Purchase this article with an account.
Hila Harris, Michael Gliksberg, Dov Sagi; Specificity in texture learning is a result of uninterrupted stimulus repetition. Journal of Vision 2012;12(9):770. doi: 10.1167/12.9.770.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Practice is known to result in long term improvement in visual sensitivity. This perceptual learning was found to be ubiquitously specific to stimulus features. For instance, sensory improvements apply only for the trained retinal location and do not transfer across space. Here we ask whether specificity is inherent to perceptual learning, or, possibly, an outcome of the repetition-based training methods universally applied.
In the experiments, spatial transfer was tested using the backward-masked texture-discrimination task (Karni & Sagi, 1991), in three main conditions: (1) standard, repetition based method; (2) dummy, background only, task-irrelevant stimuli interleaved with the trained stimuli to disrupt the repeated stimulation pattern; (3) local dummy, where the background texture presented in (2) was restricted to the trained target region.
During 4 days of training at a single location, thresholds improved from 175ms to 75ms, with no significant differences between conditions. On the 5th day, target location was altered to test generalization. The standard repetition based method (1) showed the known specificity (thresholds increased to initial level). However, complete transfer to a new location was found with both (2 & 3) "dummy" methods, as learning thresholds remained low. Additional experiments showed that the transfer was not due to the increased temporal intervals between target presentations. Moreover, transfer was found with the standard method (1) but with target presented at one of two locations, showing the importance of target location, rather than task, consistency.
Our results show that generalization of learning is obtained by breaking stimulus repeatability. Specificity is the outcome of enhanced local spatiotemporal-associations, once interfered with enables the learning of space invariant stimulus features. It seems that these associations are formed within a low level visual system, where information is locally encoded, over a time frame of a few seconds but malleable to interference.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only