce metabolic costs and operate with smaller nervous systems. Although perhaps more difficult to imagine humans as “hive minded,” especially in ways that could affect brain anatomy, humans do employ this mind-set to overcome our own human-scaled hive challenges. It appears that distributed cognition arises in humans when the cognizing power of a single person seems insufficient for a critical and complex situation, like the aircraft cockpit (Hutchins, 1995), the cardiac surgery theater (Hazlehurst, McMullen, & Gorman, 2007), or the emergency dispatch coordination center (Artman & Wærn, 1999). We view the distribution of cognition in an aircraft cockpit as nothing less than a scaled-up version of the honeybee hive, where neither would find success without the entire flight crew or hive operating as a whole cognizing unit.

Embodied Cognition

In his review, Cheng defines the historically broad term of embodied cognition as computation offloaded to the periphery rather than “central representational cognition” (p. 1). Here, one of Cheng’s key examples is the intelligent behavior of an octopus tentacle performing complex bending computations. Of interest, a relevant human example also occurs in our distal effectors. That is, human fingertips have recently been shown to perform their own complex computation, in complete isolation from central nervous control. Here, peripheral neurons within the human fingertips have been shown to signal the edge orientation of touched objects (Pruszynski & Johansson, 2014). Edge detection is a hallmark of complex feature extraction and is the exact kind of computational problem that is efficient for an organism to offload to peripheral mechanisms.

In a similar vein, Cheng’s version of embodied cognition has also been shown to occur in peripheral neurons in the visual system. It is well documented that the human retina takes in a vast amount of visual information, but the pathway from the eye to the brain presents a significant bottleneck. This is precisely the case where peripheral computation is ideal, and indeed, cells within the human retina perform computations to both systematically compress the information transmitted to the cortex and extract primitive visual features. For instance, retinal cells have been shown to respond selectively to object motion distinct from background motion in a visual scene (Gollisch & Meister, 2010). These studies show that humans, like the octopus, distribute cognition to peripheral systems, thereby reducing the computational demands on the central nervous system.

Extended Cognition

Cheng defines extended cognition as “cognition encompassing physical objects in the world, often objects constructed by the animal” (p. 2) and presents a spider’s construction and use of a web as a prototypical example. Here, the examples in the human literature are more familiar, as tool use is often held up as a hallmark of human ingenuity; one needs only reach into their pocket for a smartphone to confront exactly how much intelligence is now offloaded to external devices. However, our example here is a more foundational way in which humans extend their cognition to include tools. It has been shown that as humans use a particular physical object (e.g., a rake), the neural representation of their body schema is reorganized. For example, in right-brain-damaged patients a condition known as visuotactile extinction can arise, in which patients are unable to report a visual and/or tactile stimulus on their left hand (contralesional) when presented with a visual stimulus near their right hand (ipsilesional). Typically, the visual extinction is most severe for stimuli presented close to the body. Farnè, Bonifazi, and Làdavas (2005) showed the malleability of the body schema by giving extinction patients experience using a long rake. After tool exposure, the visual extinction effect was physically drawn out in space—from near the hand to the end of the rake. Interesting to note, this effect was not present when patients simply held the tool, confirming previous findings in monkeys (Iriki, Tanaka, & Iwamura, 1996) that humans can elongate their body schema to include tools, but only when the tool is being used to interact with the surrounding environment.

Enactive Cognition

In his review, Cheng defines enactive cognition broadly as intelligence arising out of action. He presents play behavior in dogs and, trending into human territory, human dance as examples of enactive cognition. Here we provide a brief summary of three other relevant domains of human cognitive science that fit within the context of enactive cognition.

First, a strong body of research has shown that humans are afforded increased sensitivity for cognitive processing because of specific actions or spatial orientations. For instance, the position of one’s hand in space alters vision (Abrams, Davoli, Du, Knapp, & Paull, 2008), focuses the distribution of attentional resources toward stimuli close to the hand (Reed, Grubb, & Steele, 2006), and improves detection accuracy in the blind field of a patient with unilateral damage to primary visual cortex (Schendel & Robertson, 2006). In this way, intelligence is arising out of action because performance is causally influenced by the body’s position in space.

