Abstract
Stimuli associated with aversive outcomes such as shock and monetary loss automatically capture attention. In many common experimental paradigms, aversive outcomes can be either avoided with fast and accurate responses or are unrelated to behavior and probabilistic, meaning that the stimuli associated with such outcomes are also associated with the occasional withholding of an anticipated aversive event. It therefore remains unclear whether the resulting attentional biases are driven by learning from trials on which aversive outcomes are delivered (punishment learning) or withheld (negative reinforcement), and by extension what mechanism of learning is responsible shaping the attention system. In the present study, we provide two sources of converging evidence demonstrating that learning from punishment dominates over learning from negative reinforcement when the two sources of learning compete against each other. First, in an antisaccade task, one target color predicted a possible shock, which could be averted by a fast and accurate eye movement away from the stimulus. Thus, rapid eye movements away from the stimulus were negatively reinforced. A subsequent test phase involving prosaccades to a shape-defined target provided strong evidence for a bias towards the shock-associated color -- with participants making slower and more frequently erroneous saccades to the target when coupled with a previously shock-associated distractor--even though such behavior was opposite that which was previously required to avert shock. In a second experiment, participants probabilistically received a shock immediately upon fixating one of two color distractors in an additional singleton task. Rather than this contingency reducing distractor fixations, which would have been the adaptive behavioral response, aversive outcomes ironically led to greater interference from, and more frequent saccades on, the shock-associated stimulus, directly leading to more total shocks. Altogether, our findings highlight a powerful role for learning from punishment in the shaping of attentional priority.
Acknowledgement: NARSAD Young Investigator Grant 26008