This Article Appeared in a Journal Published by Elsevier. the Attached

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy

Behavioural Processes 85 (2010) 246–251

Contents lists available at ScienceDirect Behavioural Processes

journal homepage: www.elsevier.com/locate/behavproc

Functional relationships for determining similarities and differences in comparative cognition

Anthony A. Wright ∗

Department of Neurobiology and Anatomy, Medical School, University of Texas Health Center at Houston, P.O. Box 20708, Houston, TX 77225, United States article info abstract

Article history: Pigeons, capuchin monkeys and rhesus monkeys were trained in nearly identical same/different tasks Received 6 May 2010 with an expanding 8-item training set and showed qualitatively similar functional relationships: increas- Received in revised form 22 July 2010 ing novel-stimulus transfer (i.e., concept learning) as a function of the training-set size and the level of Accepted 30 July 2010 transfer eventually becoming equivalent to baseline training performance. There were also some quantitative functional differences: pigeon transfer increases were more gradual and baseline-equivalent Keywords: transfer occurred at a larger set size (256 items) than for monkeys (128 items). Other pigeon groups Abstract concepts trained at 32 and 64-item initial set sizes showed improved transfer (relative to expanding the 8-item Same–different Cognitive modules training set), equivalent to the monkey species’ transfer at these same training set sizes. This ﬁnding Evolution of equivalent concept learning over a portion of the functional range (8, 32, and 64 items or 64–4096 Monkeys training pairs) is discussed in terms of species differences: carryover effects from smaller-set training, Pigeons evolved neural systems, cognitive and cortical modules, and general distributed learning systems for “higher-order” cognitive abilities. © 2010 Elsevier B.V. All rights reserved.

Recent articles have rekindled the debate about which human The approach taken in this article is that functional relation- cognitive abilities (e.g., tool use, teaching, social competence, ships can better identify similarities and differences in cognitive theory of mind, abstract-concept learning, analogical reasoning, ability. Functional relationships are measures of cognitive behavior domain-general cognition) are unique and distinguish humans at several points along a continuum of a critical variable. Functional from other animals (Herrmann et al., 2007; Penn et al., 2008; relationships can identify aspects that are similar and aspects that Premack, 2007, 2010). are different. Tests at only a single value (on the continuum) are Proposals of special cognitive abilities are not limited to humans. more likely to discover differences with little hope of discovering Comparative cognition has a long history of focusing on which qualitative similarity between functional relationships because lots species can or cannot learn certain cognitive tasks. Hierarchies of of variables can affect the absolute performance level. cognitive ability have been constructed on the basis of such learning Similarities and differences in cognitive abilities may be divided differences often with abstract-concept learning or analogical rea- into two general categories: qualitative similarity and quantita- soning at the top (e.g., D’Amato et al., 1985; Herman et al., 1989; tive similarity. From the standpoint of how some cognitive ability Herrnstein, 1990; Premack, 1978, 1983a,b; Thomas, 1980, 1996; works, qualitative similarity is perhaps the more important. Nev- Thompson, 1995; Thompson and Oden, 2000). For the most part, ertheless, I will argue that functional relationships are necessary to the concentration is on which species has what cognitive ability show either. Take the example of visual list memory across species and which species does not (cf., Tomasello and Call, 1997). Many as diverse as pigeons, new and old world monkeys, and humans. tests are all-or-none; the species learns or not, transfers or not, All of these species showed similar changes in their serial position etc. Learning and (transfer) performance can fail in many more functions (SPFs) as the retention delay is increased (e.g., see Wright, ways than it can succeed; such failures, including partial failures, 1998, 2007; Wright et al., 1985). The SPFs showed strong recency are difﬁcult to interpret (e.g., Macphail, 1985, 1996). May be the effects (last item memory) at short retention delays. The primacy right experiment just was not conducted for the task to be learned effect (good ﬁrst item memory) came in at intermediate delays and or transfer performance to be equivalent to the baseline training strengthens as delay is extended. The recency effect waned as the performance. primacy effect strengthened. These similar SPFs showed qualitative similarity in visual list-memory processing across these species. Notwithstanding the qualitative similarity, there were some quan- ∗ Corresponding author. Tel.: +1 713 500 5627. titative differences. For example, the time scale of these changes E-mail address: [email protected]. varied across species with these changes taking place most rapidly

