Vergence dynamics predict fixation disparity.

The neural origin of the steady-state vergence eye movement error, called binocular fixation disparity, is not well understood. Further, there has been no study that quantitatively relates the dynamics of the vergence system to its steady-state behavior, a critical test for the understanding of any oculomotor system. We investigate whether fixation disparity can be related to the dynamics of opponent convergence and divergence neural pathways. Using binocular eye movement recordings, we first show that opponent vergence pathways exhibit asymmetric angle-dependent gains. We then present a neural model that combines physiological properties of disparity-tuned cells and vergence premotor cells with the asymmetric gain properties of the opponent pathways. Quantitative comparison of the model predictions with our experimental data suggests that fixation disparity can arise when asymmetric opponent vergence pathways are driven by a distributed disparity code.

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.

Recent models and findings in visual backward masking: a comparison, review, and update.

Visual backward masking not only is an empirically rich and theoretically interesting phenomenon but also has found increasing application as a powerful methodological tool in studies of visual information processing and as a useful instrument for investigating visual function in a variety of specific subject populations. Since the dual-channel, sustained-transient approach to visual masking was introduced about two decades ago, several new models of backward masking and metacontrast have been proposed as alternative approaches to visual masking. In this article, we outline, review, and evaluate three such approaches: an extension of the dual-channel approach as realized in the neural network model of retino-cortical dynamics (Ogmen, 1993), the perceptual retouch theory (Bachmann, 1984, 1994), and the boundary contour system (Francis, 1997; Grossberg & Mingolla, 1985b). Recent psychophysical and electrophysiological findings relevant to backward masking are reviewed and, whenever possible, are related to the aforementioned models. Besides noting the positive aspects of these models, we also list their problems and suggest changes that may improve them and experiments that can empirically test them.

Nonlinear alteration of transient vergence dynamics after sustained convergence.

BACKGROUND: Adaptation models of the horizontal disparity vergence system assume a nonadaptable transient component. They also predict identical postadaptation dynamics during convergence and divergence movements. METHOD: To test the adaptation property of the transient component, a set of experiments were performed in which closed-loop vergence dynamics measured before and after sustained convergence were compared, primarily by comparing the peak vergence velocity, occurrence time of peak vergence velocity, and steady-state vergence posture. Vergence dynamics after durations of 30, 60, and 90 s of sustained convergence were compared with those after a control duration of 5 s. RESULTS: The peak divergence velocity was reduced by about 25% within 30 s of sustained vergence. However, the peak convergence velocity was unchanged for all the exposure durations. Additionally, for all durations, the peak divergence velocity was significantly higher than peak convergence velocity. In contrast to peak velocities, the occurrence time of peak convergence and divergence velocity did not differ significantly and remained unchanged for all durations. CONCLUSIONS: The transient component is adaptable. Furthermore, the adaptation is direction dependent and affects divergence and convergence dynamics differently, thereby suggesting involvement of separate pathways for convergence and divergence in the vergence sensorimotor control.

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

Simulations of a neural network model of retino-cortical dynamics (Ogmen 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, Ogmen 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, Ogmen H, Vis Res 35 (1995) 2315-2328).

Neural network model of on-off units in the fly visual system: simulations of dynamic behavior.

We analyze the dynamic properties of a neural network model for on-off spiking neurons recorded in the first optic chiasm of the fly visual system. The model consists of two parallel pathways and three sequential processing stages. The first stage models photoreceptors. At the second stage, the signal is segregated into on- and off-pathways. These pathways are proposed to correspond to two populations of amacrine cells. At the third stage, the on- and off-pathways converge to on-off neurons. Furthermore, according to the model, on-off neurons interact via recurrent connections. This stage is proposed to correspond to lamina L4 neurons. In response to luminance increments and decrements, the model exhibits a three-component response and suggests pathways for each of the components. When stimulated by a train of pulses, the model exhibits fast adaptation for frequencies higher than about 5 Hz. Furthermore, adaptation to on- and off-pulses occurs independently. When the frequency of stimulation is reduced, the unit recovers rapidly from its adapted state. The temporal modulation transfer function has its peak around 7 Hz. The phase characteristics show a phase lead for low temporal frequencies changing to a phase lag for high frequencies. These model predictions are compared with data from Jansonius and van Hateren (1991).

