Abstract
Rational economics suggests that a gain should have the same value as evasion of equivalent loss. But Kahneman and Tversky famously showed that in decision making this is not the case: how we frame the outcome matters. Is vision equally irrational? Recent studies have shown that stimuli associated with monetary reward come to draw selective attention. If this reflects a rational bias toward objects providing outcome-predictive information, stimuli predicting both gain and the evasion of loss should be prioritized. To test this hypothesis, we cued participants to detect examples of object categories – cars, trees, houses, and people – presented in images of natural scenes. For each category, correct performance resulted in reward (+150 vs. +50), the avoidance of greater loss (-50 vs. -150), or neutral outcome. We modelled results according to two predictions: if selection is driven by a bias toward outcome-predictive stimuli, both predictors of reward and of the avoidance of loss should draw attention equally well. But if selection is driven by gain, only gain-associated stimuli should be prioritized. In a behavioural experiment performance closely followed the gain-only model: accuracy was high when targets predicted reward and low when targets predicted loss avoidance. Moreover, reward-predictive stimuli disrupted search when they acted as distractors but stimuli predicting loss avoidance were easy to ignore. In a subsequent fMRI decoding study we observed the same pattern in the quality of stimuli representations in ventral visual cortex. Stimulus-evoked activity in the dopaminergic midbrain also followed the gain-only model, such that activity in this area predicted the quality of visual representation. The association of reward to a category of stimuli thus appears to have an effect on visual cognition that is economically irrational and independent of the drive to seek information.
Meeting abstract presented at VSS 2016