Interaction between Visual Attention and Goal Control for Speeding Up Human Heuristic Search

摘要

Heuristic search is the largest qualitative gap between human performance and computer performance. The dominant study in heuristic search is an empirical way in computer science especially in artificial intelligence. In this paper, we examine the factors that are impact heuristic search from the perspective of human brain. Subjects performed a set of heuristic problems in functional Magnetic Resonance Imaging (fMRI) environment. Then a computational cognitive model is set up to simulate processes of information processing of the heuristics problem solving. During the modeling, we found that two main factors, visual attention and goal control, are responsible for speeding up heuristic search in problem solving, where visual attention captures a target selectively with the goal state-directed control. The interaction between these two cognitive systems speeds up the heuristic search, which is superior to machine intelligence. We demonstrate this conclusion by results of modeling, including cost analysis in time, information processing operations, and fitness of fMRI results and model prediction.