This PDF includes a chapter from the following book:

The Cognitive Empirical and Theoretical Perspectives on Animal Cognition

© 2002 Massachusetts Institute of Technology

License Terms: Made available under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International Public License https://creativecommons.org/licenses/by-nc-nd/4.0/

OA Funding Provided By: The open access edition of this book was made possible by generous funding from Arcadia—a charitable fund of Lisbet Rausing and Peter Baldwin.

The title-level DOI for this work is: doi:10.7551/mitpress/1885.001.0001

Downloaded from http://direct.mit.edu/books/edited-volume/chapter-pdf/677490/9780262268028_c001600.pdf by guest on 29 September 2021 17 Spatial and Social Cognition in Corvids: An Evolutionary Approach

Russell P. Balda and Alan C. Kamil

research plan using controlled laboratory ex- Research Questions periments and captive . Fortunately, nut- crackers are quite willing to cache and recover The central research questions that have guided seeds in laboratory settings and do so with a high our studies since 1981 combine issues and tech- degree of accuracy, both in a sandy floor indoors niques from both comparative psychology and (Balda 1980; Balda and Turek 1984) or out of avian ecology. Most of our questions originate doors (Vander Wall 1982), as well as in a room from the cognitive implications of extensive field with a raised floor containing sand-filled cups as studies on the natural history, ecology, and potential cache sites (Kamil and Balda 1985). behavior of seed-caching corvids. Because our The ability to study caching and cache recovery questions have evolved as our studies progressed, under controlled laboratory conditions allowed we have chosen to give a historical perspective us to test hypotheses on how the nutcrackers find outlining the progression of our ideas and ques- their caches. tions (see Shettleworth, chapter 16 in this volume For example, because we were able to control for a description of a similar program with seed- when and where the birds cached, we were able caching tits and chickadees). to rule out odor, marking the site, list learning, and site preferences (Kamil and Balda 1985). We Research Paradigm also learned that these birds remember some cache sites better than others and recover food Our research program began by examining the from the better-remembered sites first, and with amazing spatial memory system of the Clark’s greater accuracy (Kamil and Balda 1990a). Birds (Nucifraga columbiana). A single sometimes revisit cache sites after they have nutcracker buries up to 33,000 food items in recovered the seeds. On these revisits they treat thousands of di¤erent subterranean sites and the cache site di¤erently than when they pre- retrieves them months later with a high degree of viously emptied it (Kamil et al. 1993). These accuracy. These birds are highly adapted for this birds also showed a long retention interval for behavior because they possess a strong, sharp bill cache memory, recovering caches with high for opening cones, extracting seeds, and burying levels of accuracy up to 9 months after creating them in the substrate; a sublingual pouch (Bock them (Balda and Kamil 1992). et al. 1973) for carrying large numbers of seeds The results of several studies showed that nut- (up to 90 pinyon , , seeds); and crackers were using visual landmarks for accu- strong wings for carrying seeds great distances rate cache recovery (Vander Wall 1982; Balda (up to 22 km). Birds have been observed digging and Turek 1984). Data obtained by using a up seeds in the field with seemingly uncanny ac- clock-shift technique, popular in studies of mi- curacy (Vander Wall and Balda 1977, 1981; gratory birds and homing pigeons, suggest that Vander Wall and Hutchins 1983). Although this seed-caching corvids may use the sun as a com- behavior occurs regularly in the field, field con- pass under some circumstances (Wiltschko et al. ditions do not allow the design of studies to ad- 2000). Thus we successfully brought a behavior dress the questions of how nutcrackers are able prominent in the field into the laboratory, where to locate their stored food. we could examine it in great detail. From these Studies of the cognitive mechanisms involved studies we concluded that nutcrackers were using in cache recovery required the development of a