A.A. Wright / Behavioural Processes 85 (2010) 246–251 247

Fig. 1. Schematic of same and different trial displays. In the actual displays, the pictures were in color without labels and the rectangle was white all on a black background. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.) for pigeons (∼10 s), intermediate for monkeys (∼30 s), and most slowly for humans (∼100 s). If list memory had been tested at one retention delay, then a likely conclusion would have been qualitative differences in species’ visual list memory due to differences in primacy and recency effects. The results showing basically similar patterns of SPF changes with retention delay are the basis for concluding qualitative species similarity in visual list memory. The results showing different time-course differences among species are the basis for concluding quantitative species differences in visual list memory. One purpose of this article is to show qualitative similarity between pigeons, rhesus monkeys and capuchin monkeys in their functions relating abstract-concept learning to size of the training set. Another purpose of this article is to show quantitative as well as qualitative similarity among these species. Pigeons will be shown to transfer to novel stimuli at a level equivalent to the mon- Fig. 2. Set-size functions of baseline and transfer performance for pigeons (blue), keys under conditions without detrimental carryover effects from rhesus monkeys (red), and capuchin monkeys (green) as their training sets were pro- prior training despite vastly different neural architectures and evo- gressively expanded from 8 items to 128 items (monkeys) or 1024 items (pigeons). lutionary histories. Although this is a special set of circumstances Error bars are standard errors of the mean. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.) producing equivalent abstract-concept learning, it is a rare example (maybe unique in abstract-concept/rule learning) where pigeons show equivalent abstract-concept learning to monkeys—and two 2. Learning, transfer, abstract-concept learning, and monkey species in this case. qualitative similarity

1. Monkeys and pigeons trained and tested with an Learning rates on the initial 8-item task were similar for the expanding 8-item training set same/different training and 3 species—30-35 sessions. Following learning, a transfer test was testing procedures conducted. Test pictures on transfer trials were novel. Correct choices were rewarded as they were on baseline trials. Transfer Rhesus monkeys (3), capuchins monkeys (3), and pigeons tests consisted of 6 consecutive transfer test sessions (90 baseline (4) were trained and tested with expanding training-sets in a trials plus 10 intermixed transfer trials). The transfer tests follow- same/different (S/D) task. They were trained with the same items, ing 8-item set training showed little or no transfer for monkeys the same displays, the same choice responses, the same visual- or pigeons. As the training set was progressively doubled to 1024 angles of the displays, the same performance criteria, the same items, transfer increased for all species. degree of set-size expansions, and the same test items. Test cham- Fig. 2 shows transfer as a function of training set size. These bers employed video monitors and touch screens mounted ﬂush in functional relationships provide a better comparison of abstract- the stimulus panels. Also mounted on the stimulus panel was either concept learning among species than can be provided by any a pellet cup (capuchins), a juice tube and a pellet cup (rhesus), or single point of comparison. Functional relationships show how a grain hopper (pigeons). For rhesus monkeys and pigeons trials abstract-concept learning changes over a substantial portion of began with the presentation of an upper picture which they had the manipulable range and multiple points for statistical power. to touch 10 times or peck 20 times before being presented with a Fig. 2 shows that the transfer function for pigeons was substantially lower picture and white rectangle, shown as gray and to the right lower than that for monkeys—across much of the set-size range (see of the lower picture in Fig. 1. Capuchin monkeys had both pictures Katz and Wright, 2006). Nevertheless, pigeons like monkeys even- along with a white rectangle presented simultaneously. In all cases, tually showed transfer greater than 80% correct and equivalent to if the two pictures were the same, then a touch/peck to the lower their baseline performance—signifying that pigeons like monkeys picture was correct; if different, then a touch/peck to the white rect- can perform as well with novel stimuli as with training stimuli. angle was correct. Sessions were composed of 100 trials (50 same Others have tested pigeons for S/D abstract-concept learning, and 50 different trials, each separated by 15-s intertrial intervals). but training difﬁculties have led some researchers to use arrays of Subjects were trained until performance was ≥85% (except at the same vs. different stimuli to train pigeons. Unfortunately, pigeons initial set size of 8 items where the criterion was three consecu- trained with stimulus arrays do not perform accurately when the tive sessions ≥80%) (see Katz and Wright, 2006; Katz et al., 2002; stimulus array is reduced to a single pair of stimuli (e.g., Cook et Wright et al., 2003; Wright and Katz, 2006 for additional details.). al., 1997; Young et al., 1997). In other cases, too few stimuli (col- Author's personal copy

