, 2011). The way in which perceptual learning is represented in the cortex may be dependent on the nature of the Ponatinib discrimination task. It is important, for example,

to distinguish between learning on lower order properties, such as those associated with inputs to the cortex (somatosensory vibration or acoustic frequency), feedforward properties such as orientation tuning, and the higher-order properties that are dependent on context, such as three-line bisection, vernier discrimination, or contour detection and shape discrimination. The cortical changes associated with contextually dependent perceptual learning have to account for its specificity. In fact, the way learning is represented in these tasks is to influence contextual interactions that are relevant to that task. This is exemplified by changes

in contour integration accompanying learning in a contour detection task (Figure 7; Li et al., 2008) and changes in modulation of responses by changing the distance between parallel lines in a three-line bisection task (Crist et al., 2001; Li et al., 2004). By enhancing the modulation in neuronal tuning to stimulus components that are relevant to the task, learning increases the task relevant information conveyed by neurons. As subjects learn a task, there is a change in the functional properties of neurons encoding the information involved in the task. Remarkably, one can see this occur even in V1. As shown in Figure 7, the ability to detect a contour composed of collinear line segments embedded in a complex background SAHA HDAC purchase improves with practice. Longer contours made of a larger number of line segments are easier to detect than those made of fewer line segments, and the number of segments required to reliably detect the contour decreases with practice. One can see from the black dashed psychometric STK38 curve the increase in detectability as a function of the number of line elements. This represents the animals’ performance early in the period of training, during the first week. This curve steepens with practice (red dashed curve), showing

the improvement in performance as a result of perceptual learning in the task. If one measures the contour related responses in V1, there is a corresponding steepening of the neurometric curve that tells how well an ideal observer can detect the embedded contours of different lengths simply based on neuronal responses. Perceptual learning can enable neurons to carry information that is required to perform complex visual discrimination tasks, not only for contour detection, as described above, but for discriminating the shapes of contours embedded in complex scenes. For animals trained in a task requiring discriminating a circle, a straight line or a wave shape, neurons take on selectivity for related shapes (Figures 8 and 9).