Specificity has always been considered one of the hallmarks of perceptual learning, suggesting that per- formance improvement would reflect changes at early stages of visual analyses (e.g., V1). More recently, however, this view has been challenged by studies documenting complete transfer of learning among dif- ferent spatial locations or stimulus orientations when a double-training procedure is adopted. Here, we further investigate the conditions under which transfer of visual perceptual learning takes place, confirm- ing that the passive stimulation at the transfer location seems to be insufficient to overcome learning specificity. By contrast, learning transfer is complete when performing a secondary task at the transfer location. Interestingly, (i) transfer emerges when the primary and secondary tasks are intermingled on a trial-by-trial basis, and (ii) the effects of learning generalization appear to be reciprocal, namely the pri- mary task also serves to enable transfer of the secondary task. However, if the secondary task is not per- formed for a sufficient number of trials, then transfer is not enabled. Overall, the results lend support to the recent view that task-relevant perceptual learning may involve high-level stages of visual analyses.
