A Logical Account of Perception Incorporating Feedback and Expectation

Murray Shanahan

Abstract

This paper presents the theoretical foundations of a vision system in which the effects of high-level reasoning percolate all the way down to influence the low-level processing of raw sensor data. This is achieved through the mechanisms of feedback and expectation. The main contribution of the paper is to present a formal framework, based on the abductive interpretation of sensor data, that incorporates the ideas of feedback and expectation in a way that marries them to logical reasoning. To enable this, two alternative measures of explanatory value are defined.