Tinkering with minds from the past

Cognitive scientists argue that in the absence of language, non-human animal conceptual representations are either impoverished or completely absent. One might make comparable claims about human infants, who enter the world with different conceptual representations from adults. Nonetheless, we often treat human infants like miniature adults. This is a mistake. I argue that research on human cognition, and in particular its domain specific knowledge systems, can only succeed if it adopts a comparative perspective. To carry out this agenda, however, we require methods that can be used across species. Focusing on how numerical abilities evolved, I describe experiments on two non-human primates, representing different phylogenetic branches. Our experimental procedure--the preferential looking time technique--was designed to assess what prelinguistic human infants know about the physical and psychological world, but it is ideal for non-human animals, especially when one wishes to explore spontaneous cognitive capacities in the absence of training. Results reveal that up to a certain age, human infants and non-human primates are indistinguishable in terms of numerical competence. We must now focus on how language, together with other cognitive facilities, bring the human child to a level of numerical sophistication that exceeds non-human animals, and why non-human animal capacities stop where they do.