Bayesian and "Anti-Bayesian" Biases in Sensory Integration for Action and Perception in the Size-Weight Illusion

摘要: Which is heavier: a pound of lead or feathers? This classic trick question belies simple but surprising truth: when lifted, the feels heavier—a phenomenon known as size–weight illusion. To estimate weight an object, our CNS combines two imperfect sources information: prior expectation, based on object's appearance, and direct sensory information from lifting it. Bayes' theorem (or law) defines statistically optimal way to combine multiple for maximally accurate estimation. Here we asked whether mechanisms combining these produce estimates both perceptions actions. We first studied ability subjects hold one hand steady other removed object it, under conditions in which about sometimes conflicted with expectations its size. Since supporting depends generation motor command that accounts lift timing weight, motion can be used gauge biases estimation by system. found system reflected integration real-time proprioceptive Bayesian, fashion discounted unexpected information. produces illusion consistently toward expectations. In contrast, compared weights objects, their defied law, exaggerating value perceptual away term this effect “anti-Bayesian” because bias opposite seen Bayesian integration. Our findings suggest fundamentally different strategies coexist nervous