Downloaded from http://direct.mit.edu/books/edited-volume/chapter-pdf/677490/9780262268028_c001600.pdf by guest on 29 September 2021 Russell P. Balda and Alan C. Kamil 130

a spatial memory system to recover their caches Specificity of Spatial Memory and that this system was of long duration and robust (Kamil and Balda 1990b). This di¤erence in the accuracy of cache recovery raises an interesting issue that is important in Comparative Studies understanding the evolution of cognitive abili- ties: Is this spatial memory restricted to remem- On the slopes of the San Francisco Peaks in bering where food has been stored or is it more Northern Arizona, five species of corvids cache general? Natural selection selects for outcomes, and recover seeds. These species di¤er in their not mechanisms. Thus it could be that the nut- degree of dependence on their seed caches to crackers’ ability is highly specific. On the other survive winter, as well as in their adaptations for hand, selection could have operated to sharpen this behavior. The Clark’s nutcracker is the most already existing spatial cognitive abilities, in highly specialized of these species and lives at which case nutcrackers should perform quite the highest elevations, where winters are harsh well on a variety of spatial tasks. Therefore, we and alternative foods are very scarce. At mid- embarked on a series of comparative studies elevations, moderately specialized Steller’s jays using di¤erent procedures to test spatial mem- ( stelleri) and pinyon jays (Gymno- ory. These included two- and three-dimensional rhinus cyanocephalus) coexist and also cache pine open-room analoges of the radial maze (Ka- seeds when they are available. Both species have mil et al. 1994; Balda et al. 1997) and operant a relatively sharp bill for extracting seeds and an nonmatching-to-sample tests. These studies all expandible esophagus for carrying pine seeds. A involved spatial memory, but not the recovery of pinyon may cache up to 26,000 pine seeds food previously cached. The results of these when cone crops are abundant (Balda 1987). At studies were consistent with our hypothesis. The lower elevations, the less specialized western species most dependent on seed caches for winter scrub jays ( californica) and Mexican survival performed at higher levels. If depen- jays (A. ultrimarina) cache and recover seeds dence on stored food has selected for improved much less intensely than the birds at higher spatial cognitive abilities, it has done so in a way elevations. Scrub and Mexican jays possess no that is not completely domain specific. morphological adaptations for the harvest, However, there is some specificity. Olson transport, caching, and recovery of seeds, and et al. (1995) tested three species in an operant the lower elevations where they live have mild nonmatching-to-sample test. In one experiment, winters and a year-round supply of arthropods, the birds were required to remember a spatial seeds, and berries. location; in another, a color. As in many other These di¤erences in natural history raise a experiments, the most seed-dependent species compelling question about evolution and cogni- performed best in the spatial test, showing tion. Are these di¤erences in morphological ad- much longer retention intervals. However, this aptations for food caching, and the concomitant di¤erence disappeared completely during the dependence on cached food, associated with dif- color test. This suggests a modularity for spatial ferences in the spatial cognitive abilities of these cognition. species? Are species that have the highest level of dependence on cached food also better at finding Social Cognition their caches? In a comparative test with three of these species, we found that although all three More recently, we have become interested in performed above chance, nutcrackers and pinyon extending our natural history-based analysis of jays recovered their caches more accurately than cognition to another domain. Primatologists western scrub jays (Balda and Kamil 1989). (e.g., Humphrey 1976) have developed a hy-

Downloaded from http://direct.mit.edu/books/edited-volume/chapter-pdf/677490/9780262268028_c001600.pdf by guest on 29 September 2021 Spatial and Social Cognition in Corvids 131

