Neural representations: A Mechanistic Challenge to Suárez's Inferential Conception

This article examines Mauricio Suárez's inferentialist account of scientific representation in light of recent advances in neuroscience and artificial intelligence (NeuroAI). While it offers valuable pragmatic insights, I argue it is insufficient to capture the dynamic, computational, and biological nature of neural representations. Drawing on the mechanistic, functionalist, and representationalist (MFR) approach and empirical findings, I maintain they are not mere abstract entities but are embodied in the physical and functional properties of neural systems. I challenge arguments against the necessity and sufficiency of similarity and isomorphism, highlighting computational transformations, functional roles, and the directionality of processing that shape content. The Hodgkin–Huxley model and neurocomputationalism employing artificial neural networks support the MFR and challenge Suárez's inferentialism. I conclude that a mechanistic and computational understanding provides a more comprehensive and empirically grounded framework for modelling and representation in the mind sciences.