From similarity to inference
Abstract
We advance a theory of inductive reasoning based on similarity, and test it on arguments involving mammal categories. To measure similarity, we quantified the overlap of neural activation in left Brodmann area 37 (lBA37) in response to pictures of different categories; the choice of lBA37 is motivated by previous literature. The theory was tested against estimated probability judgments for 160 arguments generated from 16 categories and a common predicate. The theory’s predictions (based on neural similarity) correlate strongly with these estimates. Other brain regions in the ventral visual pathway yield similarities that also allow the model to predict inductive judgments accurately whereas use of rated similarity in place of neural similarity is less successful. We conclude by extending the theory to aspects of causal reasoning and considering the relation between similarity and feature ratings of categories.