By Carel van Schaik
Culture has traditionally been considered a major dividing line between humans and animals. This position has stood largely unchallenged until the 1990s, even though the Japanese zoologist Kinji Imanishi (cited in de Waal 2001) had raised the possibility that animals have the basics of culture in the early 1950s, and Jane Goodall had suggested the same for chimpanzees in the early 1970s (Goodall 1973). How could this have happened?
As so often in scientific debates, the answer is partly a matter of definition and partly one of empirical evidence. Anthropologists cannot really agree on a definition of culture, except that it pervades all our thinking and acting. Yet, the simplest biological definition of culture is that of socially transmitted behavioral innovations, and human culture fits that too, although it tends to involve many more behaviors than those found in animals (we will not discuss here what happened during human evolution to make human culture so much richer than that seen among even our closest-living relatives, the great apes).
If we accept this broad biological definition, we must demonstrate that animals in nature acquire an innovative behavior (i.e., behavior that did not arise routinely in a given environment but was instead invented by someone) by learning it from others who already possess this innovation. This easy-sounding task is actually difficult to accomplish, because it requires demonstrating in the wild that an individual acquired a particular behavioral variant through social learning rather than through some independent process of exploration and learning (Galef 1992). Moreover, experimental studies up until the 1990s did not find that animals were very good at social learning of the more sophisticated variety needed to socially acquire the more complex kinds of innovations: copying the behavior of others through the observation of the motor patterns or their goals. Hence, the null model remained that of the seemingly more parsimonious individual learning.
The beginning of a change in attitude gradually came when great ape researchers began to uncover geographical variation in behavior, and felt they could exclude clear ecological or genetic differences between the populations to account for that variation. Although several experimental approaches are available to evaluate a cultural interpretation of geographic variation (Laland & Hoppitt 2003), these are often impossible for a slew of scientific, logistical, legal and ethical reasons. In response, field workers developed an approach to bolster the cultural interpretation of geographic variation, called the method of elimination or geographic method (Boesch 1996; Whiten et al. 1999; van Schaik 2003). A behavioral variant is considered cultural if it is common wherever it occurs, consistent with its spread and maintenance by social learning, but not closely linked to ecological differences among the areas or genetic differences among the populations. The latter clause is added because this pattern-based approach must eliminate alternative explanations that produce the same spatial pattern but do not involve social learning. An ecological explanation would claim that all individuals exposed to a particular set of habitat features independently converge on the same skill. A genetic explanation would argue that all individuals in a particular region have a strong genetic predisposition to develop the behavior, implying that the behavior is not a clear innovation that must be socially learned. This would generally lead to clear-cut geographic clusters, whose boundaries coincide with subspecies boundaries or long-term dispersal barriers. By concentrating on behavioral variants that do not show clear genetic or ecological correlations in their spatial distribution, researchers could circumvent these problems.
Additional non-experimental methods exist as well that try to capture the dynamics, most commonly intergenerational transfer. Thus, naïve animals tend to selectively attend to their mothers’ behavior when she targets foods that are rare or difficult to process (Tarnaud & Yamagiwa 2008; Jaeggi et al., in rev.). In socially tolerant species, youngsters can also do such peering with other individuals, and then focus on those individuals that show the best mastery of a particular skill, as in chimpanzees (Biro et al. 2003) or capuchin monkeys (Fragaszy et al. 2004). As a result, where adults within a single group show different variants, maturing individuals tend to eventually acquire the variants used by those with whom they associated the most (Perry & Ordoñez Jiménez 2006).
Together, these field-based, non-experimental methods served to establish the plausibility of culture in chimpanzees (Whiten et al. 1999), cetaceans (Rendell & Whitehead 2001), orangutans (van Schaik et al. 2003), and capuchin monkeys (Perry et al. 2003). This interpretation was enhanced by the demonstration of sophisticated and highly reliable observational forms of social learning in apes, especially chimpanzees (Whiten et al. 2005, 2007).