Second, a growing collection of studies argue the point that “moving is thinking” (e.g., Song & Nakayama, 2009). It is thought that movement reflects a continuously evolving cognitive state that is also influenced by the moving body. For instance, when asked to reach toward one of two targets, the trajectory of the hand in space is thought to reflect the ongoing competition between potential movements that is resolved in time (e.g., Chapman et al., 2010; for a review, see Gallivan & Chapman, 2014). Further, while this movement is taking place, information about the position of one’s own body and continual information about the environment is incorporated into the ongoing competition between options (Todorov & Jordan, 2002). In this way, everyday human movements are like Cheng’s examples of a dog playing or a dancer; cognition influences movements while movements influence cognition, which give rise to dynamic and continuously evolving thought.

Like dogs, then, this suggests that humans are continuously broadcasting a signal of their thinking via movement. This was recently tested in a study where participants observing someone move implicitly inferred their action intent (Pesquita, Chapman, & Enns, 2016). In this study, participants observing an actor reach for a target were faster to guess the end location of the observed movement when the actor was choosing where to reach as compared to being directed where to reach. These results confirm that signals of cognitive processes like decision making are evident in human movement and suggest that enactive cognition might lie at the heart of social cognition (and social constructs like language), which fundamentally requires the prediction of others’ attentional states.

Conclusion

Although not entirely novel, the idea that human cognition extends beyond the central nervous system is still a minority position. However, the target article by Cheng lends significant credence to these ideas by discussing situated cognition across animal species. Here, we support these ideas further by giving specific human examples demonstrating just how ubiquitous situated cognition is.

References

1. Abrams, R. A., Davoli, C. C., Du, F., Knapp, W. H., & Paull, D. (2008). Altered vision near the hands. Cognition, 107, 1035–1047. doi:10.1016/j.cognition.2007.09.006

2. Artman, H., & Wærn, Y. (1999). Distributed cognition in an emergency co-ordination center. Cognition, Technology & Work, 1, 237–246. doi:10.1007/s101110050020

3. Chapman, C. S., Gallivan, J. P., Wood, D. K., Milne, J. L., Culham, J. C., & Goodale, M. A. (2010). Reaching for the unknown: Multiple target encoding and real-time decision-making in a rapid reach task. Cognition, 116, 168–176. doi:10.1016/j.cognition.2010.04.008

4. Cheng, K. (2018). Cognition beyond representation: Varieties of situated cognition in animals. Comparative Cognition and Behavior Reviews, 13, 1–20. 1-20. doi:10.3819/CCBR.2018.130001

5. Farnè, A., Bonifazi, S., & Làdavas, E. (2005). The role played by tool-use and tool-length on the plastic elongation of peri-hand space: A single case study. Cognitive Neuropsychology, 22, 408–418. doi:10.1080/02643290442000112

6. Gallivan, J. P., & Chapman, C. S. (2014). Three-dimensional reach trajectories as a probe of real-time decision-making between multiple competing targets. Frontiers in Neuroscience, 8. doi:10.3389/fnins.2014.00215

7. Gollisch, T., & Meister, M. (2010). Eye smarter than scientists believed: Neural computations in circuits of the retina. Neuron, 65, 150–164. doi:10.1016/j.neuron.2009.12.009

8. Hazlehurst, B., McMullen, C. K., & Gorman, P. N. (2007). Distributed cognition in the heart room: How situation awareness arises from coordinated communications during cardiac surgery. Journal of Biomedical Informatics, 40, 539–551. doi:10.1016/j.jbi.2007.02.001

9. Hutchins, E. (1995). How a cockpit remembers its speeds. Cognitive Science, 19, 265–288. doi:10.1207/s15516709cog1903_1

10. Hutchins, E., & Klausen, T. (1996). Distributed cognition in an airline cockpit. In Y. Engeström & D. Middleton (Eds.), Cognition and communication at work (pp. 15–34). New York, NY: Cambridge University Press.

