Social Hypotheses
Primates are predominantly social animals. Although many prosimians live solitarily, all monkey and apes live in social groups (Jolly 1966; Galdikas 1994). Complex sociality may invoke its own selection pressures, favoring the evolution of social problem solving skills and other social adaptations. What would be the anatomical manifestations of such adaptations? The prefrontal and temporal cortices have been implicated in social interactions, so the degree of development of these areas in different species might reflect different levels of sociality (Sawaguchi 1992). Sociality, however, is a behavioral phenomenon and must be evaluated in behavioral terms. Two likely interconnected ideas have been proposed for investigation: complexity of social behavior and group size (presumably governing quantity of social interactions) (Byrne and Whiten 1988; Harcourt and DeWaal 1992; Cheney and Seyfarth 1990; Dunbar 1992, 1993; Sawaguchi 1988; Clutton-Brock and Harvey 1980).
Social Complexity and Machiavellian Intelligence
The Machiavellian intelligence hypothesis is an extension of speculations, especially of Jolly (1966) and Humphrey (1988), about the almost unique complexity of primate social interactions. In its present formulation, it is similar to a social version of Gibson's extractive foraging idea. Gibson proposed that increased coordination between actions governed by cortical sensorimotor areas in foraging reflected a flexible feeding strategy involving the manipulation of a matrix to acquire the food within (1987). Tool use for the purpose of extracting foods from otherwise inaccessible locations is a striking example of such a foraging strategy. Byrne and Whiten pose the ability to use other individuals as tools, manipulating the social environment in order to meet preconceived goals, is an important factor in the evolution of primate intelligence (1988). Studies of such social manipulation are for the most part confined to single species or groups of related species, in part because of the vagueness of the definition of Machiavellian intelligence itself (Byrne and Whiten 1988; Dunbar 1992). Descriptions generally fall into three subcategories: (1) transmission of novel behaviors (Caro and Hauser 1992 and references therein); (2) deception (Byrne and Whiten 1988 and references therein); and (3) alliance formation (Harcourt and DeWaal 1992 and references therein). The latter two may predicate a knowledge of rank relations between other individuals which is more complicated than simply knowing who is above and below oneself. Moreover, all three may involve altruistic interactions which could vary in kind and complexity, for example, in the time course and the nature of the objects being exchanged. Comparative studies of alliances with respect to cortex size are somewhat more advanced than work in the other two areas, and so only alliance formation will be discussed below.
Alliance formation and maintenance requires an animal to analyze a significant amount of information, including the relations between individuals involved in the alliance as well as their relations to other individuals. Alliances can also operate on different levels of associations, and an individual must be able to weigh and compare the costs and benefits of actions that may differentially affect different levels of alliances (Connor et. al 1992). These nested alliances entail a certain amount of mental sophistication, potentially involving predictions of others' actions before a situation exists. Primates in particular seem to groom their relationships with potential allies before an actual contingency arises (Harcourt 1992). The motivations underlying alliance formation -- and the degree to which intentionality comes into play -- are undoubtedly difficult or even impossible to assess, but it is clear that animals do rely on some base of knowledge about social relationships to guide them. Unfortunately, the paucity of data on alliance formation in non-primates may simply reflect expectations of researchers not to find such social complexity, instead of a real difference between taxa, and so at this stage, conclusions involving comparative social complexity can only be tentative (Harcourt 1992). Connor et. al. (1992) found evidence of nested levels of alliance formation within a population of bottlenose dolphins in Shark Bay, Australia. Because odontocetes, like primates, have relatively large brains, they proposed that there may be a relationship between brain size and social complexity, especially as revealed in alliance formation. Harcourt (1992) hypothesized that primates, more than non-primates, choose their allies based on competitive ability, not necessarily on kin relations as might be supposed. Furthermore, while many animals form alliances against individuals or parties from other groups, primates, more than other taxa, form intragroup alliances which take the form of mutualistic or protective support. These intragroup interactions allow for the manipulation of support -- rank often corresponds to desirability as an ally -- including solicitation, coercion, reciprocation, and friendships (Harcourt 1992). Alliances thereby become not necessarily a means to an end, but rather ends in themselves. However, alliances are competitive in that they do lead to tangible benefits, and so for alliances to be adaptive, these benefits must be contestable. Therefore, alliance formation must not be seen as a strictly social phenomenon; it depends ultimately on the nature of the resources contested, and differences in resource type may explain variation in degrees of alliance formation. Certain types of resources may select for alliance formation, but the degree to which alliances drive encephalization or simply build on preexisting brain structures cannot yet be tested.
