Reward motivation influences response bias on a recognition memory task

Abstract

Reward-motivated memory has been studied extensively in psychology and neuroscience. Many recognition studies follow the same type of paradigm: stimuli are cued at encoding with high or low reward values which indicate the amount the stimulus is worth if successfully recognized on a subsequent memory test. Each incorrect endorsement of a lure at retrieval is penalized with an arbitrary value between the high and low reward value, resulting in a single false alarm rate. Studies employing this type of paradigm have reported higher hit rates for high value items compared to low value items, but generally hit rate is the only measure of memory that is reported as a function of reward value. It is currently not clear what aspects of the experimental design lead to these memory effects, and other measures, like discriminability and response bias, cannot be properly calculated when there is only a single false alarm rate, but we hypothesize that these are also susceptible to motivational manipulations. To test how reward anticipation might influence memory and response bias in this type of task, we created a novel paradigm that allowed us to calculate both by associating rewards with categories (indoor vs. outdoor scenes), thus calculating separate false alarm rate as well as hit rate at each level of reward. We report results of three experiments that varied rewards and penalties for correct and error responses for the category items. In two experiments, we replicated prior findings of higher hit rates for high compared to low reward items, but consistently across three experiments, when d′ was calculated, we found no difference in memory discriminability as a function of reward. Further, Experiment 1 we found that response bias was more conservative for low reward items: participants were more likely to endorse a ‘new’ response to low compared to high reward items. This effect was significantly reduced in Experiment 2 and eliminated in Experiment 3 when the reward-penalty structure was manipulated to reduce bias. Our findings reveal that reward motivation can influence decisional biases thought to be independent of memory processes. The amount of the reward value for correct responses and the amount of the penalty for incorrect responses should be considered when designing experimental paradigms to study motivation-cognition interactions.