11. Iriki, A., Tanaka, M., & Iwamura, Y. (1996). Coding of modified body schema during tool use by macaque postcentral neurones. NeuroReport, 7(14), 2325–2330. doi:10.1097/00001756-199610020-00010

12. Pesquita, A., Chapman, C. S., & Enns, J. T. (2016). Humans are sensitive to attention control when predicting others’ actions. Proceedings of the National Academy of Sciences, 113(31), 8669–8674. doi:10.1073/pnas.1601872113

13. Pruszynski, J. A., & Johansson, R. S. (2014). Edge-orientation processing in first-order tactile neurons. Nature Neuroscience, 17, 1404–1409. doi:10.1038/nn.3804

14. Reed, C. L., Grubb, J. D., & Steele, C. (2006). Hands up: Attentional prioritization of space near the hand. Journal of Experimental Psychology: Human Perception and Performance, 32(1), 166–177. doi:10.1037/0096-1523.32.1.166

15. Schendel, K., & Robertson, L. C. (2006). Reaching out to see: Arm position can attenuate human visual loss. Journal of Cognitive Neuroscience, 16, 935–943. doi:10.1162/0898929041502698

16. Song, J. H., & Nakayama, K. (2009). Hidden cognitive states revealed in choice reaching tasks. Trends in Cognitive Sciences, 13(8), 360–366. doi:10.1016/j.tics.2009.04.009

17. Todorov, E., & Jordan, M. I. (2002). Optimal feedback control as a theory of motor coordination. Nature Neuroscience, 5, 1226–1235. doi:10.1038/nn963

Yesterday the Earwig, Today Man, Tomorrow the Earwig?

Christopher L. Hewitson, David M. Kaplan, and John Sutton

Department of Cognitive Science, ARC Centre of Excellence in Cognition and its Disorders, and Perception in Action Research Centre
Macquarie University

Reading Options

---

Abstract

In this commentary, we highlight some relevant history of the situated cognition movement and then identify several issues with which we think further progress can be made. In particular, we address and clarify the relationship between situated cognition and antirepresentational approaches. We then highlight the heterogeneous nature of the concept of morphological computation by describing a less common way the term is used in robotics. Finally, we discuss some residual concerns about the mutual manipulability criterion and propose a potential solution.

Keywords: robotics, morphological computation, antirepresentationalism, mutual manipulability, causal specificity

Author Note: David M. Kaplan, Department of Cognitive ­Science, 16 University Drive, Australian Hearing Hub, Level 3, Macquarie University, NSW 2109, Australia.

Correspondence concerning this article should be addressed to David M. Kaplan at david.kaplan@mq.edu.au.

---

A Brief History of Situated Cognition Research

We applaud Cheng’s effort to bring concepts from the framework of situated cognition to a wider biological audience, including researchers working on insect behavior. It is also delightfully ironic. Cheng notes that although the idea of “situated cognition has been bantered in philosopher and cognitive science for some time now … its connection with nonhuman animals has a more recent history” (p. 11). Although he is correct to point out that recent philosophical discussions have focused on the human case, much of the early history had a different focus, closer to home for the target audience of this article. For deep theoretical reasons, many of the earliest discussions of embodied cognition among researchers in artificial intelligence (AI) and robotics centered on the rich, adaptive behavioral repertoires of simpler organisms including insects. Therefore, in many ways, the discussion of situated cognition has truly come full circle—from insects to humans and now back to insects again.

Consider, for example, roboticist Rodney Brooks’s (1991) seminal article “Intelligence Without Representation,” which perhaps more than any other single article served to launch the embodied cognition movement. A central goal of this article and much of Brooks’s career was to build a research program in AI that drew inspiration from the intelligent behavior manifested by insects and other simpler biological organisms rather than humans. Brooks forcefully argued that researchers in AI had mistakenly assumed for too long that the hallmark of intelligence was disembodied reasoning or computation performed over explicit, language-like representations, instead of adaptive and flexible control of bodily action. According to Brooks, the latter is the real locus of natural (and so artificial) intelligence. This shift in perspective, which Brooks (1999) later termed behavior-based robotics, led him and others to extract design principles for robots from a close inspection of the behavioral competencies exhibited by nonhuman organisms like ants and bees. The Brooksian perspective should resonate with biologists who naturally (and correctly) see intelligence and cognition through an evolutionary lens.