Their major weakness to date is that these field methods cannot be used to estimate the size of the cultural repertoire, which is really what we are after if we are to get an idea of the importance of culture in nature. By design, the geographic exclusion method ignores any behavior that shows a perfect correlation with ecological variables or genetic discontinuities, e.g. inclusion in the diet of a particular food item, even if the animals are critically dependent on social learning for their maintenance (cf. Humle & Matsuzawa 2002). It therefore almost inevitably underestimates cultural repertoires in animals relying on innovation and social learning. Thus, the 40 or so variants described for chimpanzees and the 30 or so described for orangutans are likely to be the tip of the cultural iceberg. But there is no way of knowing for sure until we study systematically what it is that maturing individuals acquire during development through social learning rather than through maturation or independent exploration.
With this caveat in mind, we are now ready to ask about the content of culture in the taxa for which it has been well established.
The content of animal culture
Most known cultural variants are simple because the innovations upon which they are based are simple, e.g. the recognition of a new food or a new predator or a type of tree in which to build a nest, and hence the social learning techniques needed to adopt them from others are simple too. Experimental removal of animals or cross-fostering studies have shown that such simple behaviors are traditional, i.e. acquired by youngsters through social transmission (cf. Galef & Giraldeau 2001), such as the selection of spawning sites of a reef fish (Bluehead wrasse: Warner 1988), of travel routes in other fishes (French grunts: Helfman and Schultz 1984), or of foraging substrates in different species of tits (Slagsvold & Wiebe 2007).
These simple variants, which can be called labels, may be ubiquitous. They consist of simple innovations, and can be learned by the simplest of social learning mechanisms, technically known as enhancement, which involve nothing more than attraction to a place, a stimulus or an individual, upon which the naïve animal itself independently makes the same simple innovation. If the social system allows for horizontal social learning (i.e. toward others in the same age-sex class), they may produce geographic variation, but many of them refer to innovations that are so easily made given the species’ predispositions that they are universal in the species (e.g. the choice of foraging substrates). These universals are simply learned more efficiently when they are learned socially.
Slightly less simple labels can be established and maintained in the same way, but they do produce geographic variation. Sifakas, medium-sized lemurs, have three acoustically different calls that refer to external threats of different kinds (Fichtel & van Schaik 2006; Fichtel & Kappeler, in press). Each sifaka instinctively knows how to produce these calls but has to learn in what contexts and how to respond to them when others use them, which they do by simple social learning, including so-called observational conditioning. However, it turns out that there is (adaptive) geographic variation in their meaning, as expected from consideration of how the system works, depending on which are the most dangerous predators in their region.
Another example concerns orangutan diets. Even in very similar habitats, orangutans may eat diets that differ especially in the choice of so-called fallback foods, i.e. foods that are eaten when preferred foods are scarce, and usually taste bad and are eaten in small quantities. Given that social learning establishes the diet of developing immatures, who then expand it through their own, very limited sampling and watching the food choices of others, the easily evaluated preferred foods (which taste good and are often eaten in large quantities) will end up being the same across sites, whereas the fallback foods show remarkable variation (Bastian et al., in rev.).
Other variants represent skills that rely on more complex innovations, and thus generally require observational forms of social learning for their spread and maintenance. The best-known examples are those involving tool use, which are usually related to subsistence and have a strong effect on fitness, since they provide access to the nutritionally richest resources, such as nuts, honey or seeds (chimpanzees: Boesch & Boesch-Achermann 2000; orangutans: van Schaik 2004). But other skillful forms of subsistence behavior not involving tools may also be cultural, such as how to extract pith from spiny palms (Russon 2003). Yet other skillful variants deal with comfort behaviors, such as whether or not to add a roof to a night nest.
The signals used in social behavior tend to be innate, having evolved their features due to their effect on the recipients. Signal variants, to spread, definitely require observational or vocal learning, and must be understood by others. The exhaustively documented bird song dialects are a prime example (Tracy and Baker 1999), but we see some signal cultures in apes as well. Thus, orangutans make different kiss-squeaks, such as the various forms of kiss-squeaks found in orangutans, but place kiss-squeaks on leaves or other substrates only in some areas and not others. They also make different nesting sounds in different areas, if they make such sounds at all (van Schaik 2004), and more use different calls in different areas to call their infants in time of danger, if they do so at all (Lameira et al. 2008). In all these cases, the variants are customary in a given population.
