Abstract
[Goal]
Perceptual learning (PL), experience-induced improvement in perception, is typically highly specific to the trained location and feature. However, location specificity can be alleviated by particular training protocols. Manipulating exogenous (Donovan, Szpiro & Carrasco, 2015) or endogenous (Donovan & Carrasco, 2018) spatial attention during training facilitates learning transfer more efficiently than other protocols that require a secondary training task. Here we investigated whether feature-based attention (FBA), which enhances the representation of particular features throughout the visual field, facilitates location and/or feature transfer in PL.
[Methods]
To investigate the effects of FBA on specificity in PL, we implemented an orientation discrimination task in which observers were first presented with two reference angles simultaneously, then asked to discriminate whether the orientation of a Gabor stimulus was clockwise or counter-clockwise to either reference. In Experiment 1, we confirmed that FBA improved accuracy in this task. In Experiment 2, two groups of observers participated in a six-day PL study; the Attention group trained with a feature attention cue, indicating on a trial-by-trial basis which of the two reference angles was relevant for the discrimination, and the Neutral group trained with a neutral cue, indicating both reference angles. Observers were presented with neutral cues during both the pre-test (before training) and post-test (after training) sessions.
[Results]
For both groups of observers, performance improved for the trained feature at the trained location. Notably, training with FBA enabled complete learning transfer to the untrained location, but not to the untrained orientation. In contrast, the Neutral group exhibited both location and orientation specificity.
[Conclusion]
Our results show a perceptual benefit of FBA in an orientation discrimination task, and reveal a remarkable spatial-transfer of learning induced by FBA, reminiscent of its global effect across the visual field. This study has possible translational implications for perceptual training in visual rehabilitation.