pothesis about the evolution of intelligence based with many open holes for caching in the floor. on the cognitive demands of sociality. The observer could view all areas of the that live in large, stable social groups must be floor. Later the observer was allowed to attempt able to assess the consequences of their behav- to recover the caches it had observed being iors, classify other animals as members of various made. While pinyon jays and Mexican jays groups and coalitions, and recognize and remem- recovered caches with an accuracy above chance ber traits of many other individuals. Success levels, nutcrackers did not perform above chance within the group will be improved if an individ- (Bedneko¤ and Balda 1996a,b). ual possesses a rich internal representation of the In another experiment, pinyon jays learning a group that will allow it to adjust to the fluid na- novel task were facilitated by being able to ob- ture of the group. These cognitive demands will serve a conspecific performing the same task, but necessarily increase as group size increases and nutcrackers were not so facilitated (Templeton can be expected to be greatest for those animals et al. 1999). In a third experiment (Bond et al. living in larger, well-structured groups. Although MS), pinyon jays performed better than western the social complexity hypothesis has been con- scrub jays in an operant test of transitive infer- sidered primarily for primates, its logic is general ence. Further studies of the ability to identify and its implications can be tested with any ap- conspecifics and to form equivalence sets are propriately chosen taxon (Balda et al. 1996). planned for the near future. The species we have worked with vary con- siderably in sociality. Pinyon jays are possibly the most social bird in North America, living Methodologies in permanent groups of up to 400 individuals. Many of these individuals never leave their natal We have combined two methodologies. First, flock. Mexican jays are also highly social, living we have used a classical biological comparative in relatively stable groups of 12–18 individuals, method, comparing closely related species that where helping at the nest is especially prominent. di¤er in their adaptations. This necessarily in- The Clark’s nutcracker and western scrub jay, in volves using natural history as a clue for asking contrast, live in family units or pairs year round. relevant questions about relevant species. Princi- Young of the year disperse before the next breed- ples of Darwinian evolution are central for un- ing season. These di¤erences in social living led derstanding the dynamics of biological systems. us to hypothesize that pinyon jays and Mexican The results of our studies led us to an important jays should be able to solve more complex cog- conclusion concerning the evolution of cogni- nitive tests than the less social nutcracker and tion. The accuracy of locating seed caches is western scrub jay. If this hypothesis is correct, an adaptive trait and as such is shaped by the the ordering or gradient of the four species will actions of natural selection. be di¤erent from that along the dependency gra- Cognitive abilities are a part of the adaptive dient; that is, nutcrackers and scrub jays should arsenal that consists of a collective suite of char- perform poorly whereas pinyon jays and Mexi- acters that allows for swift and e‰cient harvest, can jays should demonstrate superior cognitive transport, and caching and then accurate recov- skills. ery of the cache at a later time. As such, cogni- The selection of appropriate tasks for test- tive traits can be viewed as playing a role in the ing this hypothesis presented a challenge. We biological success of an organism, much as mor- began with some comparative studies of obser- phological and physiological traits do. Species vational learning. In one series of experiments, di¤erences reflect, in part, di¤erences in selective nutcrackers, pinyon jays, and Mexican jays pressures among the species that are due to dif- watched a conspecific make caches in a room ferences in their ecologies. Thus, cognitive pro-

Downloaded from http://direct.mit.edu/books/edited-volume/chapter-pdf/677490/9780262268028_c001600.pdf by guest on 29 September 2021 Russell P. Balda and Alan C. Kamil 132

cesses have evolved and must be viewed as bio- another way, how does a nutcracker know when logical processes, not only because they have it is at a cache site? We know that they use the their roots in neurophysiology, but also because position of landmarks to locate caches (e.g., they are biologically significant as adaptations Balda and Turek 1984). Kamil et al. (1999) com- that contribute to the biological success of the pared the body orientations used during caching organisms so endowed. with those used during recovery. We found that Second, we have dealt with the learning– nutcrackers perform just as accurately when they performance distinction that bedevils compar- use di¤erent orientations as when they use the ative studies of cognition by using multiple same orientation. This suggests that they re- behavioral test procedures. That is, if one species member each site separately. Recently, Kamil performs di¤erently from another in any specific and Cheng (2001) suggested that when the dis- test, this could be due to di¤erences in how well tances between the landmarks and the goal are the test situation is suited to each species rather relatively great, as they often are in nature, the than to di¤erences in ability. However, when birds remember the directional relationship be- substantially di¤erent tasks (such as cache re- tween the goal and each of several landmarks. covery and operant tests) produce similar pat- We are beginning a series of experiments to test terns across species, the likelihood that the this hypothesis. behavioral di¤erences are due to real species dif- ferences in ability grows. These varied tests pro- Selection for Spatial Cognition vide converging operations, an approach first outlined by Kamil (1988). The results of comparative studies such as those reviewed here that indicate that spatial cognition is correlated with dependence on cached food Internal States are consistent with the hypothesis that in food- storing, scatter-hoarding species, natural se- Our experiments are designed to inform us about lection has favored spatial abilities. We are the knowledge that our birds possess about their currently attempting a more direct test of the physical and social environments. We do not hypothesized link between cognitive abilities and generally think of this knowledge in terms of in- biological fitness in natural populations by ob- ternal psychological states because the meaning taining measures of spatial abilities and measur- of this term is unclear. We take the term internal ing reproductive success in a wild population of psychological states to be equivalent, at least for pinyon jays. some people, to internal subjective states (e.g., awareness or consciousness), and we have never Social Cognition found it useful to speculate about the subjective states of our birds. Our initial results of the social cognition hy- pothesis (outlined earlier) are quite exciting. We Future Work hope that we and other investigators will be able to expand this work in two ways. First, many Cache Location more taxa need to be studied, so that we have many independent comparisons, each testing the Although it is well established that Clark’s nut- hypothesis (Felsenstein 1985; Kamil 1988). Sec- crackers (and other seed-caching birds) remem- ond, more tests of social cognition need to be ber their cache sites, we know relatively little developed and used in this e¤ort. about exactly what they remember. To put it