248 A.A. Wright / Behavioural Processes 85 (2010) 246–251 ored geometric shapes) may have been used, limiting the number abstract concept and applying this rule on the first presentation of of rule exemplars, and resulting in incomplete (i.e., partial) concept new training pairs (i.e., transfer) with increasing frequency as the learning (e.g., Blaisdell and Cook, 2005). Even humans require vari- set size was expanded. ation in the exemplars to adequately learn rules (e.g., Chen and Mo, 2004). Partial concept learning tends to be inconclusive. A subject 4. Training and testing pigeons on initial 32- and 64-item that has learned the S/D concept ought to be as accurate with pairs sets of novel stimuli as with pairs of training stimuli. If not, then it is hard to say what is controlling behavior. In this section, evidence is presented from groups of experimen- The result of full concept learning (transfer equivalent to tally naïve pigeons trained on larger (32 and 64) initial training baseline performance) in Fig. 2 shows qualitative similarity in sets to evaluate effects of prior training on subsequent transfer abstract-concept learning across rhesus monkeys, capuchins mon- (Nakamura et al., 2009). The training procedures were the same keys and pigeons in these similar S/D tasks. The pigeons’ more as those previously described for pigeons, except that one group shallow transfer function resulted in pigeons requiring a set size of began training with the 32-item training set and another group 256 items to achieve full concept learning as opposed to a set size began training with the 64-item training set. The training items of 128 items for monkeys. This latter finding is a quantitative dif- were the same items as in the 32- and 64-item sets for the expanded ference for these species in this similar S/D task. This quantitative 8-item group training. We were surprised to find better transfer difference between monkeys and pigeons is not particularly sur- in these experiments, suggesting that there was an adverse carry- prising; we have seen similar differences many times over several over effect from the prior 8-item group that retarded subsequent decades of cognitive research. What was surprising, however, was transfer when the training set was expanded. that pigeons do have the cognitive ability to fully learn an abstract S/D concept. In a subsequent section of this article, the quantitative difference in species transfer is further explored with a surprising 5. Improved pigeon learning and transfer finding that pigeons under special training conditions show quan- titatively similar transfer to monkeys. But first, it will be necessary The learning rates of the 32-item and 64-item groups were sim- to show that the transfer results shown in Fig. 2 were not the result ilar to each other and similar to the previous 8-item group (see of stimulus generalization. Nakamura et al., 2009 for details). This was a surprising result because these groups had 16 times or 64 times more training pairs than the 8-item group. The only way these groups could have 3. Learning and transfer is not explained by stimulus learned at rates similar to the 8-item group was if they learned the generalization relationship between the pairs of stimuli and the resulting relational learning (i.e., abstract-concept learning) fed back on itself to We tested whether the transfer results shown in Fig. 2 was based enhance learning of other stimulus pairs. On one hand, it is some- upon generalization from training stimuli as opposed to learning a what strange that a more complex task with more stimuli should rule or an abstract concept (Wright and Katz, 2007). For general- be learned at the same rate as a simpler one with fewer stimuli. ization to account for transfer, subjects would have to be learning On the other hand, the greater number of stimuli produced more individual training pairs. Generalization cannot be based on indi- exemplars of