T only to a reduce of activity but also a sharpening from the representation in

T only to a reduce of activity but also a sharpening from the representation in visual cortex. They discover that perceptual expectation leads to a reduction in neural activity in V1, but improves the stimulus representation, as measured by multivariate pattern analysis (Kok et al., 2012). In line with this notion, there has been much attention towards the selectivity of neurons involved in learning.Frontiers in Human Neurosciencewww.frontiersin.orgOctober 2013 Volume 7 Article 668 Seri and SeitzLearning what to expectPRIORS Within the SELECTIVITY Of the NEURONSA natural way in which (structural) priors could be represented within the brain is in the selectivity with the neurons along with the inhomogeneity of their preferred attributes (Ganguli and Simoncelli, 2010; Fischer and Pena, 2011; Girshick et al., 2011). In this framework, the neurons representing the expected attributes of the environment could be present in larger numbers (Girshick et al., 2011), or be much more sharply tuned PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/21366472 (Schoups et al., 2001), or extra strongly connected to larger processing stages (Raiguel et al., 2006) than neurons representing non-expected characteristics. By way of example, as discussed above, a Bayesian model using a prior on Ceruletide cardinal orientations (reflecting the fact that they may be a lot more frequent inside the natural environment) can account for the observed perceptual bias toward cardinal orientations. These effects can also be simply accounted for in a model in the visual cortex where additional neurons are sensitive to cardinal orientations, with these neurons getting also additional sharply tuned (as observed experimentally), combined using a very simple population vector decoder (Girshick et al., 2011). Comparable models have been proposed inside the auditory domain to clarify biases in localization of sources (Fischer and Pena, 2011) and formalized theoretically. Ganguli and Simoncelli (2010), one example is, supplied a thorough analysis of how priors may be implicitly encoded inside the properties of a population of sensory neurons, so as to supply optimal allocation of neurons and spikes offered some stimulus statistics. Interestingly, their theory makes quantitative predictions about the connection among empirically measured stimulus priors, physiologically measured neural response properties (cell density, tuning widths, and firing rates), and psychophysically measured discrimination thresholds (see also: Wei and Stocker, 2012). No matter whether all structural priors correspond to inhomogeneities in cell properties is unclear. The light-from-above prior is thought to become related to activity in early visual cortex (Mamassian et al., 2003), but, as far as we know, its precise relation with neural responses is yet unclear. The slow-speed prior, nevertheless, may very well be implemented in such a way, by way of an over-representation of very slow speeds in MT or even a shift in the tuning curves toward reduced speeds when contrast is decreased (Krekelberg et al., 2006; Seitz et al., 2008). Accordingly, there’s some proof that prolonged expertise with high-speeds leads to a shift with the MT population to favor greater speeds (Liu and Newsome, 2005).PRIORS Inside the NEURONS’ SPONTANEOUS ACTIVITYbe computationally advantageous, driving the network closer to states that correspond to most likely inputs, and therefore shortening the reaction time with the technique (Fiser et al., 2010). Berkes et al. (2011) lately supplied further evidence for this thought by analyzing spontaneous activity inside the main visual cortex of awake ferrets at diverse stages of development. They fou.