Gopathy Purushothaman

Moving ahead through differential visual latency

The time it takes to transmit information along the human visual pathways introduces a substantial delay in the processing of images that fall on the retina. This visual latency might be expected to cause a moving object to be perceived at a position behind its actual one, disrupting the accuracy of visually guided motor actions such as catching or hitting, but this does not happen. It has been proposed that the perceived position of a moving object is extrapolated forwards in time to compensate for the delay in visual processing.

Quantum neural networks (QNNs): inherently fuzzy feedforward neural networks

This paper introduces quantum neural networks (QNNs), a class of feedforward neural networks (FFNNs) inherently capable of estimating the structure of a feature space in the form of fuzzy sets. The hidden units of these networks develop quantized representations of the sample information provided by the training data set in various graded levels of certainty. Unlike other approaches attempting to merge fuzzy logic and neural networks, QNNs can be used in pattern classification problems without any restricting assumptions such as the availability of a priori knowledge or desired membership profile, convexity of classes, a limited number of classes, etc. Experimental results presented here show that QNNs are capable of recognizing structures in data, a property that conventional FFNNs with sigmoidal hidden units lack.

Gating and control of primary visual cortex by pulvinar

The primary visual cortex (V1) receives its driving input from the eyes via the lateral geniculate nucleus (LGN) of the thalamus. The lateral pulvinar nucleus of the thalamus also projects to V1, but this input is not well understood. We manipulated lateral pulvinar neural activity in prosimian primates and assessed the effect on supra-granular layers of V1 that project to higher visual cortex. Reversibly inactivating lateral pulvinar prevented supra-granular V1 neurons from responding to visual stimulation. Reversible, focal excitation of lateral pulvinar receptive fields increased the visual responses in coincident V1 receptive fields fourfold and shifted partially overlapping V1 receptive fields toward the center of excitation. V1 responses to regions surrounding the excited lateral pulvinar receptive fields were suppressed. LGN responses were unaffected by these lateral pulvinar manipulations. Excitation of lateral pulvinar after LGN lesion activated supra-granular layer V1 neurons. Thus, lateral pulvinar is able to powerfully control and gate information outflow from V1.

Gamma-range oscillations in backward-masking functions and their putative neural correlates

Backward-masking functions have been hitherto categorized into two types, commonly named Type A and Type B. The analysis of a model of Retino-Cortical Dynamics produces the prediction that spatially localized stimuli should reveal an oscillatory metacontrast function. The predicted new type of metacontrast masking function was investigated in a psychophysical experiment. The results show oscillatory metacontrast functions with significant power in the gamma range (30-70 Hz). A marked decrease in the oscillations is observed when the spatial extent of the stimuli is increased. The theoretical basis of the study relates the oscillations found in the metacontrast function to gamma-range oscillations observed in scalp and intracerebral recordings. The qualitative agreement between the model and data provides support for this putative relationship.

Motion deblurring in a neural network model of retino-cortical dynamics.

Simulations of a neural network model of retino-cortical dynamics (Oğmen H, Neural Netw 6 (1993) 245-273) are presented. The temporal-step response of the model to a single dot (spatial impulse) consists of three post-retinal phases: reset, feed-forward dominant and feedback dominant. In response to a single moving dot, the model predicts the perception of extensive blur. This extensive blur is proposed to be due to the relative spatial and temporal offsets between transient and sustained signals conveyed from retina to post-retinal levels. In response to a pair of horizontally separated dots moving in the horizontal direction, the model predicts extensive blur for the trailing dot irrespective of dot-to-dot separation. For the leading dot, the model predicts a decrease in perceived blur for long exposure durations when dot-to-dot separations are small. The reduction of perceived blur at long exposure durations for small dot-to-dot separations is proposed to stem from the spatio-temporal overlap between the transient activity generated by the trailing dot and the sustained activity generated by the leading dot. The model also predicts that targets moving at higher speeds generate more blur even when blur is normalized with respect to speed. The mechanism in the model generating this effect is a slow inhibition within the sustained channel. These predictions are compared with recent psychophysical data (Chen S, Bedell HE, Oğmen H, Vis Res 35 (1995) 2315-2328) and are found to be in excellent agreement. The model is used to offer a coherent explanation for several controversial findings published in the literature. This computational study shows that a model without any motion-compensation mechanism can give a good account of motion deblurring phenomenon and supplements our recent experimental study which provided evidence against motion-compensation type models in explaining the motion deblurring phenomenon (Chen S, Bedell HE, Oğmen H, Vis Res 35 (1995) 2315-2328).