the rule (concept). Thus, if pigeons in the 32-item and vidual pictures because all pictures were randomized with regard 64-item groups are learning the rule or abstract concept, as they to same and different trials and top and bottom positions. We tested appear to do, then this faster than expected learning makes perfect the trials-to-acquisition for expanding training sets according to sense. the generalization hypothesis. Predictions about trials to acqui- Additional support for abstract-concept learning enhancing sition for expanded training sets were made according to initial acquisition comes from the transfer results. Fig. 3 shows transfer 8-item learning rates (learning trials per training stimulus pair), from the groups initially trained with 32 and 64 items compared to adjusted for transfer according to the generalization hypothesis as transfer at these same training sets from the group trained initially the training set was expanded. Predictions according to the gener- with 8 items and then expanded to 32 and 64 training items. The 32- alization hypothesis were that trials-to-acquisition would increase item and 64-item groups transferred better than the 8-item group quite rapidly with successive doublings of the training set (train- did at those same 32- and 64-item training sets following set-size ing pairs increase as the square of number of training stimuli). expansion. Apparently, the pigeons in the expanded 8-item group By contrast, the obtained results showed no increase in trials-to- experienced carryover effects from their prior smaller set training acquisition and in most cases actually decreased with successive that adversely affected subsequent learning and transfer. Indeed, doublings of the training set. The predicted vs. obtained func- the group trained with the initial 64-item set transferred at a level tions diverged so rapidly that at set sizes for full concept learning equivalent to their baseline performance with 64 training items as (128/256 items for monkeys and pigeons, respectively), there were opposed to 256 training items for the 8-item group with expand- differences of as much as three orders of magnitude (1000 times). ing training sets. In this respect alone, it might be concluded that This analysis showed that stimulus generalization was not instru- pigeons have the cognitive ability to achieve full concept learning at mental in the learning or transfer by these animals. a smaller training set size (64 vs. 128) than either monkey species. The results of this analysis do, however, make perfect sense from But such a conclusion would not take into account the increasing the standpoint of learning the abstract concept based upon learn- and superior baseline performance exhibited by the monkeys as ing the relationship between the stimuli. Progressively increasing the set size was expanded, or the fact that (other) monkeys were the number of new stimuli progressively increased the number of not tested for adverse carryover effects like pigeons were. exemplars of the rule and would therefore be expected to accelerate A possible explanation for the adverse carryover effects is that abstract-concept learning. Additionally, when these species were pigeons might experience more severe restricted-domain rela- transferring at a level comparable to their baseline performance, tional learning than monkeys (Wright and Katz, 2009). The domain they often reached criterion so rapidly that the vast majority of representing the pigeons’ relational learning might persist and new training pairs (99% for rhesus and capuchin monkeys and 98% carryover into subsequent training (e.g., with the next expanded for pigeons) had not been seen even once. The only way that these training set). The role of old training items in the expanded train- subjects could produce such a result would be by learning the S/D ing set might “set the occasion” for restricted-domain relational Author's personal copy

