Causal Inference in Sensorimotor Learning and Control

This chapter focuses on the issue of causal inference in perception and action, arguing that ambiguous sensory cues only make sense when the brain understands their causes. It takes a normative view, which focuses on how the nervous system could optimally infer properties of the body or world for perception and sensorimotor control given assumptions about noise in the body and the environment. The normative approach aims to understand why the nervous system works the way it does and not the specific mechanisms that give rise to behavior. Specifically, it asks how the nervous system should estimate the causal relation of events (e.g., errors and movements) and then compare the predictions of these optimal inference models to the way humans actually behave.