Abstract
Perceptual learning can transfer to a new retinal location if additional training with an irrelevant task primes the new location (“double-training”, Xiao et al., 2008), suggesting perceptual learning a more general process occurring in non-retinotopic high brain areas. However, learning is most useful if it can transfer to other untrained locations after double-training. We measured the transfer of Vernier learning, which is very location specific, to untrained locations after double-training. (1) Two orthogonal Vernier stimuli were trained at separate visual quadrants. After training learning transferred to other untrained quadrants. (2) Vernier and a new task were trained at diagonal quadrants, respectively. Vernier learning always transferred to the diagonal quadrant. However, whether it also transferred to other untrained quadrants depended on the location specificity of the new learning task. The location-non-specific motion direction or orientation learning can piggyback Vernier learning to other locations at the same or different eccentricities, but location-specific contrast learning cannot. (3) Similar trends were observed when double-training was performed at the same location. Two orthogonal Vernier training showed no transfer of learning to other locations. However, Vernier learning did transfer to other locations with motion direction/orientation training. We proposed that location specificity results from the learned high-level decision failing to functionally connect to the new inputs representing untrained locations. Double-training reactivates the new inputs to establish the functional connections to enable learning transfer. The current study shows that the brain may have learned to discount the location information and make the learning solely feature-based when a same task is trained at two locations. Moreover, for motion/orientation learning the learned decision unit may always connect to other locations that are constantly primed by motion/orientation stimuli. Vernier learning, once becoming contingent with these tasks, can be piggybacked to other locations through these tasks' active network.
NSFC Grant 30725018 (C.Y.) and NIH Grants R01-04776 (S.A.K) and R01-01728 (D.M.L).