Fuzzy pattern classification using feedforward neural networks with multilevel hidden neurons

This paper introduces feedforward neural networks inherently capable of fuzzy classification of non-sparse or overlapping pattern classes. These networks are unique in that the hidden layers consist of multilevel neurons. The multilevel hidden neurons allow the networks to learn the fuzziness in the input data and also to minimize the within-class variances. The performance of the proposed networks over an overlapping pattern set is compared with that of conventional feedforward networks trained for crisp classification and those trained for fuzzy classification. The results show that the proposed networks reduce misclassification errors and have considerably better generalization ability.<<ETX>>

Morphological and neurochemical comparisons between pulvinar and V1 projections to V2

The flow of visual information is clear at the earliest stages: the retina provides the driving (main signature) activity for the lateral geniculate nucleus (LGN), which in turn drives the primary visual cortex (V1). These driving pathways can be distinguished anatomically from other modulatory pathways that innervate LGN and V1. The path of visual information after V1, however, is less clear. There are two primary feedforward projections to the secondary visual cortex (V2), one from the lateral/inferior pulvinar and the other from V1. Because both lateral/inferior pulvinar and V2 cannot be driven visually following V1 removal, either or both of these inputs to V2 could be drivers. Retinogeniculate and geniculocortical projections are privileged over modulatory projections by their layer of termination, their bouton size, and the presence of vesicular glutamate transporter 2 (Vglut2) or parvalbumin (PV). It has been suggested that such properties might also distinguish drivers from modulators in extrastriate cortex. We tested this hypothesis by comparing lateral pulvinar to V2 and V1 to V2 projections with LGN to V1 projections. We found that V1 and lateral pulvinar projections to V2 are similar in that they target the same layers and lack PV. Projections from pulvinar to V2, however, bear a greater similarity to projections from LGN to V1 because of their larger boutons (measured at the same location in V2) and positive staining for Vglut2. These data lend support to the hypothesis that the pulvinar could act as a driver for V2. J. Comp. Neurol. 521:813–832, 2013.

Retinotopic maps in the pulvinar of bush baby (otolemur garnettii)

Despite its anatomical prominence, the function of the primate pulvinar is poorly understood. A few electrophysiological studies in simian primates have investigated the functional organization of pulvinar by examining visuotopic maps. Multiple visuotopic maps have been found for all studied simians, with differences in organization reported between New and Old World simians. Given that prosimians are considered closer to the common ancestors of New and Old World primates, we investigated the visuotopic organization of pulvinar in the prosimian bush baby (Otolemur garnettii). Single‐electrode extracellular recording was used to find the retinotopic maps in the lateral (PL) and inferior (PI) pulvinar. Based on recordings across cases, a 3D model of the map was constructed. From sections stained for Nissl bodies, myelin, acetylcholinesterase, calbindin, or cytochrome oxidase, we identified three PI chemoarchitectonic subdivisions, lateral central (PIcl), medial central (PIcm), and medial (PIm) inferior pulvinar. Two major retinotopic maps were identified that cover PL and PIcl, the dorsal one in dorsal PL and the ventral one in PIcl and ventral PL. Both maps represent central vision at the posterior end of the border between the maps, the upper visual field in the lateral half and the lower visual field in the medial half. They share many features with the maps reported for the pulvinar of simians, including the location in pulvinar and the representation of the upper–lower and central–peripheral visual field axes. The second‐order representation in the lateral map and a laminar organization are likely features specific to Old World simians. J. Comp. Neurol. 521:3432‐3450, 2013.

Effect of exposure duration, contrast and base blur on coding and discrimination of edges.

We extend a neural network model, developed to examine neural correlates for the dynamic synthesis of edges from luminance gradients (Oğmen, 1993), to account for the effects of exposure duration, base blur and contrast on the perceived sharpness of edges. This model of REtino-COrtical Dynamics (RECOD) predicts that (i) a decrease in exposure duration causes an increase in the perceived blur and the blur discrimination threshold for edges, (ii) this increase in perceived blur is more pronounced for sharper edges than for blurred edges, (iii) perceived blur is independent of contrast while the blur discrimination threshold decreases with contrast, (iv) perceived blur increases with increasing base blur while the blur discrimination threshold has a nonmonotonic U-shaped dependence on base blur, (v) the perceived location of an edge shifts progressively towards the low-luminance side of the edge with increasing contrast, and (vi) perceived contrast of suprathreshold stimuli is essentially independent of spatial frequency over a wide range of contrast values. These predictions are shown to be in quantitative agreement with existing psychophysical data from the literature and with data collected on three observers to quantify the effect of exposure duration on perceived blur.

Perceptual decision related activity in the lateral geniculate nucleus

Fundamental to neuroscience is the understanding of how the language of neurons relates to behavior. In the lateral geniculate nucleus (LGN), cells show distinct properties such as selectivity for particular wavelengths, increments or decrements in contrast, or preference for fine detail versus rapid motion. No studies, however, have measured how LGN cells respond when an animal is challenged to make a perceptual decision using information within the receptive fields of those LGN cells. In this study we measured neural activity in the macaque LGN during a two-alternative, forced-choice (2AFC) contrast detection task or during a passive fixation task and found that a small proportion (13.5%) of single LGN parvocellular (P) and magnocellular (M) neurons matched the psychophysical performance of the monkey. The majority of LGN neurons measured in both tasks were not as sensitive as the monkey. The covariation between neural response and behavior (quantified as choice probability) was significantly above chance during active detection, even when there was no external stimulus. Interneuronal correlations and task-related gain modulations were negligible under the same condition. A bottom-up pooling model that used sensory neural responses to compute perceptual choices in the absence of interneuronal correlations could fully explain these results at the level of the LGN, supporting the hypothesis that the perceptual decision pool consists of multiple sensory neurons and that response fluctuations in these neurons can influence perception.