Symbols are a special class of signal variants, in that their meaning is completely arbitrary and geographically variable. Think of styles of clothing or of words in humans. To date, there are only some tentative examples of those in chimpanzees, such as the leaf clipping display described by Boesch (1996). In some populations, this display serves as, and is understood as, an invitation to play, but in others it is an invitation to sex. Likewise, if the mother-infant calls of orangutans are truly learned, they qualify as symbols, because the relationship between sound and meaning is completely arbitrary (unlike the alarm calls mentioned above, which are therefore considered functionally referential without being truly referential in the human sense of denoting something by arbitrary convention). The same thing would hold for the nesting sounds, if they turn out to have a communicative meaning. Thus, at this moment, it is not clear whether symbols are widespread among non-humans, but further work may show there is no qualitative difference between apes and us in this respect.
Variation in animal cultures
To properly assess the distribution of culture in nature, i.e. which species have which components of culture and how do humans differ from other animals we need to escape from semantic debates that have overshadowed debates about substance. There is probably a lot of meaningful variation in socially transmitted behaviors in nature that should not be ignored, and using one term to capture it all (culture) may make us lose sight of this important task: mapping and understanding this variation.
I therefore propose a new terminology that breaks out of the scala naturaenature of the debate to date (“do animals have [human-like] culture, or don’t they?”), and stresses the strength of three potentially independent dimensions of socially transmitted innovations: (i) the complexity of the innovation; (ii) the complexity of the social learning techniques required to acquire the behavior socially; and (iii) the extent to which there is a geographic imprint.
At this stage, a simple dichotomy on each dimension will suffice to frame the approach. Thus, innovations can be cognitively simple or cognitively complex, depending on whether they could have arisen by chance and trial and error or instead required some form of insight to have been performed the first time (cf. Whiten & van Schaik 2007). In the terminology developed by Strasser et al. (in prep.), innovations can be accidental or goal-directed. Likewise, they could have been acquired by simple forms of social learning that do not require observation of the motor patterns or understanding of the goals of the actions (thus, social facilitation or enhancement, cf. Whiten et al. 2004) or do require the more observational forms, such as motor imitation and goal emulation. Finally, we can divide cultural variants into those that do produce geographic variation with those that do not, either because they are universal and ubiquitous everywhere or are interspersed with other variants at many different sites. Each behavioral variant could fall into any of eight possible combinations, but correlations will reduce the number of states in practice.
As was already noted implicitly, the innovation and social learning dimensions will tend to be correlated, because cognitive abilities limit the complexity of both innovation and social learning, and thus the content and extent of cultural repertoires. The variants that are easily innovated and transmitted through simple mechanisms such as social facilitation or stimulus or local enhancement may be most widespread (and may also be linked to the third dimension, because where innovations arise so easily we may not observe any geographic variation). As the cognitive complexity of the innovation increases, more dedicated mechanisms of social learning are required for social transmission. These make the most interesting cultural variants, because social transmission is likely to be essential for their spread and maintenance, and geographic variation almost inevitably arises. In practice, therefore, the complexity of innovation and that of social learning are tightly correlated, and this correlation may have arisen through correlated evolution of these two abilities (cf. van Schaik & Burkart in press). Thus, we can collapse the dimensions of innovation and of social learning into a single one.
This new dimension can, however, be extended with a third state. Cumulative culture refers to innovations that are beyond the reach of individual inventors (Tomasello 1999) and arose through the step-by-step accumulation of modifications to a variant that improved its function but moved it further and further away from the original innovation. These innovations could be called cumulative innovations, which go beyond complex innovations. One could also make the case that these most complex innovations require more than just observational social learning, but require the active involvement of the role model in the form of intentional teaching (sensu Caro & Hauser 1992). Again, these two steps on the innovation and the social learning scale are likely to be correlated. This leaves us with a three-step scale of complexity of innovation and social learning.
Although socially transmitted behaviors usually produce geographic variation, there are two reasons why they might leave no geographic imprint. The first reason we already encountered. Labels may be so easy to invent that most individuals could do so themselves, and even if all individuals in a given species actually acquire them socially, they may therefore be universal, and thus not recognized as cultural. In other words, some of the behavior patterns in the routine repertoire of a species, found in all populations where the relevant ecological or demographic conditions hold, may turn out to be socially learned and partly maintained by social learning (even if it could easily be reinvented if social learning failed). The reason for this is that maturing individuals of all species may find it easier to copy the behaviors of knowledgeable adults, even if they do so by default by following them around, because that is infinitely more efficient than independent exploration.
