Abstract
Specificity and transfer of perceptual learning has generally been measured in two-alternative tasks. And, although learning has rarely been reported for spatial frequency judgments, we recently showed learning in a majority of observers in 8-alternative spatial frequency (8AFC) identification (Dosher, Liu, & Lu, VSS 2017). However, little is known about specificity or generalization in nAFC tasks. Here, we examined whether learning in 8-alternative spatial frequency identification was specific to the orientation of the training stimulus. Stimuli were Gabors of 8 spatial frequencies (0.50 to 5.68 cpd, in half-octave steps) in 5 orientations (−55.5° to 34.5° relative to vertical, in 22.5° steps) and embedded in Gaussian external noise (sigma=0.24); observers were trained to identify which of the 8 spatial frequencies was displayed in each trial. Learning and transfer were evaluated by comparing 8AFC judgments in a post-test after training to a pre-test. Four different groups were trained for 5 sessions of 960 trials each with either the leftmost, middle, or rightmost orientation, or a mixture of all five orientations in three Gabor contrasts (0.3, 0.6, 1.0). Consistent with the earlier report, there was modest learning in each training condition; training in the mixed condition (“roving” the non-judged orientation) or in the middle orientation condition if anything showed larger improvements. There was orientation-specificity in the leftmost and rightmost training conditions, seen in the degree of transfer over orientations. A simulation of the n-AFC integrated reweighting theory (IRT, Dosher et al., 2013; Dosher, Liu, & Lu 2017) without orientation invariance spatial frequency representations predicts more specificity than observed in the human data. Example simulations suggest a possible role for orientation-invariant representations mediating orientation transfer in a way analogous to the role of location-invariant representations in transfer over retinal locations, consistent with orientation invariance in higher visual cortical regions.
Acknowledgement: Supported by the National Eye Institute Grant # EY–17491.