Computational Neuroscience of Vision

Abstract

UNIVERSITY PRESS χ Ι Contents 2.4 Computational processes that give rise to V1 simple cells 49 2.4.1 Linsker's method: Information maximization 50 2.4.2 Olshausen and Field's method: Sparseness maximization 53 2.5 The computational role of V1 for form processing 2.6 Backprojections to the lateral geniculate nucleus Extrastriate visual areas 57 65 3.4.2 Depth perception The parietal cortex 4.1 Introduction 4.2 Spatial processing in the parietal cortex 4.2.1 Area LIP 71 4.2.2 Area VIP 4.2.3 Area MST 4.2.4 Area 7a 4.3 The neuropsychology of the parietal lobe 4.3.1 Unilateral neglect 4.3.2 Balint's syndrome 77 4.3.3 Gerstmann's syndrome 79 Contents xvii 13.4 Continuous neuronal dynamics allows fast network processing 13.5 Hierarchical feature analysis 13.6 Trace learning rule for invariant representations 13.7 Spatial feature binding by feature combination neurons 13.8 IT provides a representation for later memory networks 13.9 Face expression and object motion 13.10 Attentional mechanisms 13.11 Visual search 13.12 Egocentric vs allocentric representations 13.13 Short term memory as the controller of attention 13.14 Output to object selection and action systems 13.15 'What' versus 'where' processing streams 13.16 Short term memory must be separated from perception 13.17 Backprojections must be weak 13.18 Long-term potentiation and short-term memory 13.19 "Executive control" by the prefrontal cortex 13.20 Reward processing occurs after object identification 13.21 Effects of mood on memory and visual processing 13.22 Visual outputs to Long Term Memory systems 13.23 Episodic memory and the operation of mixed discrete and continuous attractor networks 13.24 Visual outputs to behavioural response systems 13.25 Multimodal representations in different brain areas 13.26 Visuo-spatial scratchpad and change blindness 13.27 Invariant object recognition and attention 13.28 Conscious visual perception 13.29 Attention-future directions 13.30 Integrated approaches to understanding vision 13.31 Apostasis

Object Class Recognition and Localization Using Sparse Features with Limited Receptive Fields

“…We wanted to investigate the option of retaining some geometric information above the S2 level. In fact, neurons in V4 and IT do not exhibit full invariance and are known to have receptive fields limited to only a portion of the visual field and range of scales [30]. To model this, we simply restrict the region of the visual field in which a given S2 feature can be found, relative to its location in the image from which it was originally sampled, to ±t p % of image size and ±t s scales, where t p and t s are global parameters.…”

Distributed Computation using Evolutionary Consciousness : An Approach

“…In the bottom-up neocortical model, the two distributed neuronal nodes perform cognitive actions considering chained unidirectional information flow. It is observed that, at the algorithmic levels, the cognitive functions become computationally intractable although living species perform such functions without delay [15]. The reason is that, the living organisms often employ heuristics and greedy algorithmic models in problem solving.…”

An Approach to the Parameterization of Structure for Fast Categorization

“…Neural networks describe contours by a hierarchical integration of their local orientations in a pyramid space (e.g. (Serre et al, 2007;Hansen and Neumann, 2004;Amit and Mascaro, 2003;VanRullen and Thorpe, 2002;Rolls and Deco, 2002)). Yet such a description is essentially template matching in a more refined form.…”