A second reason for lack of a geographic imprint is where the species’ social organization is such that there are no clear-cut, geographically separate social units, and social transmission is strictly vertical. Think of dolphins, where infants learn their foraging specializations (innovations!) exclusively from their mothers (Mann & Sargent 2003), but there may be several different specializations within a single locality, and indeed in most localities. Thus, pattern alone will not give us a hint that social transmission maintains these behavioral variants, and detailed field studies are needed to reveal them as cultural. We do not know how common this pattern is because in black bears, where one might have expected a similar system as in dolphins, feeding specializations depend on the habitat they grew up in rather than the mother’s food specializations (Mazur & Seher 2008), suggesting independent exploration as the source.
A preliminary taxonomy of animal culture
These two dimensions, cognitive complexity of innovation and social learning and the extent to which a geographic imprint is present, can be linked to the content-based classification proposed earlier (Figure 1). Labels are simple innovations requiring simple social learning mechanisms that need not produce geographic variation. As we saw, most cases of social transmission in nature of labels may concern the acquisition during development of species-wide foraging patterns or predator recognition, with its main effect one of speeding up their acquisition. Thus, this first box in the phase space in Figure 1 is shaded separately, and may be occupied by a great many species, largely undiscovered until made visible by some experimental manipulation.
Great apes and perhaps some monkeys and cetaceans show the kinds of cultural variants in the second box of Figure 1. Skills are complex innovations that require complex social learning mechanisms and usually produce geographic variation. Signal variants tend to be simple, although they require observational forms of social learning, but almost inevitably involve geographic variation. Symbols can be complex and are geographically variable; they may or may not be common among apes. To date, only humans occupy the area covered in the next box, which also contains cumulative skills (as well as symbols, and those quite amply).
While this scheme is preliminary, it takes the arbitrariness out of the debate on terminology by focusing on the relevant aspects of the whole phenomenon of culture. Many would limit use of the word culture to the third box, albeit for different reasons. Galef (1992) and Tomasello (1999) did so, because at the time they proposed this only humans were known to show imitation and teaching and homology of mechanisms was important to them (now we know that apes can imitate but rarely teach, but several other species are now known to teach). The cultural anthropologists did so to, but because only humans show extensive symbolism in the form of ethnic marking and the institutions built upon them. Other might draw the line somewhere else. Thus, Whiten & van Schaik (2007) propose that species are considered cultural when they show geographically variable socially transmitted innovations in multiple domains. This would grant culture to the occupants of boxes 2 and 3. Yet others might be more generous and grant culture to species in all three boxes. The point of the Figure is that we can now discuss this topic without chauvinism, recognizing the broad spectrum of cultural behaviors in animals and humans.
Despite the recent surge of studies, it is perhaps more remarkable how little evidence we have for culture in nature. Why are there no examples from such well-studied animals such as horses, great tits or sticklebacks? The scheme in Figure 1 suggests that many of the culturally transmitted innovations in those taxa are labels that leave no geographic imprint. A scheme like this is mainly of heuristic value, and if new work necessitates its revision it will have served its purpose.
If, as we suggested here, most species rely on social learning whenever possible, even if this produces little geographically variable behavior, then this may prompt us to develop a new null model. The current null hypothesis is that animals acquire their distinct behavioral repertoires through maturation of genetically anchored developmental programs (instincts) or through individual exploration of environmental affordances (innovation). But note that the parsimony of this position is based entirely on its being formulated before we knew much about social learning in the era that animals were studied in small laboratory cages rather than in nature. Given their reluctance to do much exploring in natural conditions and given their strong preference for social learning, even if it is of a very simple kind, social learning may be the predominant form of learning. Thus, I predict that the parsimonious position will be to assume that many animals learn their labels socially rather than individually, at least those in which parents and offspring associate show the labels of box 1.
Many more labels than expected may also move out of that box. To give one example, when different populations of a species eat a different diet, the reigning paradigm prompts us to assume that each animal independently came up with the optimum diet set through experimentation and estimation of profitabilities, rather than as a result of copying others’ decisions, with a bit of checking of non-eaten food items on the side. However, field data are starting to suggest that the latter is what actually happens in orangutans, so that in a decade or so, the null model for a certain range of species may well have become that animals acquire most of their behavior through at least some form of social learning. I expect that the species for which this prediction should hold the most will be those possessing large brains with large cortical regions developing largely after birth and in interaction with environmental inputs. The coming decade will tell us which kind of learning predominates in nature and in which species and conditions.
