Water. Cat. Pencil. Fear. Reward. Categories that exist inside the signal processing structure we call the human brain can range from concrete and tangible to amorphous and abstract. But they all serve the practical purpose, from an evolutionary point of view, of increasing our metabolic efficiency and thus evolutionary fitness. From a neurologist’s point of view, they can be described as part of the brain’s signal compression process. Lisa Feldman Barrett and Earl K. Miller are the authors of a recent paper published in the journal Nature Reviews Neuroscience which hypothesizes, based on decades of studies, that categorization is an even more fundamental part of our thought processes than traditional neuroscience has thought.
Scientists have long known that categorization is done by all animals; categorization allows us to combine incoming sensory signals from various receptors on our body into more manageable groups or packets. Our brains receive a constant influx of information from all kinds of sensory surfaces such as our eyes, nose, and even things like glucose monitors in our intestines. All this information creates signals with high dimensionality, meaning that they are full of many specific details — so many details that trying to process each one individually, every second of every day, would be an impossible task. Or, in the language of evolutionary fitness, it would simply incur way too much of a metabolic cost. By the time we processed all those details, we would be eaten by a lion. This is where signal compression comes in: These high-dimensional, granular signals are constantly transformed by our brain’s neuronal pathways into lower-dimensional, more abstract signals. They can be thought of as our brain’s way of looking at the forest instead of the trees. Traditionally, these lower-dimensional signals have been referred to as the static categories that are reached at the end of signal compression. For example, the signals providing information like purring, furry, tabby, and soft can be compressed into a single category: cat.
Barrett and Miller, however, think about categorization in a slightly different way. A category, they claim, is not just a thing that the brain reaches after signal compression, but rather a part of a dynamical process involving both feedback and feedforward that helps the brain to make predictions during signal compression. Feedback signals are the predictive ones traveling towards the sensory neurons, sending them information about what may come next, and feedforward signals are the incoming sensory signals being sent towards the decision-making parts of the brain (limbic edge), taking many little details of information about the world and compressing them. Feedforward signals continually correct prediction errors from the feedback, interacting in real time as the information is processed.
These predictions are part of a state called allostasis, in which the brain anticipates energy needs before they arise, preparing for them in order to boost metabolic efficiency. An example of this would be the brain preparing an organism for action by increasing heart rate, blood pressure, and glucose utilization when it has reason to anticipate a period of stress or uncertainty. Allostasis is different from homeostasis because the former is predictive, while the latter is reactive. Allostasis is our brain’s way of staying ahead of the game.
One part of this, compressing signals, allows the brain to generalize across both space and time, to learn from past events and context clues and predict what is likely to happen. Using a prototype or algorithm to predict likely future signal inputs can minimize the cost (in this case, metabolic cost — energy resources) of signal processing. This is where categorization comes in. It comes in, however, as a trial-and-error process of creating similarities from differences, because the world is not presented to us as a pre-sorted group of categories. We, humans, have to constantly deal with an overwhelming noise of information (signals) screaming at us from every direction. Our brain’s job is to sort that information into useful groups — and what makes useful groups is different for everyone, varying with past experiences and cultural background. Interestingly, as Barrett and Miller note at the end of their paper, some neuropsychiatric disorders may be understood as differences in the complexity and organization of people’s categories — a highly interesting topic for future research that could enhance our understanding of neurodivergence and depressive disorders.