Brooks was by no means alone in embracing this perspective in the early development of the embodied cognition movement. In a seminal book that formed the foundation for much current work on embodied and extended cognition, Andy Clark (1997) used case studies of organisms like the “humble” cockroach to argue for an embodied view of intelligence. According to Clark (1997), the essence of intelligence and cognition is rooted in an organism’s basic capacities to “sense, act, and survive” (p. 4), and so can be found throughout the biological world. Barbara Webb’s (1994) influential robotic modeling of the mechanisms of cricket phonotaxis similarly emphasized how bodily structures can in some instances obviate the need for internal representation and computation in the service of intelligent behavior. Cognitive scientists in the 1990s thus had to take seriously the idea that the adaptive behavior exhibited by relatively simple mobile robots and insects should form the basis for scaled-up approaches to flexible decision making and action control in humans. The title of David Kirsh’s (1991) now famous critical response to Brooks (1991) succinctly (and somewhat sarcastically) captures the view: “Today the Earwig, Tomorrow Man?”

It is both interesting and exciting that debates about situated cognition (which, according to Cheng, subsumes embodied and extended cognition) have now come full circle. Early discussions revolved around insect models; “intermediate” discussions around humans; and at least some future discussions, one hopes, will revolve around insects again. Situated cognition research is in an important sense returning to its roots. With a nod to Kirsh (1991), we may summarize the past, present, and future of situated cognition as follows: Yesterday the earwig, today man, tomorrow the earwig.

Situated Cognition and Representationalism: Enmity, Friendship, or Neutrality?

Theorists in some areas of situated cognition, as Cheng notes, have embraced the revolutionary rhetoric by which Brooks and others, for a time at least, hoped to convince cognitive scientists to scale up from insects to humans without appealing to computations over representations. From the start, however, many argued that creatures like us require more sophisticated control systems because we have rangier and more conflicting goals, and face more representation-hungry problems in more complex environments. For Kirsh (1991) and Clark (1997, 2005), revised and nonclassical notions of dynamic and action-oriented representation would remain in the situated cognition theorist’s toolkit. From one perspective, the subsequent history of situated cognition can look like an unproductive sequence of standoffs between various strands of radical antirepresentationalism (enactivism, dynamical systems theory, direct realism), and various defensive bands of moderates who remained friends of the representational theory of mind. But this story of entrenched enmity between situated cognition and representationalism is partial and misleading.

Cheng sometimes identifies situated cognition, and especially its more “liberal versions,” with the claim that cognition is “fundamentally different from the standard cognition of representations” (p. 2). On this view, it is a “conservative” move to argue that accounts of complex cognitive phenomena are unlikely to “escape references to representations” (p. 12). But these ways of drawing the lines unhelpfully collapse two distinct issues. Situated cognition, we suggest, is not fundamentally a thesis about representations at all. The right target, instead, has always been internalism or individualism, the view that cognitive processes occur entirely in the individual head, fundamentally distinct and separate from body and environment. Although issues about mental representation have often been confused with issues about individualism, they are orthogonal (Sutton, 2015; Wilson, 1994). Nothing about the truth or falsity of internalism is settled by adopting a particular view about mental representation.

By treating situated cognition as neutral with respect to representationalism, we open up appropriate space for empirical inquiry to address a range of possible relations between internal and external resources in different cognitive systems. Sometimes the relevant computations occur over external representations in public symbol systems. Intelligent behavior and cognition can reflect many different combinations of neural, bodily, environmental, and social resources. Empirical study is needed to identify the different ways in which such heterogeneous but complementary resources are integrated across distinctive cognitive ecologies (Hutchins, 2010; Menary, 2007; Sutton, 2015). To the extent that the situated cognition movement is radical, it is not because it dispenses in any blanket fashion with computations and representations but because it does not assume a solely internal location for cognitive states and processes in the heads of individual organisms.

