Purchase this article with an account.
Emily Ward, Marvin Chun; Neural coding of perceptual features is enhanced when they are task relevant. Journal of Vision 2014;14(10):22. doi: 10.1167/14.10.22.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Multi-voxel pattern analysis has allowed us to investigate neural coding of stimulus-specific visual information by constructing high-dimensional representational spaces. Despite their utility for exploring visual representation, the extent to which multi-voxel relationships change as a function of task or attentional demands has not been widely explored. We scanned 10 participants while they viewed items that varied along three feature dimensions: shape, color, and texture. Participants either viewed the items passively, or were instructed to attend to one of the dimensions (e.g. "shape") and indicate the feature value for each item (e.g. "circle", "triangle", or "square"). This allowed us to examine neural discriminability among feature values with and without an explicit task, and more importantly, when a particular dimension was task relevant or not. In the lateral occipital cortex, feature values could not be classified during passive viewing, but could be classified when participants performed the feature-relevant task. Critically, across all tasks, task-relevant features (e.g. "circle" when attending to shape) could be classified, but task-irrelevant features (e.g. "red" when attention to shape) could not. These results are consistent with previous studies that show category learning can fine tune neural representations (Folstein et al., 2012), and widespread cortical tuning towards the task-relevant objects and away from task-irrelevant objects (Cukur et al., 2013). Our results show that task relevance impacts the representation of even simple object features. This suggests that representation in high-level visual areas may dynamically shift to facilitate behavior.
Meeting abstract presented at VSS 2014
This PDF is available to Subscribers Only