Abstract
Every choice is guided by the interaction between sensory information and internal goals. Choice behaviour however evolves with the course of actions and changing environments. To assess how sensory information and internal drives interact over time, we employed a free choice saccadic reaction time task. Participants responded to two choice targets that varied randomly in time of relative onset. Selection was faster and more probabilistic when the two options appeared close in time, while large temporal gaps led to slower, predictable selection of the early target. A competitive stochastic decision model investigated the relationship between reaction time distributions in terms of sensory evidence accumulation towards a decision and the prior probability of choice, and captured how target asynchrony influences choice and reaction time. Using a reinforcement learning component, we found that future choices are biased by the way we accumulate and expect sensory information relative to recent choice history; this manifested as competing alternation rate and repetition probability biases respectively. This tradeoff was mediated by the previous decision time and the growing urgency to act, providing reactive and proactive components of bias formation. Using fMRI-guided HD-tDCS, we probed choice history in human right frontal eye field, known to play a key role in the planning and execution of eye movements. While alternating choices were unaffected, repeated choices were more likely to be made to the visual hemifield contralateral to stimulation and under conditions of greater urgency. Thus, our results depict choice bias as an unfolding competition between sensorimotor processing and internal expectations involving the frontal eye field, and cast HD-tDCS as a novel method to target the mechanisms promoting expectation in choice behaviour.
Meeting abstract presented at VSS 2018