Figure 1. The dimensions of culture and their distribution across animals. (V= vertical, O= oblique, H= horizontal, T= transmission)
Carel van Schaik is a Dutch primatologist who is professor and director of the Anthropological Institute and Museum at the University of Zurich in Switzerland since 2004. He is a leading authority on the evolution of culture and cognition in primates. Van Schaik studied biology at the University of Utrecht, graduating in 1979. He was a researcher for the Netherlands Foundation for the Advancement of Tropical Research until 1984 and finished his doctoral dissertation for the Utrecht University in 1985. After positions at Utrecht and at Princeton University, he joined the Department of Biological Anthropology and Anatomy at Duke University in Durham in 1989, initially as Associate Professor and later as Professor, besides holding adjunct positions in Zoology and the School of the Environment. In 2004 he moved to the University of Zurich.
Van Schaik has studied wild primates in all continents since 1976, albeit increasingly with a strong focus on Indonesia. His work on orangutan socioecology and conservation biology has underscored the existence of complex signaling systems which have evolved with respect to particular ecologies and social networks and the semiotic dynamics of culture. Van Schaik’s revolutionary research shows how the making and the use of tools, sounds, gestures and movements that demand complex calculations are transmitted socially, thus creating cultural assemblages with great variations when populations are isolated from one another by distance or physical barriers to movement.
Professor van Schaik’s acclaimedAmong Orangutans: Red Apes and the Rise of Human Culture (2004, Belknap/ Harvard University Press) was reviewed in SemiotiX 10. For a list of van Schaik’s numerous publications, see his homepage.
Anthropological Institute and Museum, University of Zurich, Zurich, Switzerland; E-mail: email@example.com
Bastian ML, Zweifel N, Vogel ER, Wich SA, van Schaik CP (in review) Geographic variation in orangutan diets: evidence for social learning
Biro D, Inoue-Nakamura N, Tonooka R, Yamakoshi G, Sousa G, Matsuzawa T (2003) Cultural innovation and transmission of tool use in wild chimpanzees: evidence from field experiments. Anim Cogn 6: 213-223.
Boesch C (1991) Teaching among wild chimpanzees. Anim. Behav. 41:530-532
Boesch C (1996) Three approaches for assessing chimpanzee culture. In: Russon AE, Bard KA, Parker ST (eds) Reaching into Thought: the minds of the great apes. Cambridge University Press, Cambridge, pp 404-429
Boesch C, Boesch-Achermann H (2000) The Chimpanzees of the Taï Forest: Behavioural Ecology and Evolution. Oxford, Oxford University Press
Caro TM, Hauser MD (1992) Is there teaching in nonhuman animals? Q Rev Biol 67:151-174
de Waal FBM (2001) The Ape and the Sushi Master: Cultural Reflections by a Primatologist. Basic Books, New York
Fichtel C, van Schaik CP (2006) Semantic differences in sifaka (Propithecus verreauxi) alarm calls: a reflection of genetic or cultural variants? Ethology 112: 839-849
Fragaszy DM, Perry S (2003) Towards a biology of traditions. In Fragaszy DM, Perry S (eds) The Biology of Traditions: Models and Evidence. Cambridge University Press, Cambridge, pp 1-32
Fragaszy DM, Izar P, Visalberghi E, Ottoni EB, Gomes de Oliveira M (2004) Wild capuchin monkeys (Cebus libidinosus) use anvils and stone pounding tools. Amer J Primatol 64: 359-366
Galef BG Jr. (1992) The question of animal culture. Human Nature 3:157-178
Galef BG Jr, Giraldeau L-A (2001) Social influences on foraging in vertebrates: causal mechanisms and adaptive functions. Anim Behav 61:3-15
Goodall J (1973) Cultural elements in a chimpanzee community. In E. W. Menzel (ed) Precultural Primate Behavior. Basel, Karger, pp 144-184.
Helfman GS, Schultz ET (1984) Social transmission of behavioural traditions in a coral reef fish. Anim Behav 32: 379-384
Humle T, Matsuzawa T (2002) Ant-dipping among chimpanzees of Bossou, Guinea, and some comparisons with other sites. Am J Primatol 58:133-148
Jaeggi AV, Dunkel LP, van Noordwijk MA, Wich SA, Sura AAL, van Schaik CP (in review) Social learning of diet and foraging skills among wild orang-utans (Pongo pygmaeus wurmbii): implications for culture.
