Abstract
The study of numerical abilities in non-human species speaks to three major issues: 1) the evolutionary origin of mathematical knowledge; 2) the extent to which such knowledge represents a modularized cognitive domain, and the limits of this module in the absence of language; and 3) the validity of various models of nonlinguistic numerical representation. Current data suggest that non-human animals possess considerable numerical abilities in the absence of language, and the majority of these results appear consistent with the Accumulator Model of numerical representation. However, these data come predominantly from paradigms that have involved extensive training of subjects. More recent evidence suggests that spontaneous, untrained numerical abilities may be more consistent with the Object File Model. The experiments presented in this thesis address these issues by further examining the spontaneous numerical abilities of free-ranging rhesus macaques with methods that require no training of subjects and mimic natural foraging problems. These experiments show that the Accumulator Model may not be consistent with spontaneous mathematical abilities for small numbers because it is less able to account for the ability of rhesus to represent subtraction events and the quantity zero. The Object File Model better accounts for these abilities, as well as the discovery of a Recency effect in certain tasks, which may be due to the process of updating numerical information. These data also address the general issues of the independence of mathematical abilities from language as well as their evolutionary origins.