This slight corrective to Cheng’s tendency to identify situated cognition with antirepresentationalism in fact highlights some of his most promising claims about the possibility of a “merger” between situated and representational explanations (p. 10). As he notes, it is far from obvious that nonrepresentational explanation will prove sufficient to explain even insect cognition in its more complex and demanding forms. But this does not mean that situated cognition fails in these cases. Body and environment may still be playing key roles in larger cognitive systems, even if the internal components of the processes in question are computational. Over the timescales of cultural evolution and development alike, it may be that those internal computational processes are deeply transformed and shaped by the situated nature of biological intelligence.

Morphological Computation: A Heterogeneous Concept

The concept of morphological computation, which Cheng introduces in his interesting discussion of octopus fetching movements, might be one basis of such transformation. Morphological computation is the notion that certain processes can be carried out by the body that would otherwise be handled through computations performed by the central nervous system (CNS; Pfeifer & Bongard, 2006; Pfeifer, Iida, & Lungarella, 2014; Pfeifer, Lungarella, & Iida, 2007). As Cheng notes, claims about morphological computation provide especially clear examples of embodied cognition hypotheses. But the term is ambiguous, and several distinct phenomena are collapsed under a single concept (Hoffmann and Müller, 2017). In this section, we try to pull them apart.

Morphological computation most commonly describes a reduction in the computational load placed on the CNS by exploiting material properties of the organism’s body such as its shape, structure, and dynamics. In this sense, bodily properties effectively change—and simplify—the computation to be performed. This is what Cheng means when he uses the term, and this is why he prefers to call it “decomputation.” Instead of trying to solve the computationally demanding inverse kinematics problem (mapping a desired outcome into motor commands; Flash & Sejnowski, 2001) for all the degrees of freedom of one of its hyperflexible arms, the octopus temporarily reconfigures its arm into a stiff quasi-jointed structure to transfer an object from one place to another (Sumbre, Fiorito, Flash, & Hochner, 2006). More specifically, precise patterns of muscle activations function to dynamically set joint locations and divide the arm into proximal, medial, and distal segments, which drastically reduces the degrees of freedom in the soft-body arm from near infinite to just 7. This vertebrate-like quasi-joint strategy in turn reduces the computational load on the associated neural circuitry and greatly simplifies the motor control problem to be solved. We therefore agree with Cheng that it is probably more appropriate to call this decomputation because bodily properties are harnessed to reduce the computational demands on the CNS rather than it being the case that computations are literally being performed in the non-neural body.

But Cheng’s brief discussion of morphological computation, especially his decision to rename it, hints that there might be another—more literal—way to use the term. We agree. Robotics researchers do in fact seem to use the term to describe a genuine division of computational labor across the CNS and body, where the overall amount of computation to be performed remains essentially unchanged, but parts of the body literally do some of the computational processing. In this case, neural computation would quite literally be offloaded onto the non-neural body. Although Cheng provides some hints that there must be something more to the idea of morphological computation beyond that of decomputation, he offers no examples of genuine morphological computation. We briefly describe one kind of example proposed in the robotics literature to illustrate this alternative conception and merely raise the possibility that analogous biological examples may be found.

In a series of network simulations, Hauser and colleagues (Hauser, Ijspeert, Füchslin, Pfeifer, & Maass, 2011, 2012) provided evidence that nonrigid or compliant physical bodies or body parts (modeled as recurrent networks of mass-spring systems) can be used to implement certain nonlinear computational operations. More specifically, they argued that the nonlinear input–output transformation achieved by these networks is implemented by the interconnected mass-springs that function as the network weights. Critically, it is the morphological structure and dynamic properties of these mass-spring systems themselves that are supposed to provide the nonlinearity. Along similar lines, Nakajima et al. (2013) and Hoffmann and Müller (2017) maintained that the complex nonlinear dynamics of soft body structures such as the octopus arm might serve as computational reservoirs, high-dimensional dynamic systems, that can be exploited to perform nonlinear computations.

