Abstract
Gaze direction is not only decided voluntarily by the goal or intention but also affected involuntarily by visual salience, novelty and acquired experience with no specific goal and no feedback. Previous monkey study (Kim and Hikosaka, 2013) showed that long-term learned values of visual objects guide saccades in automatic manners. Then, do the long-term learned values also guide saccades of human? If so, which brain regions represent the long-term learned values? To address these questions, 26 human participants learned object-value association for 5 days. Three groups of fractal objects were associated with different monetary values: good (gain), bad (loss), and neutral (neither gain nor loss). Pairs of objects with different values were simultaneously presented, and participants chose higher valued objects to maximize monetary reward. Participants achieved the maximum performance after 4 days, indicating that they gradually learned the value-guided choice behavior. Before the first and 5-day of learning, participants performed a free gazing task in which nine learned objects were simultaneously presented, and participants freely looked at them for 8 seconds without any outcome. After 4 days of learning, the number of first gaze and gaze frequency to learned good objects were increased (+0.8077 times/session; +4.65 times/trial) but the view durations of learned bad and neutral objects were reduced (-45.29ms/trial; -48.85ms/trial). To identify the brain structures guiding involuntary gaze, fMRI data were acquired before and after the learning. We found that subjects showing better performances in free gazing task had greater activation differences (good-bad contrast) in the caudate tail BOLD after long-term learning. This suggests a role of the caudate tail in long-term value memory-based gaze bias. These results indicate that the past learned values guide involuntary gaze in no-goal condition, and the human caudate tail might be involved in the historical value process for the gaze bias.
Meeting abstract presented at VSS 2018