DSCN4361

articl.es Evolutionary Anthropology 181

For this reason, Jerison derived his encephalization ąuotienL Ali subse-ąuent studies have used body size as the appropriate baseline against which to measure relative deviations in brain size. However, a problem has sińce emerged: Brain size is determined early in development and, compared to many other body Systems, appears to be highly conservatlve in evolution-ary terms. As a result, body size can often change dramat! cally both ontoge-netically across populations in re-sponse to local environmental condi-tions²¹ and phylogenetically²²*²³ without corresponding changes in brain size. This is particularly con-spicuous in the case of phyletic dwarfs (e.g., callitrichids and perhaps modem humans and hylobatids²²) and spe-cies in which body size may have increased in response to predation pressure following the occupation of morę o pen terrestrial habitats (e.g., papionids²⁴).

The lability of body size therefore makes it a poor baseline, though one that probably is adeąuate for analyses on the mouse-elephant scalę. Conse-ąuently, it is necessary to find an inter-nally morę consistent baseline for taxo-nomically fine-grained analyses. Willner²² suggested that either molar tooth size or brain size may be suit-able because both are developmen-tally conservative. Because we are eon-cemed with brain part size, some aspect of brain size seems the most appropriate.

At this point, three options are avail-able. One is to compare the neocortex, the brain part of interest, with the whole brain; the second is to use the rest of the brain other than the part of interest; the third is to use some less variable primitive component of the brain, such as the medulla, as a baseline. Two options are in turn available as mechanisms for controlling for brain size in each of these cases. One is to use residuals from a common regression linę against the baseline (e.g., the residual of neocortex volume on total brain volume or medula vol-ume). The other choice is to use ratios.

We have considered and tested all these options¹⁰*²⁴ (see Box 1). The re-sults are virtually identical irrespec-tive of which measure is used. One explanation for this may be that all these measures actually index the same thing, absolute neocortex size, mainly because the neocortex is such a large component of the primate brain. In-deed, the use of absolute neocortex size produces results that are simllar to those obtained from relativized indi-ces of neocortex volume.^24,25 This makes some sense in computational terms: As Byme²⁶ has pointed out, a 10% inerease in the processing capac-ity of a smali Computer is worth a great deal less in information-process-ing terms than is a 10% inerease in a large Computer. Although residuals from a common regression linę would conventionally be considered the saf-est measure, and have been used in many recent analyses,²⁷-²⁸ I shall con-

The neocortex is generally regarded as being the seat of those cognitive processes that we associate with reasoning and consciousness, and therefore may be expected to be under the most intense selection from the need to inerease or improve the effectiveness of these processes.

tinue to use my original ratio index because it provides the best predictor (see Box 1).

Finally, it is now widely appreciated that comparative analyses need to con-trol for the effects of phylogenetic inertia. Closely related species can be expected to have similar values for many anatomical and behavioral di-mensions merely by virtue of having inherited them from a recent common ancestor. In such cases, plotting raw data would result in pseudoreplica-tion, artificially inflating the sample size by assuming that closely related species are actually independent evolu-tionary events. The ways of dealing with this problem include plotting means for higher taxonomic units, performing nested analyses of varlance using phylogenetic leveis as factors, comparing matched palrs of species, and making independent contrasts that control directly for phylogeny. Each method has its own advantages and disadvantages, but the flrst and third procedures are particularly associated witii loss of Information and smali sample sizes. I shall use the flrst and last method, the last because it allows individual species to be compared. but the flrst because it allows grade shifts within data sets to be identified (a problem that independent-contrasts methods have difflculty dealing with). I shall take the genus as a suitable basis for analysis because genera typi-cally represent different reproductive or ecological radiatlons and thus are morę likely to constitute independent evolutionary events.

The resulting analyses are relatively straightforward: Figurę 2 presents the data for neocortex ratio for the anthro-poid primate species in the data base of Stephan, Frahm, and Baron.²⁹ Neo-cortex size, however measured, does not correlate with any index of the ecological hypotheses, but does correlate with social group size. Similar flndings were reported by Sawaguchi and Kudo,³⁰ who found that neocortex size correlated with mating system in primates. Barton and Purvis³¹ have confirmed that using both residuals of neocortex volume on total brain vol-ume and the method of independent contrasts yields the same result. Both Barton¹⁰ and T. Joffe (unpublished) have repeated the analyses using the medulla as the baseline for compari-son. Morę importantly, Barton and Purvis³¹ have shown that while rela-tive neocortex volume correlates with group size but not the size of the ranging area, the reverse is true of relative hippocampus size. A correla-tion between rangę area and hippocampus size is to be expected because of hippocampal involvement in spatial memory.³²-³³ This correlation demon-strates that it is not simply total brain size that is important (a potential problem, given the overwhelming size of