Comparative tests of primate cognition: different scaling methods produce different results.

Although early comparative studies supported hypotheses that ecological demands selected for primate cognition, later work indicated that social demands were more important. One difference between earlier and later studies is that earlier studies scaled brain structures by (A) taking residuals from an interspecific regression of the brain structure in question on body mass, whereas later studies scaled them by (B) taking residuals from an interspecific regression of the brain structure in question on another brain structure or by (C) taking ratios of the brain structure in question to another brain structure. We conducted a series of comparative tests to explore the possibility that the different methods are responsible for the discrepancy between earlier and later studies. Specifically, we tested the ability of a social variable - group size - and an ecological variable - home range size - to explain variation in the non-V1 isocortex (isocortex minus primary visual cortex) when this structure was scaled with the three different methods. In multiple regression analysis, group size was a better predictor of the non-V1 isocortex with method (B). With methods (A) and (C), however, results were ambiguous: either home range size or group size explained more of the variation, depending on the inclusion of outliers, the use of independent contrasts, and whether home range size was scaled relative to body mass. We examine the three scaling methods and find no reasonable basis for preferring any of them. Hence, our results do not allow a distinction between social and ecological hypotheses. The general implications of our study are that (1) previous comparative studies are inconclusive and (2) further research is needed to develop a scaling method where relative measures of brain structure size are demonstrated to correspond with behavioral performance.