Downloaded from http://direct.mit.edu/books/edited-volume/chapter-pdf/677490/9780262268028_c001600.pdf by guest on 29 September 2021 Spatial and Social Cognition in Corvids 133

tory. In Current Ornithology, Vol. 13, V. Nolan and E. From Limits to Opportunities D. Ketterson, eds., pp. 333–366. New York: Plenum. Balda, R. P., Kamil, A. C., Bedneko¤, P. A., and Hile, Yesterday’s limits are often today’s opportu- A. G. (1997). Species di¤erences in spatial memory on nities, and we are reluctant to set limits on the a three-dimensional task. Ethology 103: 47–55. study of animal cognition. However, the study of Bedneko¤, P. A. and Balda, R. P. (1996a). Social the adaptive nature of animal cognition and its caching and observational spatial memory in pinyon evolution is much more di‰cult and challenging jays. Behaviour 133: 807–826. than the study of morphological and physiologi- Bedneko¤, P. A. and Balda, R. P. (1966b). Observa- cal traits for a number of reasons. Animals are tional spatial memory in Clark’s nutcrackers and not necessarily programmed to maximize the Mexican jays. Animal Behaviour 52: 833–839. performance of cognitive behaviors with the Bock, W. J., Balda, R. P., and Vander Wall, S. B. same degree of certainty that they often exhibit (1973). Morphology of the sublingual pouch and in morphological and physiological experiments. tongue musculature in Clark’s nutcrackers. The Auk Cognitive behaviors are often more subtle, are 90: 491–519. governed by very complex and involved neural Bond, A. B., Kamil, A. C., and Balda, R. P. (MS). circuitry that is not obvious, and are not often Social complexity predicts cognitive di¤erences be- performed with maximum intensity. As Hum- tween two corvid species. phrey (1976) once pointed out, we would learn Felsenstein, J. (1985). Phylogenies and the comparative little from watching Albert Einstein through a method. American Naturalist 125: 1–15. pair of binoculars! Humphrey, N. K. (1976). The social function of intel- lect. In Growing Points in Ethology, P. P. G. Bateson and R. H. Hinde, eds., pp. 307–317. Cambridge: References Cambridge University Press. Kamil, A. C. (1988). Synthetic approach to the study Balda, R. P. (1987). Avian impacts on pinyon- of animal intelligence. In Comparative Perspectives in woodlands. In Proceedings of the Pinyon-Juniper Con- Modern Psychology: Nebraska Symposium on Motiva- ference, R. L. Everett, ed., pp. 525–533. Reno, Nev.: tion, Vol. 35, D. W. Leger, ed., pp. 230–357. Lincoln: USDA Forest Service General Technical Report, INT- University of Nebraska Press. 215. Kamil, A. C. and Balda, R. P. (1985). Cache recovery Balda, R. P. (1980). Recovery of cached seeds by a and spatial memory in Clark’s nutcrackers (Nucifraga captive Nucifraga caryocatactes. Zeitschrift fu¨r Tier- columbiana). Journal of Experimental Psychology: Ani- psychologie 52: 331–346. mal Behavior Processes 11: 95–111. Balda, R. P. and Kamil, A. C. (1989). A comparative Kamil, A. C. and Balda, R. P. (1990a). Di¤erential study of cache recovery by three corvid species. Animal memory for di¤erent cache sites by Clark’s nutcrackers Behaviour 38: 486–495. (Nucifraga columbiana). Journal of Experimental Psy- Balda, R. P. and Kamil, A. C. (1992). Long-term spa- chology: Animal Behavior Processes 16: 162–168. tial memory in Clark’s nutcracker, Nucifraga colum- Kamil, A. C. and Balda, R. P. (1990b). Spatial mem- biana. Animal Behaviour 44: 761–769. ory in seed-caching corvids. Psychology of Learning Balda, R. P. and Turek, R. J. (1984). The cache- and Motivation 26: 1–25. recovery system as an example of memory capabilities Kamil, A. C. and Cheng, K. (2001). Way-finding and in Clark’s nutcracker. In Animal Cognition,H.L. landmarks: The multiple-bearings hypothesis. Journal Roitblat, T. G. Bever, and H. S. Terrace, eds., pp. 513– of Experimental Biology 204: 103–113. 532. Hillsdale, N.J.: Lawrence Erlbaum Associates. Kamil, A. C., Balda, R. P., Olson, D. J., and Good, S. Balda, R. P., Kamil, A. C., and Bedneko¤, P. A. (1993). Revisits to emptied cache sites by Clark’s nut- (1996). Predicting cognitive capacity from natural his-

