New Framework Says Brain Builds Categories to Meet the Body's Energy Needs

Every moment, the brain is flooded with sensory signals — light, sound, smell, taste, touch — which it compresses into objects, people, and emotions through a process called categorization. The traditional view treats this as the last step of sensory processing, with the brain passively matching details to templates in memory, like a filing cabinet. But that model cannot explain why a rhythmic patter is heard as a pigeon taking flight on a bright street, yet as a stranger's footsteps in a dark alley.

In a new framework published in Nature Reviews Neuroscience, neuroscientists Lisa Feldman Barrett of Northeastern University and Earl Miller of the Massachusetts Institute of Technology (MIT) argue that the brain reconstructs its categories moment to moment, using not only the senses and memory but the body's physiological and energy needs. Rather than reflecting objective reality, the brain projects categories onto the world to serve survival — its predictions, not accumulated sensory information, shape what is perceived. Categorization is not confined to one brain region but spans the whole nervous system.

The framework merges each researcher's earlier work. Miller studies predictive coding: the brain generates feedback signals predicting what happens next, and these normally dominate incoming, or feedforward, sensory signals; a mismatch creates a prediction error, sometimes felt as surprise. Barrett studies allostasis, how the body predictively regulates its energy use, and holds that emotions such as fear are predictive action plans: elevated heart rate, fast breathing, and tense muscles, in the context of being chased by a dog, are categorized as fear to trigger fight-or-flight behavior — rather than hardwired from birth.

Barrett first approached Miller in 2025 to combine their ideas into a joint paper. Luiz Pessoa, a neuroscientist at the University of Maryland who was not involved in the work, called the framework "a fresh perspective on categorization" and said pursuing energetic constraints as fundamental to categories is "really important to pursue."