There are several key issues that underlie determinations of relative brain size in primate evolution. The first is allometric correction—long recognized as necessary due to general patterns of negative allometry or progressive relative diminution of brain size compared to body size in mammals and other vertebrates. Some of the earliest studies of allometry dealt with ontogenetic, interspecific, and phylogenetic scaling of brain size (see Gould, 1966: Huxley, 1932: for references). Any number of excellent review papers may be consulted for discussion of various empirical and theoretical issues involved in recognizing and explaining general patterns of brain-body scaling, as well as utilizing these scaling baselines for computation of individual species' encephalization quotients (EQ) or other such residualized determinations (e.g., Gould, 1966, 1975a; Jerison, 1979; Martin, 1989). Here I follow traditional broad interspecific scaling analyses, and the use of these baselines to determine residuals for individual specimens or species. These approaches work reasonably well for addressing general questions, such as whether extant primates tend to cluster above the size-corrected norm for all mammals in static comparisons of extant or fossil species (Figure 1). However, it is vital to keep in mind that such static patterns mask true phylogenetic contexts and comparisons. Evolution consists of transformations of antecedent states, and only hypothetical ancestral values can provide reliable baselines for such assessments—no species ever evolved from the predicted value of a regression line best-fit to a static scatter of diverse species! Moreover, the expectation of
log Body weight
Figure 1. A schematized plot of brain-body scaling in an interspecific series of mammals. Relative brain size (EQ) is traditionally determined by comparing observed values (here for species X and Z as representative examples) against regression-adjusted predicted values for species of that body size, defined as the solid line with arrowhead. These residualized values are shown here as arrows from the predicted values for species X (negative residual, smaller-than-average brain size) and species Z (positive residual, relatively large brain size).
Residuals may also be determined within a more phylogenetically controlled series, either by using genus-level regressions or fitting a line of slope 0.33 through a smaller sister species (see text and Lande, 1979). Three cases of the latter are illustrated here, with the dashed lines fit to the selected values as shown. The species indicated by the "a," "b," and "c" are here taken as phyletically-enlarged descendants of these three species. The disparities between residualized values determined relative to these intra-generic lines, as opposed to the overall line of best fit (solid line), are significant. For species a, encephalization has decreased relative to the expectations based on the intrageneric fit, but the species still has a positive residual value relative to the overall trend. For species b, the opposite pattern holds, i.e., an increase in relative brain size is observed in comparison to the intrageneric prediction, but species b still has a negative residual relative to the broad sample regression. In the case of species c, the observed brain size is exactly as predicted given intrageneric allometric scaling, but the ancestral species has a positive residual and the descendant species a negative residual relative to the line of overall best-fit. The key point here is that the EQ residual values in these cases are coded quite differently, and thus would pattern differently in comparative studies associating relative brain size with factors such as variance in diet, habitat, social structure, etc. See text and Shea (1983), and Williams (2002) for additional discussion.
brain size change correlated with short-term selection on body size yields a different, and substantially lower, scaling exponent than the static broad interspecific value of 0.66-0.75 (Gould, 1975a; Lande, 1979, 1985; Riska and Atchley, 1985; Shea, 1983, 1987, 2005; Williams, 2002). I have previously argued that such lowered scaling exponents result because the body size diversification is generated by differential selection on postnatal growth rates, which exhibit reduced correlations with early (prenatal) growth periods, when brain size is increasing differentially (Shea, 1983, 1992a,b). This principle is also illustrated in Figure 1, where lower-than-average residualized values relative to the all-mammal plot may actually be coincident with a derived increase in encephalization, once the proper baseline is recognized. Other complex and counterintuitive arrangements of the standard EQ values are also possible (see Figure 1 and accompanying legend).
An important issue related to the use of interspecific scaling baselines for calculation of EQ residuals is the specific slope value of the broad interspecific trend. Jerison's (1973, 1979) EQ values utilized slopes of 0.66, originally derived from empirical scatters of extant mammal, and subsequently fit to the grand means of various brain-body assemblages. Following Jerison's early study, various researchers (e.g., Armstrong, 1983; Eisenberg, 1981; Hofman, 1983; Martin, 1981) concluded that a slope of 0.75 provided a better fit to the extant mammal scatter. This will of course alter the calculation of residuals, such that smaller species will on average exhibit higher EQ values in the new calculation, as the line of best-fit through the grand mean is repositioned lower within the scatter at small values of body size, and larger species will on average exhibit lower EQ values as the line is repositioned higher within the scatter at large values of body size. Nevertheless, I utilized Gould's (1975a), Jerison's (1973, 1979), and Radinsky's (1970, 1975) original values here for several reasons. First, these are relative assessments, and thus the ordering and comparability is robust to such changes in the overall slope value. In this chapter, I am not focused on either the absolute values of these EQ's, or the underlying slope from which they are derived. In any case, the many discussions of the "new" (circum 0.75) versus "old" (circum 0.66) slope values are in all probability off the mark, since the best-sampled scatters evidence considerable curvilinear-ity (see Martin, 1981, Figure 1), and the overall slope is unlikely to be linked to any single underlying factor, be it metabolic rate or some other input. A convincing criterion of "functional equivalence" based on neural capacities or brain functions has never been advanced for the 0.66 or 0.75 coefficients.
In fact, arguments linking measured cognitive performance to size-corrected EQ values in anthropoids and hominoids of divergent body size support a criterion of subtraction value in the 0.2-0.4 range (Williams, 2002). Much additional work will be needed to address these complex issues in the comparative study of brain size and scaling.
Jerison's (1973, 1979) computation of EQ in living and fossil mammals represented landmark advances in our understanding of the evolution of gross brain size. However, various other difficulties with these and other such studies must be acknowledged. One is the question of the biological meaning of total brain size and the myriad empirical and theoretical issues related to gross brain size, the brain's internal allometries, and the evolvability of its components. Here I follow Jerison (1979) in acknowledging that gross size of the brain and/or its components serve only as surrogate measures and correlates of more salient aspects of organismal performance and fitness (Arnold, 1983). While brain size scaling regularities—externally with body size and internally among its components—are indisputable, there is obviously also considerable residual variance attesting to the evolvability of total brain size and its localized regions. Much of this variance is related to particular sensory and cognitive specializations of various mammalian species.
Was this article helpful?