Laland KN, Hoppitt W (2003) Do animals have culture? Evol Anthro 12:150-159
Lameira AR, Hardus ME, Naso A, van Noordwijk MA, van Schaik CP, Sterck, EHM, Wich SA (in review) Call cultures in orangutans
Mann J, Sargent B (2003) Like mother, like calf: the ontogeny of foraging traditions in wild Indian Ocean bottlenose dolphins (Tursiops sp.). In Fragaszy DM and Perry S (eds) The Biology of Traditions: Models and Evidence. Cambridge, Cambridge University Press, pp 236-266.
Mazur R, Seher V (2008) Socially learned foraging behaviour in wild black bears, Ursus americanus. Anim Behav 75: 1503-1508
Perry S, Panger M, Rose LM, Baker M, Gros-Louis J, Jack K, MacKinnon KC, Manson J, Fedigan L, Pyle K (2003) Traditions in wild white-faced capuchin monkeys. In: Fragaszy DM, Perry S (eds) The Biology of Traditions: Models and Evidence. Cambridge, Cambridge University Press, pp 391-425
Perry S, Ordoñez Jiménez, JC (2006) The effects of food size, rarity, and processing complexity on white-faced capuchins’ visual attention to foraging conspecifics. In Hohmann G, Robbins MM, Boesch C (eds) Feeding Ecology in Apes and Other Primates: Ecological, Physical and Behavioral Aspects.Cambridge, Cambridge University Press, pp. 201-232
Rendell L, Whitehead H (2001) Culture in whales and dolphins. Behav. Brain Sci. 24:309-382
Russon AE. 2003. Innovation and creativity in forest-living rehabilitant orang-utans. In Reader SM, Laland KN (eds) Animal Innovation. Oxford, Oxford University Press, pp 279-306
Slagsvold T, Wiebe K (2007) Learning the ecological niche. Proc Roy Soc B 274: 19-23
Tarnaud L, Yamagiwa J (2008) Age-dependent patterns of intensive observation on elders by free-ranging Japanese macaques (Macaca fuscata yakui) within foraging context on Yakushima. Am J Primatol 70: 1-11
Tomasello M (1999) The Cultural Origins of Human Cognition. Harvard University Press, Cambridge, MA
Tracy TT, Baker MC. 1999. Geographic variation in syllables of house finch songs. Auk 116: 666-676
van Schaik CP (2003) Local traditions in orangutans and chimpanzees: social learning and social tolerance. In: Fragaszy DM, Perry S (eds) The Biology of Traditions: Models and Evidence. Cambridge, Cambridge University Press, pp 297-328
van Schaik CP (2004) Among Orangutans: Red Apes and the Rise of Human Culture. Harvard University Press (Belknap), Cambridge, MA
van Schaik CP, Ancrenaz M, Borgen G, Galdikas B, Knott CD, Singleton I, Suzuki A, Utami SS, Merrill MY (2003) Orangutan cultures and the evolution of material culture. Science 299:102-105
van Schaik CP, Burkart JM (in press) Why culture makes smart: the cultural intelligence hypothesis. Anim Cogn
Warner RR (1988) Traditionality of mating-site preferences in a coral reef fish. Nature 335: 719-721
Whiten A, Goodall J, McGrew WC, Nishida T, Reynolds V, Sugiyama Y, Tutin CEG, Wrangham RW, Boesch C (1999) Cultures in chimpanzees. Nature 399: 682-685
Whiten A, Horner V, Litchfield CA, Marshall-Pescini S (2004) How do apes ape? Learning & Behavior 32: 36-52
Whiten A, Horner V, de Waal FBM (2005) Conformity to cultural norms of tool use in chimpanzees. Nature 437: 737-740
Whiten A, Spiteri A, Horner V, Bonnie KE, Lambeth SP, Shapiro SJ, de Waal FBM (2007) Transmission of multiple traditions within and between chimpanzee groups. Current Biology 17, 1-6
Whiten A, van Schaik CP (2007) The evolution of animal ‘cultures’ and social intelligence. Phil Trans Roy Soc 362: 603–620