Neural network model of short-term horizontal disparity vergence dynamics.

We present a neural network model of short-term dynamics of the human horizontal vergence system (HVS) and compare its predictions qualitatively and quantitatively with a large variety of horizontal disparity vergence data. The model consists of seven functional stages, namely: (1) computation of instantaneous disparity; (2) generation of a disparity map; (3) conversion of the disparity into a velocity signal; (4) push-pull integration of velocity to generate a position signal; (5) conversion of the position signal to motoneuron/plant activity for each eye; (6) gating of velocity overdrive signal to motoneuron/plant system; and finally (7) discharge path for position cells. Closed-loop (normal binocular viewing) symmetric step and staircase disparity vergence data were collected from three subjects and model parameters were determined to quantitatively match each subject's data. The simulated closed-loop as well as open-loop (disparity clamped viewing) symmetric step, sinusoidal, pulse, staircase, square and ramp wave responses closely resemble experimental results either recorded in our laboratory or reported in the literature. Where possible, the firing pattern of the neurons in the model have been compared to actual cellular recordings reported in the literature. The model provides insights into neural correlates underlying the dynamics of vergence eye movements. It also makes novel predictions about the human vergence system.

Two-dot alignment across the physiological blind spot.

Three competing hypotheses have been proposed for the cortical representation of the blind spot. These are: (i) the regions surrounding the blind spot maintain their spatial values; (ii) the opposite sides of the blind spot are represented adjacently at the cortex, so that the blind spot is "sewn-up"; and (iii) the blind spot is sewn-up with compensation occurring in the immediate surround of the blind spot, so that spatial values are distorted only in the immediate surround of the blind spot. To distinguish between these hypotheses we used a two-dot alignment task, with the two dots straddling the blind spot at varying dot separations. Thresholds in the two-dot alignment task are limited by the cortical separation of the two dots. When thresholds for alignment across the blind spot are compared with thresholds over intact retina at the same eccentricity, the three hypotheses predict: (i) no change in thresholds; (ii) a lowering of thresholds; and (iii) a lowering of thresholds but only at separations slightly greater than the diameter of the blind spot. Thresholds across the blind spot were closely similar to thresholds across intact retina. The results do not support a sewing-up (with or without compensation) of the blind spot. Rather, our results are consistent with a preservation of spatial values around the blind spot.

A target in real motion appears blurred in the absence of other proximal moving targets.

For exposure durations longer than about 40 msec, a field of dots in sampled motion has been reported to appear less smeared than predicted from the visual persistence of static displays. This reduction of perceived smear has been attributed to a motion "deblurring" mechanism. However, it has been long recognized that an isolated target moving continuously in a dark field appears to be extensively smeared. To reconcile these apparently contradictory observations, we investigated the effect of dot density on the extent of perceived smear for a single moving dot and for fields of dots with densities ranging from 0.75 to 7.5 dots/deg2. Bright targets were presented in continuous motion against a photopically illuminated background field. The results reconcile previous conflicting observations by showing that the length of perceived smear decreases systematically with dot density for exposure durations longer than about 50 msec. In three additional experiments, we arranged the spatial configuration of the targets to evaluate whether motion deblurring results primarily from a motion compensation mechanism (such as integration within the spatiotemporally oriented receptive fields of putative motion mechanisms) or from inhibition exerted by spatiotemporally adjacent targets. The results show that the activation of motion mechanisms is not a sufficient condition for motion deblurring and that the reduction of perceived smear requires the presence of spatiotemporally adjacent targets. Taken together, these findings suggest that motion deblurring results primarily from masking exerted by spatiotemporally proximal targets.

Perceived length across the physiological blind spot.