A.A. Wright / Behavioural Processes 85 (2010) 246–251 249

Fig. 3. Mean baseline and transfer performance for the 32-item and 64-item groups trained and tested initially at those item sets (32- and 64-item initial sets) on the right compared to mean baseline and transfer performance by the 8-item group on the left trained initially with an 8-item and subsequently expanded to 32- and 64-item training sets (8-item expanding set). Dotted line is chance performance. Error bars are standard errors of the mean. learning and thereby retard abstract-concept learning that other- al., 2009 for details). Pigeons may have an unfair advantage because wise would mutually beneﬁt performance with new item pairs and monkeys were not tested for adverse carryover effects. But in our enhance novel-item transfer. experience monkeys are less prone to such effects (e.g., monkeys learned expanded training sets more rapidly than pigeons; see 6. Improved pigeon concept learning approaches that for Wright and Katz, 2006; Wright and Katz, 2007); moreover, the monkeys expense of those monkey experiments would be prohibitive. Nev- ertheless, the similar functional relationships for abstract-concept Improved levels of transfer from the 32- and 64-item pigeon learning shown in Fig. 4 for monkeys and pigeons are remarkable groups are combined with the 8-item group’s (initial) transfer in showing that pigeons do have the cognitive capability to show to form a functional relationship. This functional relationship is abstract-concept learning at levels equivalent to monkeys under at then compared to the functional relationships from the rhesus least some condition. and capuchin monkeys in Fig. 4. In this case, the different pigeon groups show transfer levels similar to monkeys (see Nakamura et 7. Discussion

The goal of comparative cognition is to compare different species’ cognitive abilities. The most valid comparisons, in my opinion, come from tasks that are as identical as possible, that manipulate critical parameters over a substantial range, and that compare functional relationships for different species. In this article, pigeons and two monkey species were trained and tested in nearly identical S/D abstract-concept learning tasks with the same items, same visual displays, similar parameters and manip- ulations, producing functional relationships for concept learning over the set-size range of 8–256 training items. The functional relationships of Fig. 2 show: qualitatively similar relationships of abstract-concept learning increases as a function of training set size for pigeons and monkeys; a more rapidly increasing function for monkeys than pigeons; and levels of transfer performance for each species that (eventually) matched that species’ baseline performance demonstrating “full” (vs. partial) concept learning. The functional relationships of Fig. 4 show that pigeons are quantita- tively as well as qualitatively similar to monkeys when pigeons were initially trained at set sizes of 8, 32, or 64 items. The comparison of functional relationships shown in Fig. 4 are remarkable because they show that pigeons ultimately have the cognitive ability to learn the abstract S/D concept at the same level as monkeys over several set-size values where numbers of rule exemplars Fig. 4. Initial transfer performance of three pigeon groups trained with 8, 32, or 64 increased from 64 to 4096 pairs. items compared to transfer performances by rhesus and capuchin monkeys initially trained with the 8-item set and progressively expanded to the 32- and 64-item Similarities and differences among the species’ functional training sets. Error bars are standard errors of the mean. relationships have implications for the role of different neu- Author's personal copy

