Reward produces learning of a consciously inaccessible feature

Reward has a significant impact on behaviour and perception. Most past work in associative reward learning has used perceptually distinct visual cues to associate with different reward values. Thus, it remains unknown to what extent the learned bias towards reward-associated stimuli depends on consciousness of the apparent differences between stimuli. Here, we resolved this issue by using an inter-ocular suppression paradigm with the monetary rewarding and non-rewarding cues identical to each other except for their eye-of-origin information. Thus, the reward coding system cannot rely on consciousness to select the reward-associated cue. Surprisingly, the targets in the rewarded eye broke into awareness faster than those in the non-rewarded eye. We further revealed that producing this effect required both top-down attention and inter-ocular suppression. These findings suggest that the human's reward coding system can produce two different types of reward-based learning. One is independent of consciousness yet fairly consuming attentional resources. The other one results from volitional selection of stimuli of behavioural significance.