LAUSR

Borghi et al.’s ambitious contribution [1] covers a lot of ground: they delineate types of abstract concepts, review theories of word learning, relate abstractness to modality effects in learning and recognition, and most importantly, advance the argument that words are “social tools” that facilitate (or more strongly, enable) the learning of abstract concepts. I agree with Borghi et al. that we need to focus on language for understanding how we learn abstract concepts. But I worry that in framing the Words as Tools theory so firmly in the “embodied and grounded” approach to cognition (to which I am generally sympathetic), the theory risks being overlooked by researchers who do not already subscribe to this approach. Therefore, I would like to highlight the main reason why I think it is so important to look to language to understand abstract concepts, even for someone who does not embrace the embodied view. Briefly, words allow us to chunk sensory inputs in a way that is not possible by relying on (nonlinguistic) perception and action alone. In neural network terms, words are “targets” that supervise category learning.