Group Size, Social Structure, and the Multiple Factor Hypothesis
If social interactions act as selection pressures, the strength of the pressure probably depends in some way upon the salience of social interactions in an animal's life. An animal that does not interact with others regularly faces a different environment from one surrounded by conspecifics, so social pressures will shape the two differently. Also, the nature of the interactions can be directly related to socio-ecological variables.[13] Whether behavioral changes co-opt anatomical structures or whether social pressures mold ontogeny involves currently untestable causal relationships.
Clutton-Brock and Harvey (1980) found that brain size correlated with home range size in the cercopithecines[14] and that monogamous species have significantly smaller brains than polygynous ones. These two socio-ecological factors are related, for home range varies with troop size, and monogamous species have smaller troops than polygynous ones. Dunbar (1992) claimed that after separating the various related factors, only group size, of a host of behavioral ecology variables, remained important. However, Sawaguchi's work demonstrates that not only do social structure and diet correlate with neocortex size across primates -- separating the primates into three grades -- but also that within each grade selection pressures may be differentially influential (Sawaguchi 1989, 1990, 1992; Sawaguchi and Kudo 1990).
Sawaguchi divided primates depending on their social structure (solitary and troop-making for the prosimians and monogynous and polygynous for the anthropoids), habitat (arboreal, terrestrial), and diet (foliovorous, frugivorous), into congeneric groups in order to eliminate phylogenetic influence.[15] Using indices of `extra' cortical parts (ECIs) (Sawaguchi 1989 after Jerison 1973 and Hofman 1982),[16] relative brain size (RBS) (Sawaguchi 1990 after Clutton-Brock and Harvey 1980), and relative size of the neocortex (RSN) (based on allometry between neocortex volume and brain size) (Sawaguchi and Kudo 1990; Sawaguchi 1992; Dunbar 1993),[17] Sawaguchi examined correlations between brain and neocortex size and social structure and ecology and found that the different cerebral measures gave different results. Only the ECI results and the 1992 RSN results will be discussed here.
Due to inadequate comparative sample sizes, ECI correlations with social structure could not be evaluated for old-world monkeys or for diet and habitat for new-world monkeys. In new-world monkeys, polygynous groups had higher ECIs than monogynous groups, and in old-world monkeys, while terrestrial groups had higher ECIs than arboreal groups, differences in diet were not significant.[18] Furthermore, terrestrial old-world monkeys had larger troop and individual home ranges than arboreal ones. Therefore, based on the extra-cortical indices, Sawaguchi divided the anthropoids (excluding the apes due to a small sample size) into three grades: (1) the old-world terrestrial/ frugivorous/ polygynous monkeys; (2) both old and new world arboreal/ frugivorous/ polygynous monkeys; and (3) the new-world arboreal/ frugivorous/ monogynous monkeys.
RSN, as formulated in Sawaguchi (1992), although potentially confounded by body size effects, is theoretically an appropriate measure of neocortical expansion (Barton 1993). "Neocortical size relative to residual brain size is related to the allocation of brain material to neocortical functions." However, the exact neuroanatomical components reflected by this measure are not yet known (Sawaguchi 1992). With respect to RSN, frugivores displayed higher values than foliovores (contradicting the ECI based finding (1989)), and values for polygynous and monogynous species did not significantly differ. Within the frugivores, however, polygynous species exhibited higher values than monogynous ones. Habitat played no significant role, but troop size was significantly related to RSN.
These results suggest that the factors controlling primate neocortical expansion are not uniform across the order and vary in their strength according to the nature of the animal's environment. This conclusion, that the salience of different factors would depend to a large extent on the interplay between factors in an animal's environment, has too often been overlooked by researchers who tend to lump many primates together without adequately controlling for potential confounds and thus inadvertently test more than one variable at a time, invalidating the results. Sawaguchi (1992) emphasizes this interaction between factors and also delineates the limitations of gross measures of brain function in attributing causal relations stating that intelligence is an amalgamation of different processes:
Both diet and social interactions appear to be associated with the degree of neocortical development in anthropoids. The anthropoid neocortex consists of multiple, parallel circuitries which are involved in multiple parallel functions...Problems arising from foraging may differ from those associated with social interactions, and different neocortical circuitries may be responsible for solving different problems...It is, therefore, likely that multiple, parallel factors associated with diet and social interactions may have been associated with the development of multiple, parallel neocortical circuitries of anthropoids.