A Suboptimal Model Explained the Spatiotemporal Control of Human Eye Movements in Visual Search

Efficient eye movement planning in both space and time is vital for visual perception. Previous studies showed that human eye movements were consistent with an optimal searcher during visual search; however, we found that such a model could not fully explain the spatial control of eye movements and it completely ignored the temporal control aspect. Here, we first measured the temporal course of human visibility map, and then measured eye movement dynamics during a visual search task. We further built a continuous-time eye movement model that determined both fixation duration and location. When the participants’ saccade amplitude bias and memory capacity limitations were introduced, this model could describe multiple aspects of human visual search behavior much better than the optimal searcher with fixed fixation duration. Interestingly, this model predicted that the human visual memory capacity is approximately 8 fixations. The results showed that human performed suboptimal in visual search, and implied that human may use a mixed control strategy to select fixation location. This strategy may have adaptive advantages considering the limitations of the human brain and statistics of the natural environment.