Cognitive Scientists Propose Predictive Brain Model for Categorization in Nature Review

Cognitive scientists Lisa Feldman Barrett and Earl K. Miller challenge traditional neuroscience models in a new review article published in Nature Reviews Neuroscience. They argue that categorization is not a late-stage cognitive process but a fundamental predictive function operating throughout sensory processing. The brain constructs categories dynamically based on prior experience to anticipate bodily needs and motor plans, rather than comparing sensory inputs to fixed prototypes. Anatomical evidence shows cortical architecture favors feedback connections over feedforward ones, with up to 90% of visual cortex synapses dedicated to memory-based filtering. Functional studies reveal beta frequency waves carrying goal information constrain gamma waves carrying sensory data. This framework redefines categorization as a signal processing event that renders sensory inputs meaningful in terms of predicted action plans. The model has implications for understanding disorders like depression and autism, where categorization processes may become overly broad or insufficiently compressed. Funding for the research came from multiple sources including the National Institutes of Health, U.S. Army Research Institute, Office of Naval Research, and several foundations.