Purchase this article with an account.
Brian Anderson, Hiroto Kuwabara, Dean Wong, Joshua Roberts, Arman Rahmim, James Brašić, Susan Courtney; Learning Mechanisms Underlying Value-Driven Attention. Journal of Vision 2017;17(10):1101. doi: https://doi.org/10.1167/17.10.1101.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Reward learning can shape attentional priorities such that previously reward-associated stimuli capture attention. The neural correlates of value-driven attention have become a topic of increasing interest, with studies revealing multiple brain areas throughout the visual system that show elevated responding to previously reward-associated distractors. These prior studies, however, have tended to focus on the consequences of reward learning, without addressing the learning mechanisms responsible. We will present two studies that begin to tackle this difficult and important issue. Using whole-brain fMRI, we measured reward processing in the training phase of the value-driven attentional capture paradigm (Anderson et al., 2011). The receipt of high reward evoked elevated activity in regions commonly attributed to reward processing, but also in several visual areas previously implicated in value-driven attention. These reward signals in visual areas contained information about the immediately preceding target stimulus, such that its position and identity could be reliably decoded from the reward-evoked activity on high- but not low-reward trials. In a second study using PET, we measured dopamine release during visual search with and without reward feedback. Attentional capture by previously reward-associated stimuli in a subsequent test phase was strongly correlated with changes in endogenous dopamine levels within the striatum attributable to the processing of reward during training. Together, these studies suggest a possible learning mechanism underlying value-driven attention. Dopaminergic reward signals from the striatum predict subsequent capture, and could serve as the putative teaching signals to visual areas measured using fMRI. With repeated stimulus-reward pairings, these teaching signals could serve to potentiate the rewarded visual representation, allowing the associated object to compete more effectively for selection. The proposed learning mechanism can account for classical value-driven attentional capture, attentional capture by reward cues that are never targets, and reward-mediated priming, offering a unifying account of reward-related attention effects.
Meeting abstract presented at VSS 2017
This PDF is available to Subscribers Only