Author: Sepide MovaghatiPublisher:
ISBN:
Category :
Languages : en
Pages :
Book Description
"According to classical models of visual information processing, the visual cortex acts as a hierarchical structure in which information flow is unidirectional and transmitted from the primary visual cortex (V1) to higher visual areas. This model considers V1 as an inflexible processing unit, which filters local basic features of visual input, such as the spatial frequency and orientation of the changes in luminance. Contrary to this classical model, recent studies have reported variant activity in V1 based on the presence of higher-order features within visual inputs. In other words, specific characteristics, such as phase content or spatial coherence, potentially modulate the activity of V1. However, the explicit characteristics and the precise timing of their modulatory effect remain unclear. In this study, MagnetoEncephaloGraphy (MEG) was used to test whether the spatiotemporal response to natural images of animals is different from the response to their phase-shuffled counterparts. The frequency content and the total contrast of the images were maintained throughout the phase-scrambling approach. Meanwhile, higher-order image statistics, such as Kurtosis, were manipulated. We investigated the effect of this manipulation on the evoked response of V1 by taking advantage of (1) the high temporal resolution of the MEG technique and (2) the Minimum norm source localization method. The results of this thesis show that the modulation of activity associated with the presentation of natural images appears as early as 50ms post-stimulation. This effect can be observed in the power of the signals recorded by MEG sensors as well as in sources localized in the primary visual cortex. The phase scrambling of visual input increased the amplitude of the early response (30-70ms post-stimulation) and the M170 component (140-200ms post-stimulation). The weaker population response to natural images during the early response is consistent with a model proposing sparse neuronal coding for natural images. The increased M170 activation of V1 to non-natural phase-scrambled images is consistent with the predictive coding model. According to this model, the prediction error in lower order sensory areas increases with decreasing predictability of the sensory input. An increase in the power of gamma frequency (60-80Hz) was observed in response to phase-scrambled images. These findings are in accordance with a previous study that found prediction errors in auditory cortex to be predominantly conveyed via the high (gamma) frequency range. In addition, spatially coherent features in visual input induce a stronger surround suppression effect and reduce the population activity in V1. Therefore, the mechanisms underlying the late modulatory effect on the M170 component may also involve surround suppression in the visual cortex. Our findings suggest that feedback modulates the neuronal response of the primary visual cortex. Further, they identify V1 as a dynamic filter that can be modulated by the higher-order characteristics of visual input, as opposed to previous models that deem V1 to be an inflexible low-level processing unit." --