250 A.A. Wright / Behavioural Processes 85 (2010) 246–251 ral architectures and evolutionary histories in abstract-concept 459), this concept of sameness may also be the keel and backbone learning. Among neural structures that play a critical role in of nonhuman species’ cognition as well. abstract-concept learning, the prefrontal cortex (PFC) stands out. The PFC is a relatively late appearing neural structure and shows Acknowledgements perhaps the most variability across primates, including new world (e.g., capuchins) and old world (e.g., rhesus) primates (e.g., Rosa and This research was supported by NIH grants MH-61798 and MH- Tweedale, 2005). The prefrontal cortex is among the most phyloge- 072616. I thank Jeff Katz, Mauricio Papini, Caitlin Elmore, and netically recent regions of the human brain, and ontogenically, it is Allison Foote for their many insightful comments on an earlier draft one of the last to mature (Fuster, 2002; Gogtay et al., 2004; Harris et of this manuscript. al., 2009). Pigeons have neither neocortex nor PFC (e.g., Diekamp et al., 2002). Nevertheless, recent evidence indicates a possible ana- logue to the mammalian PFC, the caudolateral nidopallium, with References several common characteristics including: Pallium derived (as is Baddeley, A.D., 2002. Is working memory still working? Eur. Psychol. 7, 85–97. the PFC in primates) but with nuclear instead of mammalian lami- Barrett, H.C., Kurzban, R., 2006. Modularity in cognition: framing the debate. Psychol. nar organization (e.g., Emery, 2006; Jarvis et al., 2005; Reiner et al., Rev. 113, 628–647. 2005); similar functions of top-down control, attention, planning, Blaisdell, A.P., Cook, R.G., 2005. Two-item same-different concept learning in pigeons. Learn. Behav. 33, 67–77. concept learning (e.g., Emery, 2006; Manns and Gunturkun, 2009); Chiappe, D., MacDonald, K.B., 2005. The evolution of domain-general mechanisms and functions manipulated by dopamine (e.g., Rose et al., 2010). in intelligence and learning. J. Gen. Psychol. 132, 5–40. These differences in neural-architecture raise the issue of how Chen, Z., Mo, L., 2004. Schema induction in problem solving: a multidimensional analysis. J. Exp. Psychol. Learn. Mem. Cogn. 30, 583–600. these different species might have evolved to produce (some) Cook, R.G., Katz, J.S., Cavoto, B.R., 1997. Pigeon same—different concept learning with functional similarities of abstract-concept learning. One possibility multiple stimulus classes. J. Exp. Psychol. Anim. Behav. Process. 23, 417–433. is that these species evolved similar cognitive modules (homo- Cosmides, L., Tooby, J., 1994. Origins of domain specificity: the evolution of functional organization. In: Hirschfeld, L.A., Gelman, S.A. (Eds.), Mapping the Mind. plasies) for abstract-concept learning despite biologically (neural, Cambridge University Press, Cambridge, pp. 85–116. chemical, molecular) different cognitive modules for this func- D’Amato, M.R., Salmon, D.P., Colombo, M., 1985. Extent and limits of the matching tion (e.g., Papini, 2002, 2003). Traditionally, proposed cognitive concept in monkeys (Cebus apella). J. Exp. Psychol. Anim. Behav. Process. 11, modules have been bottom-up processing with domain-specific 35–51. Diekamp, B., Kalt, T., Gunturkun, O., 2002. Working memory neurons in pigeons. J. input mechanisms, for example, a face-specific mechanism for con- Neurosci. 22, 210–215. specific recognition (e.g., Cosmides and Tooby, 1994; Kanwisher, Emery, N.J., 2006. Cognitive ornithology: the evolution of avian intelligence. Phil. 2000, 2006). More recently, however, there have been proposed Trans. R. Soc. B 361, 23–43, doi:10.1098/rstb.2005.1736. Fuster, J.M., 2002. Frontal lobe and cognitive development. J. Neurocytol. 31, cognitive modules with top-down processing and domain-general 373–385. mechanisms, for example, working memory (e.g., Chiappe and Gogtay, N., Giedd, J.N., Lusk, L., Hayashi, K.M., Greenstein, D., Vaituzis, A.C., Nugent MacDonald, 2005) as well as the examples mentioned in the open- 3rd, T.F., Herman, D.H., Clasen, L.S., Toga, A.W., et al., 2004. Dynamic mapping of human cortical development during childhood through early adulthood. Proc. ing sentence of this article. Certainly, top-down, domain-general Natl. Acad. Sci. U.S.A. 101 (21), 8174–8179. mechanisms would more applicable to a rule-based process such Harris, L.W., Lockstone, H.E., Khaitovich, P., Weickert, C.S., Webster, M.J., Bahn, as abstract-concept learning. S., 2009. Gene expression in the prefrontal cortex during adolescence: implications for the onset of schizophrenia. BMC Med. Genomics 2 (28), 1–14, Substantiating and defending modular accounts of cognition, doi:10.1186/1755-8794-2-28. domain specific as well as domain general, is fraught with difficul- Herman, L.M., Hovancik, J.R., Gory, J.D., Bradshaw, G.L., 1989. Generalization of visual ties. First of all, modularity is by its very nature—fractal. As Barrett matching by a bottlenosed dolphin (Tursiops truncates): evidence for invariance of cognitive performance with visual and auditory materials. J. Exp. Psychol. and Kurzban (2006) pointed out, working memory is thought to Anim. Behav. Process. 