Abstract
Reward motivation enhances memory formation (Adcock et al., 2006). Here we examine the retroactive effect of reward on memory formation occurring during visual statistical learning, a type of incidental learning. In a learning phase, participants were required to passively view a sequence of images. Unbeknownst to participants, the thirty-two images were statistically structured as eight quadruplets, and quadruplets were equally divided into two types of sequences --rewarded sequences (e.g., ABCD) and non-rewarded sequences (e.g., A’B’C’D’). Subsequently, in the value-learning phase, participants were told there were quadruplets in the previous phase, and half of the quadruplets were rewarded sequences, while the other half were non-rewarded sequences. During this phase, the last items of quadruplets (e.g., D and D’) were shown, and participants had a chance to learn whether those last items belong to a rewarded sequence or a non-rewarded sequence. Following these two learning phases, two memory tests were performed. In the sequence recognition test, two sequences (one target and one foil) were presented, and participants had to choose which was more familiar. Critically, target and foil sequences consisted of only the first three items of quadruplet (e.g., target: ABC or A’B’C’; foil: AFK or A’F’K’). We excluded the last item to ask whether the learned reward association retroactively refreshed memory for other members of a sequence. We found that recognition rate of rewarded sequences was significantly higher than non-rewarded sequences, despite the absence of the image learned during value-learning and despite the absence of reward during the passive viewing phase. In a final reward memory phase, participants showed above chance ability to identify reward vs. non-reward images. Our results are the first to show that memory associations can be enhanced by being refreshed with novel reward information, even when memory was indirectly associated with reward.