The perception-cognition-action fallacy

A longstanding idea in Western science is the supposition that the behavior of an agent, like an animal or a person, can be broken down into three distinct and largely non-overlapping processes: perception, cognition, and action. In this view, perception is thought to be essentially passive: light waves and sound waves hit your eyes and ears, and all that your sense organs do is simply play the resulting movie for your inner homunculus to watch. Cognition is then the process of taking the sense data and thinking about what to do based on it. Once a decision is made, action is the passive result, the body an inert puppet moving according to the will of the brain. All three of these processes can happen without each other and can be studied and analyzed in isolation without considering the cognitive properties of perception or the perceptive properties of action.

There are lots of ways to massage the lines between perception, cognition, and action into something blurrier. Perception involves lots of chemical reactions; the sense organs do not passively receiving incoming signals but instead transform them in non-trivial ways. Many experiments show the role that cognitive properties like beliefs, attention, and memory play on sensory perception, and phenomena like the rabbit-duck illusion show that what we perceive is a function of how we choose to look at something. Cognition is something we only detect indirectly through the perception and action of the cognitive entity in question, and cognition consists of the perceptions and actions of the constituent parts of the collective intelligence. And action can be seen as an expression of perception and cognition; action is the externally visible output of the cognitive-perceptual transformation of the agent’s internal model.

In active inference, perception and action are two sides of the same coin: perception is the process of using externally originating signals to update the internal model, and actions are taken to generate new perceptual data to test and update the internal model. Cognition is the updates of the model, updates that themselves control and result in perceptions and actions. There is no “first perception, then cognition, and finally action”. It’s all one continuous process. I call the standard belief that perception, cognition, and action are separate and distinct processes that proceed in stages from perception to cognition to action the perception-cognition-action fallacy.

An example of the scientific missteps created by the perception-cognition-action fallacy comes from neuroscience. The belief that perception, cognition, and action are separate and either unrelated or very simplistically related processes led neuroscientists to believe that the brain contains distinct and architectural structures for handling perception, cognition, and action, much like how a kitchen contains distinct structures for producing heat (stove), water (sink), and cold (refrigerator). So neuroscientists theorized that there is a special part of the brain for perception, a special part for motor behavior, a special part for higher reasoning, etc. New evidence, however, shows that behavior is a whole-brain phenomenon not localized to specialized, premade structures. As evidence continues to pile up and wear away at the perception-cognition-action fallacy, new theories about the brain and how perception, cognition, and action are constructed will emerge.