As Cheng highlights, morphological computation is a fruitful concept for researchers interested in situated cognition. But, as we have noted, it is also heterogenous. We agree with Cheng that the quasi-articulated octopus arm is a case of morphological decomputation because the non-neural body reduces the computational load on the CNS but does not itself play a direct computational role. However, the theoretical work just cited suggests a stronger or more literal kind of morphological computation in which the non-neural body actually performs computations, although it remains to be seen whether this kind of computation occurs in real biological systems.

Mutual Manipulability: New Applications and Next Steps

It is encouraging to see that the mutual manipulability criterion (MM) is finding application in biology (Japyassú, 2017; Japyassú & Laland, 2017). The explicit goal in the original publication by Kaplan (2012) was to shift the direction of the debate about the embodiment and extension away from what he called “proprietary demarcation criteria” for determining the boundaries of cognition, which require problematic assumptions about the nature of cognition. MM was proposed as a way to steer the discussion down more fruitful and empirically grounded paths. Because MM reflects general interventional strategies used by scientists to experimentally test and determine mechanism boundaries, there was always latent potential for wider application of these ideas beyond the initial context of human cognition. Now that MM has proven useful for probing questions about extended cognition in spiders (Japyassú & Laland, 2017), it will be interesting to see how widely it can be applied to other biological taxa.

MM thus appears to be useful generally for detecting when some bodily or environmental feature should count as a real component as opposed to serving merely as a causally relevant background condition. But there is a residual worry that MM is still not restrictive enough, if MM is satisfied merely, as Cheng puts it, “when causal influence flows both ways, from object to brain and from brain to object” (p. 6). This worry should give experimental biologists wishing to employ MM some reason to pause. Craver (2007, p. 158) worried, for example, that MM might allow the heart to qualify as a component in the mechanism underlying performance in word-stem completion tasks, even though it seems to be more appropriately described as a causal background condition for that performance. Lesioning or stimulating the heart in a bottom-up experiment would almost certainly disrupt a subject’s ability to perform word-stem completion. One could imagine that by engaging a subject in an especially arduous version of a word-stem completion task, the heart might start beating faster. According to MM, then, the heart would count as a component.

Craver suggests that in cases like this we might appeal to the additional fact that interventions on background conditions as opposed to components will tend to be relatively nonspecific in their effects. Lesioning or stimulating the heart, to the extent that it has any effects at all, will likely have diffuse effects that disrupt or compromise performance in the target task of word-stem completion and many other tasks besides. Relatedly, interventions on background conditions will tend not to have subtle, differential effects on word-stem completion (e.g., error rates are unlikely to vary parametrically with lesion size or stimulation intensity). Instead, these interventions will tend to have unsubtle, all-or-none effects (e.g., complete extinguishing of task performance).

This idea of specificity has been subject to considerable discussion under the rubric of causal specificity (Griffiths et al., 2015; Waters, 2007; Woodward, 2010). Woodward (2010) used a simple example to help illustrate the basic idea. He asked us to consider the difference between the tuning dial and the on/off switch on a radio. Both are capable of exerting a causal influence on what the listener hears: The dial must be appropriately tuned to a specific frequency (e.g., 105.7 MHz), and the on/off switch must be in the “on” state. But important to note, the dial and switch seem to differ with respect to the degree to which they are “specific” to their effects. The tuning dial has relatively fine-grained causal influence over which station is heard by the listener (assuming the switch is on) in the sense that there are many possible values the dial can take (the cause variable), which result in correspondingly many differences in the station that is heard (the effect variable). The dial is thus a relatively specific cause of which station is heard. By contrast, although the on/off switch is also causally relevant to whether any station is heard at all, it has no causally specific influence on which particular station is heard. The switch is only a relatively nonspecific cause of whether a station is heard.

