Furthermore, each neuron’s shape selectivity depends on the shape the animal is currently looking for, reflecting the importance of top-down influences on the functional

properties of these neurons. The same neurons change their shape selectivities according to the cued shape the monkey is looking for, and shape expectation induces a global shift in the set of shape Sunitinib chemical structure selectivities of the population of superficial layer V1 neurons. One can think of the properties developed as a result of learning this task in terms of the association field: the anatomical circuitry (Figure 3) allows a wide range of potential shape selectivities, which represent the full Ibrutinib extent of the association field. At any given time only a subset of these connections are effective, and only a portion of the association field expressed, depending on the shape that the animal is expecting. Perceptual learning may also change which cortical areas represent

the trained stimulus. In visual search tasks the ability of a stimulus to pop-out from an array of distracters depends on familiarity with the stimulus (Wang et al., 1994). One can follow the development of this pop-out quality during the period of training. Subjects learn to identify the target one location at a time, as if the target is being represented at multiple locations within a retinotopically organized area (Figure 10; Sigman et al., 2005). Consistent with this idea, cortical activation measured with fMRI shows a shift

in activation, below from lateral occipital cortex, when the array contains untrained stimuli, to early visual cortex (V1/V2), when the array contains the trained stimulus. The training is useful for enabling subjects to identify shapes rapidly and in parallel with other shapes. Engaging early visual cortex in the task allows such parallel processing of shape features. This finding suggests that extensive training can shift the cortical representation of the learned shape from higher to lower visual areas for more efficient and less effortful processing. This idea is supported by the evidence that extensive training on a perceptual task significantly reduces activity in the frontoparietal attentional network (Mukai et al., 2007; Pollmann and Maertens, 2005; Sigman et al., 2005). As a consequence, the automatic and pop-out quality of visual search targets differing in attributes associated with early, retinotopically mapped areas (Treisman, 1998; Treisman and Gelade, 1980) can be extended to more complex objects as a result of training.