Downloaded from http://direct.mit.edu/books/edited-volume/chapter-pdf/677490/9780262268028_c001600.pdf by guest on 29 September 2021 Russell P. Balda and Alan C. Kamil 134

crackers (Nucifraga columbiana): A puzzle revisited. Animal Behaviour 45: 241–252. Kamil, A. C., Balda, R. P., and Olson, D. J. (1994). Performance of four seed-caching corvids in the radial- arm analog. Journal of Comparative Psychology 108: 385–393. Kamil, A. C., Balda, R. P., and Good, S. (1999). Pat- terns of movement and orientation during caching and recovery by Clark’s nutcrackers (Nucifraga colum- biana). Animal Behaviour 57: 1327–1335. Olson, D. J., Kamil, A. C., Balda, R. P., and Nims, P. J. (1995). Performance of four seed-caching corvid species in operant tests of nonspatial and spatial mem- ory. Journal of Comparative Psychology 109: 173–181. Templeton, J. J., Kamil, A. C., and Balda, R. P. (1999). Sociality and social learning in two species of corvids: The pinyon jay (Gymnorhinus cyanocephalus) and the Clark’s nutcracker (Nucifraga columbiana). Journal of Comparative Psychology 113: 450–455. Vander Wall, S. B. (1982). An experimental analysis of cache recovery in Clark’s nutcracker. Animal Behav- iour 30: 84–94. Vander Wall, S. B. and Balda, R. P. (1977). Coad- aptations of the Clark’s nutcracker and the pinon pine for e‰cient seed harvest and dispersal. Ecological Monographs 47: 89–111. Vander Wall, S. B. and Balda, R. P. (1981). Ecology and evolution of food-storage behavior in conifer-seed- caching corvids. Zeitschrift fu¨r Tierpsychologie 56: 217–242. Vander Wall, S. B. and Hutchins, H. E. (1983). De- pendence of Clark’s nutcracker (Nucifraga columbiana) on conifer seeds during the postfledgling period. Cana- dian Field Naturalist 97: 208–214. Wiltschko, W., Balda, R. P., Jahnel, M., and Wiltschko, R. (2000). Sun compass orientation in seed- caching corvids: Its role in spatial memory. Animal Cognition 2: 215–221.

Downloaded from http://direct.mit.edu/books/edited-volume/chapter-pdf/677490/9780262268028_c001600.pdf by guest on 29 September 2021