Objects falling across the physiological blind spot appear "complete" despite the absence of photoreceptors. Completion of objects may occur across the blind spot because (1) the blind spot is filled in with the background (the associative explanation); (2) the opposite sides of the blind spot may be contiguously represented in the cortex (i.e. the blind spot is simply sewn up-the retinotopic explanation); or (3) the blind spot may be sewn up, with compensatory expansion occurring around the blind spot (the compensation explanation). These theories would predict no size distortions regardless of object size; constant size distortions regardless of object size; and distortions that depend on the size of the object, respectively. To evaluate these explanations, we measured size distortions at the blind spot. We measured length distortions at the blind spot using a criterion-free two-alternative forced-choice method with feedback. Observers compared the lengths of test bars presented across the blind spot with lengths of reference bars presented at the corresponding location in the fellow eye. Test bar lengths ranged from 7-14 deg. Reference bar lengths were in the range of +/- 3 deg of test bar length. From the observers' responses the perceived length of each bar at the blind spot was estimated. Estimates of the precision of length discrimination at the blind spot were also obtained. Our results were consistent with the associative explanation. In all seven observers, length distortions at the blind spot were smaller than 1 deg (< 20% of the vertical height of the blind spot) for all bar lengths tested. For bars that were presented across the blind spot, the precision with which observers could discriminate length was comparable to that of normal periphery (Weber fraction approximately 20%). Both the veridicality and precision of perceived length are preserved around the blind spot.

Quantitative studies of fly visual sustained neurons.

We present a quantitative study of a neural network model [1] proposed for the sustained neurons in the fly visual system. Electrophysiological recordings of sustained neurons [2] are digitized and transferred to a computer. A numerical ordinary differential equation solver is used to simulate the model. In order to obtain an initial set of parameters, we introduce approximations to the model and obtain fits to parts of the response characteristics. These initial parameter values are then refined by optimization routines. The model is compared to data in 4 different experimental paradigms and in general is in good agreement with data. We conclude that the simplified versions of temporal and spatial adaptation mechanisms of the model capture the essential features of the dynamics of sustained neurons and that the refinement of the model requires further experimental studies to elucidate the number of stages involved in temporal adaptation as well as the precise shape of the relationship between the membrane potentials and the spike frequency for the sustained neurons.

A neural model for nonassociative learning in a prototypical sensory-motor scheme: the landing reaction in flies.

Nonassociative learning is an important property of neural organization in both vertebrate and invertebrate species. In this paper we propose a neural model for nonassociative learning in a well studied prototypical sensory-motor scheme: the landing reaction of flies. The general structure of the model consists of sensory processing stages, a sensory-motor gate network, and motor control circuits. The paper concentrates on the sensory-motor gate network which has an agonist-antagonist structure. Sensory inputs to this circuit are transduced by chemical messenger systems whose dynamics include depletion and replenishment terms. The resulting circuit is a gated dipole anatomy and we show that it gives a good account of nonassociative learning in the landing reaction of the fly.

On the Mechanisms Underlying Directional Selectivity.

Recent efforts in the understanding of motion detection and directional selectivity include electrophysiological studies using single photoreceptor stimulations and a combination of electrophysiology and neuropharmacology. Results of the former have been interpreted in favor of facilitator motion detection models while results of the latter have been interpreted in favor of inhibitory models. In this paper, this conflicting data interpretation problem is addressed by mathematically modeling some effects of neuropharmacological substances and by applying this formalism to a neural network model of directionally selective motion perception. The study offers a possible resolution to the paradox.

Neural models for SUSTAINED and ON-OFF units of insect lamina.

We present models based on transmitter gating and competition for SUSTAINED and ON-OFF units located in the lamina of the fly visual system. Predictions of the models are compared with data in several experimental paradigms. The overshoot and the plateau of the response are explained by the adaptation process of transmitter dynamics. Furthermore, it is demonstrated that this transmitter adaptation when coupled with fast competition can explain response features arising from different combinations of ON and OFF field inputs.