This qualitative distinction between specific and nonspecific causes has recently been made more precise by applying tools from information theory. Griffiths et al. (2015) suggested that causal specificity can be measured in terms of the reduction in uncertainty about the value of the effect variable E that results from intervening to set the value of the cause variable C (i.e., the mutual information between E and C).

How, then, might this distinction between dial-like or specific causes, on one hand, and switchlike or nonspecific causes, on the other, be leveraged for thinking about extended components in cognition? Consider the interesting case of extended spider cognition discussed at some length by Cheng. The experimental results seem to show that MM is satisfied. Specifically, experimental interventions to increase thread tension in particular areas of the web change spider foraging behavior. In the reverse direction, interventions to induce changes in the internal state of the foraging system (changes in spider satiation levels) can lead to changes in web thread tension. But we may also want to ask additional questions about the specificity of these effects. For example, do fine-grained changes in web thread tension change spider foraging behavior in specific or nonspecific ways? Put another way, is web thread tension more dial-like or more switchlike in its influence on spider behavior? There is some evidence that relatively fine-grained changes in web tension can have similarly fine-grained or differential effects on spider foraging behavior and attentional state (Japyassú & Laland, 2017; Nakata and Zschokke, 2010; Watanabe, 2000). In the other direction, do fine-grained changes in spider satiation levels change web thread tension in specific or nonspecific ways? At least one experimental study seems to suggest that they do (Japyassú & Laland, 2017; Watanabe, 2000).

An interesting next step for investigations of extended cognition like these might be to compare the specificity “profiles” of internal versus external components using well-defined information-theoretic measures like mutual information (Griffiths et al., 2015). Because interventions on typical (i.e., internal) components will tend to produce relatively specific downstream effects as compared to those on causal background conditions, might the same be true for external components? If interventions on external components are equally specific in their effects, this could provide an additional line of support for situated cognition. If their specificity profiles turned out to be systematically different in some way, what would this mean?

We pose these questions not because we have answers but because we strongly suspect that by addressing these and other related questions, research on embodied and extended cognition in nonhuman animals might be propelled into some fruitful and unexpected directions.

References

1. Brooks, R. A. (1991). Intelligence without representation. Artificial Intelligence, 47, 139–159. doi:10.1016/0004-3702(91)90053-M

2. Brooks, R. A. (1999). Cambrian intelligence. Cambridge, MA: MIT Press.

3. Clark, A. (1997). Being there. Cambridge, MA: MIT Press.

4. Clark, A. (2005). Word, niche and super-niche: How language makes minds matter more. Theoria, 20, 255–268.

5. Craver, C. F. (2007). Explaining the brain: Mechanisms and the mosaic unity of neuroscience. New York, NY: Oxford University Press.

6. Flash, T., & Sejnowski, T. J. (2001). Computational approaches to motor control. Current Opinion in Neurobiology, 11, 655–662. doi:10.1016/S0959-4388(01)00265-3

7. Griffiths, P. E., Pocheville, A., Calcott, B., Stotz, K., Kim, H., & Knight, R. (2015). Measuring causal specificity. Philosophy of science, 82, 529–555. doi:10.1086/682914

8. Hauser, H., Ijspeert, A. J., Füchslin, R. M., Pfeifer, R., & Maass, W. (2011). Towards a theoretical foundation for morphological computation with compliant bodies. Biological Cybernetics, 105, 355–370. doi:10.1007/s00422-012-0471-0

9. Hauser, H., Ijspeert, A. J., Füchslin, R. M., Pfeifer, R., & Maass, W. (2012). The role of feedback in morphological computation with compliant bodies. Biological Cybernetics, 106, 595–613. doi:10.1007/s00422-012-0516-4

10. Hoffmann, M., & Müller, V. C. (2017). Simple or complex bodies? Trade-offs in exploiting body morphology for control. In G. Dodig-Crnkovic& R. Giovagnoli (Eds.), Representation and reality in humans, other living organisms and intelligent machines (pp. 335–345). New York, NY: Springer. doi:10.1007/978-3-319-43784-2_17

