Purchase this article with an account.
Brad Wyble, Chloe Callahan-Flintoft; A system level model of visual attention: Targets and distractors are figments of your experimental design. Journal of Vision 2017;17(10):960. doi: 10.1167/17.10.960.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Attentional capture studies provide clues as to how the visual system addresses the challenge of selecting task-relevant information, while remaining sensitive to distracting information. This ability allows us to notice unanticipated stimuli when performing a challenging visual task like driving a car. Clearly, distracting information captures attention and task-defined control settings guide attention towards relevant information, but our understanding of how these processes are implemented at the neural level is lacking. For example, in additional singleton paradigms, theories of "distractor suppression" cannot elucidate how the visual system determines which neurons should be suppressed. To better understand the neural implementation of attentional control, a system-level model of visual attention has been developed that simulates a broad set of behavioral and neural constraints (i.e. ERPs related to attention). In the model, stimuli evoke representations in an attentional map that mediates a competition between them. Attentional set is implemented as weights that bias attention towards task-relevant stimuli, but any stimulus with sufficient salience can activate attention. This model elucidates candidate attentional control mechanisms, and at the mechanism-level there are no targets or distractors per se. Rather, stimuli have varying levels of intrinsic salience and relevance to the task. The model's attention map mediates a competition between representations of those stimuli and then deploys attention to enact that decision, enhancing the winner(s) and suppressing the loser(s). The model helps to resolve ongoing debates as to whether distractor suppression is deployed as a target-surround, or selectively towards distractors. The model suggests that both occur, being differentially observable in different tasks. The model also resolves the distinction between spotlight and divided attention theories by showing that either can occur in different task contexts. Having been developed, the model now becomes part of a cyclical process of prediction, testing and model refinement.
Meeting abstract presented at VSS 2017
This PDF is available to Subscribers Only