15, 124–136. have subsystems (e.g., visuospatial sketchpad, phonological loop, Herrmann, E., Call, J., Hernandez-Lloredo, M.V., Hare, B., Tomasello, M., 2007. episodic buffer; Baddeley, 2002), and each subsystem can be Humans have evolved specialized skills of social cognition: the cultural intelli- thought of as a (more specialized) module—perhaps resulting in gence hypothesis. Science 317, 1360–1366. Herrnstein, R.J., 1990. Levels of stimulus control: a functional approach. Cognition an assembly of interconnected modules. But where does it end? 37, 133–166. Second, it is unclear how cognitive modules, “hypothetical men- James, W., 1890/1950. The Principles of Psychology, vol. 1. Dover Publications, New tal subroutines for performing specialized cognitive functions”, are York (Original work published 1890). Jarvis, E., et al., 2005. Avian brains and a new understanding of vertebrate brain related to cortical modules, “cortical circuits that consist of verti- evolution. Nat. Rev. Neurosci. 6 (2), 151–159. cal, interconnected columns of neurons” (e.g., Mercado, 2009,p.2). Katz, J.S., Wright, A.A., 2006. Same/different concept learning by pigeons. J. Exp. Relating cognitive to cortical modules is a basic issue for all of com- Psychol. Anim. Behav. Process. 32, 80–86. Katz, J.S., Wright, A.A., Bachevalier, J., 2002. Mechanisms of same/different abstract- parative cognition and cognitive neuroscience, particularly since all concept learning by rhesus monkeys (Macaca mulatta). J. Exp. Psychol. Anim. cognition must have a neural basis. Third, cognitive modules may be Behav. Process. 28, 358–368. useful devices for explaining (or speculating about) species’ cogni- Kanwisher, N., 2000. Domain specificity in face perception. Nat. Neurosci. 3, 759–763. tive differences. But in their present state of development, cognitive Kanwisher, N., 2006. What’s in a face? Science 311, 617–618. modules are much less useful in explaining cognitive similarities. Macphail, E.M., 1985. Vertebrate intelligence: the null hypothesis. Phil. Trans. R. Soc. Consider some findings discussed in the current article. How could Lon. Ser. B 308, 37–51. Macphail, E.M., 1996. Cognitive function in mammals: the evolutionary perspective. invoking differently evolved cognitive modules for pigeons and Cogn. Brain Res. 3, 279–290. old- and new-world monkeys possibly account for similarities in Manns, M., Gunturkun, O., 2009. Dural coding of visual asymmetries in the pigeons these species’ abstract-concept learning? Indeed, the real issue may brain: the interaction of bottom-up and top-down systems. Exp. Brain Res. 199, be just the reverse. How could differently evolved cognitive and 323–332, doi:10.1007/s00221-009-1702-z. Mercado III, E., 2009. Cognitive plasticity and cortical modules. Curr. Dir. Psychol. “cortical” modules possibly end up—by chance—producing similar Sci. 18 (3), 153–158, doi:10.1111/j.1467-8721.2009.01627.x. abstract-concept learning under any conditions, including those in Nakamura, T., Wright, A.A., Katz, J.S., Bodily, K.D., Sturz, B.R., 2009. Abstract-concept Fig. 4? learning carryover effects from the initial training set in pigeons (Columba livia). J. Comp. Psychol. 123, 79–89. Possibly S/D abstract-concept learning is a general (rather Papini, M.R., 2002. Pattern and process in the evolution of learning. Psychol. Rev. than specific) property of learning systems because of the rela- 109 (1), 186–201. tional requirements for learning and cognition. Taking a lead from Papini, M.R., 2003. Comparative psychology of surprising nonreward. Brain Behav. Evol. 62 (2), 83–95. William James, who 120 years ago said that the concept of same- Penn, D.C., Holyoak, K.J., Povinelli, D.J., 2008. Darwin’s mistake: explaining the dis- ness is “the very keel and backbone of our thinking” (James, 1890,p. continuity between human and nonhuman minds. Behav. Brain Sci. 31, 109–130. Author's personal copy

