Abstract
Practice helps improve performance on a variety of visual tasks. Previous studies have shown that the magnitude of these improvements is inversely proportional to initial levels of performance, with subjects who perform more poorly at the start tending to improve most during perceptual training. If initial performance levels determine the absolute magnitude of learning, it follows that equating performance at the start of training should lead to equivalent amounts of learning. Here we test this prediction by comparing learning on an abutting Vernier alignment task with stimuli presented at two retinal eccentricities (5 and 15 deg) equated in terms of either retinal size (unscaled stimuli) or cortical size (scaled stimuli). Prior to learning, unscaled stimuli produced larger alignment thresholds at the more peripheral eccentricity, whereas scaled stimuli produced equivalent alignment thresholds. Consistent with previous work, we found that the magnitude of learning for participants who trained over eight daily sessions with the unscaled stimuli (n=11) was significantly larger at 15 than 5 degrees eccentricity. However, when stimuli were spatially scaled (n=11), we found equivalent amounts of learning at each location. These data suggest differences in the magnitude of learning can be accounted for by differences in the cortical representation of stimuli. Cortical scale may set not only the initial performance level but also the upper limit for the magnitude of performance improvements following training.