11. Hutchins, E. (2010). Cognitive ecology. Topics in Cognitive Science, 2, 705–715. doi:10.1111/j.1756-8765.2010.01089.x

12. Japyassú, H. F. (2017). Plasticity and cognition in spiders. In C. Viera & M. O. Gonzaga (Eds.), Behaviour and ecology of spiders (pp. 383–396). New York, NY: Springer. doi:10.1007/978-3-319-65717-2_14

13. Japyassú, H. F., & Laland, K. N. (2017). Extended spider cognition. Animal Cognition, 20, 375–395. doi:10.1007/s10071-017-1069-7

14. Kaplan, D. M. (2012). How to demarcate the boundaries of cognition. Biology and Philosophy, 27, 545–570. doi:10.1007/s10539-012-9308-4

15. Kirsh, D. (1991). Today the earwig, tomorrow man? Artificial Intelligence, 47, 161–184. doi:10.1016/0004-3702(91)90054-N

16. Menary, R. (2007). Cognitive integration: Mind and cognition unbounded. New York, NY: Springer. doi:10.1057/9780230592889

17. Nakajima, K., Hauser, H., Kang, R., Guglielmino, E., Caldwell, D. G., & Pfeifer, R. (2013). A soft body as a reservoir: Case studies in a dynamic model of octopus-inspired soft robotic arm. Frontiers in Computational Neuroscience, 7. doi:10.3389/fncom.2013.00091

18. Nakata, K., & Zschokke, S. (2010). Upside-down spiders build upside-down orb webs: Web asymmetry, spider orientation and running speed in Cyclosa. Proceedings of the Royal Society of London B: Biological Sciences, 277(1696). doi:10.1098/rspb.2010.0729

19. Pfeifer, R., & Bongard, J. (2006). How the body shapes the way we think: A new view of intelligence. Cambridge, MA: MIT Press.

20. Pfeifer, R., Iida, F., & Lungarella, M. (2014). Cognition from the bottom up: On biological inspiration, body morphology, and soft materials. Trends in Cognitive Sciences, 18, 404–413. doi:10.1016/j.tics.2014.04.004

21. Pfeifer, R., Lungarella, M., & Iida, F. (2007). Self-organization, embodiment, and biologically inspired robots. Science, 318(5853), 1088–1093. doi:10.1126/science.1145803

22. Sumbre, G., Fiorito, G., Flash, T., & Hochner, B. (2006). Octopuses use a human-like strategy to control precise point-to-point arm movements. Current Biology, 16, 767–772. doi:10.1016/j.cub.2006.02.069

23. Sutton, J. (2015). Remembering as public practice: Wittgenstein, memory, and distributed cognitive ecologies. In Mind, Language & Action: Proceedings of the 36th International Wittgenstein Symposium (pp. 409–443). Berlin, Germany: De Gruyter. doi:10.1515/9783110378795.409

24. Watanabe, T. (2000). Web tuning of an orb-web spider, Octonoba sybotides, regulates prey-catching behaviour. Proceedings of the Royal Society of London B: Biological Sciences, 267(1443), 565–569. doi:10.1098/rspb.2000.1038

25. Waters, C. K. (2007). Causes that make a difference. The Journal of Philosophy, 104, 551–579. doi:10.5840/jphil2007104111

26. Webb, B. (1994). Robotic experiments in cricket phonotaxis. In D. Cliff, P. Husbands, J.-A. Meyer, & S. W. Wilson (Eds.), From animals to animats 3: Proceedings of the Third International Conference on Simulation of Adaptive Behavior (pp. ٤٥–54). Cambridge, MA: MIT Press.

27. Wilson, R. A. (1994). Wide computationalism. Mind, 103, 351–372. doi:10.1093/mind/103.411.351

28. Woodward, J. (2010). Causation in biology: Stability, specificity, and the choice of levels of explanation. Biology & Philosophy, 25, 287–318. doi:10.1007/s10539-010-9200-z