Proceedings of the Annual Meeting of the Cognitive Science Society

An ongoing debate about the structure of conceptual space is based on two competing mathematical theories of similarity that make distinct predictions about the structure of mental representations and how to model the representational space they are stored in. These are known as metric (Shepard, 1962) and ultrametric (Tversky, 1977) theories, modeled by multidimensional scaling and additive trees respectively. Turning to the brain to resolve this conflict, we propose a computational framework to assess behavioral and neural data’s underlying structure and investigate whether the behaviorally known spaces for colors (metric) and letters (ultrametric) can be reproduced from neural data. Our results show that the metric color wheel can be reproduced from brain area V4, but that neural activations of the letters from extrastriate cortex (V2-V5) are also metric instead of being ultrametric. Finally, we discuss three possibilities for the brain’s similarity structure, including a potential metric bias.