Abstract
Recent studies indicate that stimuli with reward associations are especially effective at capturing attention and at gaining access to visual working memory. However, these effects generally show large individual variability. We investigated whether this stems from individual differences in visual orienting during learning, a well-known characteristic of animals undergoing Pavlovian conditioning (Boakes, 1977). Specifically, some animals (cue-trackers) predominantly orient toward visual cues that precede reward availability, whereas others (goal trackers) predominantly orient toward reward locations (goal box). Although all animals show learning, the cue-trackers produce dopaminergic striatal responses to cues similar to those made in response to actual rewards; goal-trackers show no such cue-induced responses. Perhaps value-based attention capture in humans stems from cue tracking during learning. Mimicking animal studies, we monitored the eye movements of adults learning to associate coloured shapes (cues) with monetary rewards. In blocks of trials, cues and rewards appeared reliably in different locations with each location marked by a checkerboard during inter-trial intervals. Eye movements during this period were used to index cue- versus goal-tracking behaviour. Cues appeared briefly; then participants pressed a key to reveal their reward. After each block, each cue was rated for "luckiness" to provide an explicit learning measure. Key press speed provided an implicit measure of learning. As expected, both explicit and implicit learning in all participants was observed. Cues with the highest expected value produced the highest ratings and the fastest key presses. Although the majority of participants showed consistent fixation biases toward cue versus goal locations during the inter-trial interval, we observed substantial individual differences in cue/goal biases, showing that human variation in orienting during learning may mirror effects found in animals. We anticipate that this novel method of quantifying visual orienting biases during associative learning may usefully predict the strength of value-driven attention capture measured subsequently.
Meeting abstract presented at VSS 2013