Abstract
Uncertainty brings people not only the joys but also the pain of life. It is therefore fundamentally important for we humans to process uncertain information accurately and appropriately from daily investments to long-term return forecasts. Though risk processing during value-based decisions has been widely studied, whether the expected reward and reward risks could be processed automatically in a non-choice situation is still under-explored. Using a modified monetary incentive delay task (MIDT) in which mean reward and reward variance were parametrically manipulated and orthogonalized, we examined the behavioral and attentional coding of expected reward and risks at a more implicit level with strategical use of behavioral response time (RT) and simultaneous recording of eye movements. Behaviorally, we showed that RTs (n=90) varied as a function of mean reward but not reward variance, suggesting that reward magnitude is the main driver of response vigor. Moreover, fixation difference (n=68) between high and low value (total fixation duration/counts/ percentage and first fixation latency) was reduced with larger mean reward and smaller reward variance, providing the first attentional evidence coding for mean reward and reward variance automatically with similar visual patterns. Altogether, we provide behavioral and attentional evidence coding for expected utility and risks both similarly and differentially, irrespectively of whether the response is decision-related or not. Our findings may help to explain or predict pathological gambling or addictive behaviors based on atypical processing of reward and risks, i.e., augmented (reduced) weight of reward (or risks) during value-based decision-making. Our study may provide insights into behavioral intervention via attentional manipulation during value-based risky decisions.