A.A. Wright / Behavioural Processes 85 (2010) 246–251 251

Premack, D., 1978. On the abstractness of human concepts: why it would be difﬁ- Thompson, R.K.R., Oden, D.L., 2000. Categorical perception and conceptual judg- cult to talk to a pigeon. In: Hulse, S.H., Fowler, H., Honig, W.K. (Eds.), Cognitive ments by nonhuman primates: the paleological monkey and the analogical ape. Processes in Animal Behavior. Erlbaum, Hillsdale, NJ, pp. 423–451. Cogn. Sci. 24, 363–396. Premack, D., 1983a. Animal cognition. Ann. Rev. Psychol. 34, 351–362. Tomasello, M., Call, J., 1997. Primate Cognition. Oxford University Press, New York. Premack, D., 1983b. The codes of man and beasts. Behav. Brain Sci. 6, 125–167. Wright, A.A., 1998. Auditory and visual serial position functions obey different laws. Premack, D., 2007. Human and animal cognition: continuity and discontinuity. Proc. Psychol. Bull. Rev. 5, 564–584. Natl. Acad. Sci. U.S.A. 104, 13861–13867. Wright, A.A., 2007. An experimental analysis of memory processing. J. Exp. Anal. Premack, D., 2010. Why humans are unique: three theories. Pers. Psychol. Sci. 5 (1), Behav. 88, 405–433. 22–32. Wright, A.A., Katz, J.S., 2006. Mechanisms of same/different concept learning in Reiner, A., Yamamoto, K., Karten, H.J., 2005. Organization and evolution of the avian primates and avians. Behav. Process. 72, 234–254. forebrain. Anat. Rec. 287A, 1080–1102. Wright, A.A., Katz, J.S., 2007. The generalization hypothesis of abstract-concept Rosa, M.G.P., Tweedale, R., 2005. Brain maps, great and small: lessons from compar- learning: learning strategies and related issues in Macaca mulatta, Cebus apella ative studies of primate visual cortical organization. Phil. Trans. R. Soc. B 2005, and Columba livia. J. Comp. Psychol. 121, 387–397. 665–691, doi:10.1098/rstb.2005.1626. Wright, A.A., Katz, J.S., 2009. A case for restricted-domain relational learning. Psy- Rose, J., Schiffer, A.M., Dittrich, L., Gunturkun, O., 2010. The role of dopamine in main- chol. Bull. Rev. 16, 907–913 (PMID: 19815797). tenance and distractability of attention in the “prefrontal cortex” of pigeons. J. Wright, A.A., Santiago, H.C., Sands, S.F., Kendrick, D.F., Cook, R.G., 1985. Memory Neurosci. 167, 232–237. processing of serial lists by pigeons, monkeys, and people. Science 229 (19), Thomas, R.K., 1980. Evolution of intelligence: an approach to its assessment. Brain 287–289. Behav. Evol. 17, 454–472. Wright, A.A., Rivera, J.J., Katz, J.S., Bachevalier, J., 2003. Abstract-concept learning Thomas, R.K., 1996. Investigating cognitive abilities in animals: unrealized potential. and list-memory processing by capuchin and rhesus monkeys. J. Exp. Psychol. Cogn. Brain Res. 3, 157–166. Anim. Behav. Process. 29, 184–198. Thompson, R.K.R., 1995. Natural and relational concepts in animals. In: Roitblat, Young, M.E., Wasserman, E., Garner, K.L., 1997. Effects of number of items on the H.L., Meyer, J.A. (Eds.), Comparative Approaches to Cognitive Science. MIT Press, pigeon’s discrimination of same from different visual displays. J. Exp. Psychol. Cambridge, MA, pp. 175–224. Anim. Behav. Process. 23, 491–501.