Search PubMed⌕ Search
Home/Archive/PubMed

PubMed · 8234315

Nonlinear processes in visual pattern discrimination.

H R Wilson
View original source ↗Export metadata ↓

Abstract

Visual performance on many simple pattern discrimination tasks can be accurately predicted by quasi-linear models composed of oriented linear filters followed by simple contrast nonlinearities. However, many complex discrimination tasks require highly nonlinear processes for their explanation. Evidence is provided for two nonlinear processes in pattern discrimination: (i) one process involves a sequence of filtering, rectification, and subsequent filtering to extract texture boundaries; (ii) the second process results from contrast gain-control processes. It is suggested that quasi-linear processes and nonlinear texture-boundary processes, each with an appropriate contrast gain control, may operate in parallel to provide the basis for all higher-level visual analyses.

Explore related subjects

Discrimination, PsychologicalFeedbackHumansMathematicsModels, BiologicalModels, PsychologicalPattern Recognition, VisualVisual AcuityJournal ArticleReview

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

H R Wilson. 1993-11-01. Nonlinear processes in visual pattern discrimination.. https://doi.org/10.1073/pnas.90.21.9785

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed1996

Neural network models of perceptual learning of angle discrimination.

We study neural network models of discriminating between stimuli with two similar angles, using the two-alternative forced choice (2AFC) paradigm. Two network architectures are investigated: a two-layer perceptron network and a gating network. In the two-layer network all hidden units contribute to the decision at all angles, while in the other architecture the gating units select, for each stimulus, the appropriate hidden units that will dominate the decision. We find that both architectures can perform the task reasonably well for all angles. Perceptual learning has been modeled by training the networks to perform the task, using unsupervised Hebb learning algorithms with pairs of stimuli at fixed angles theta and delta theta. Perceptual transfer is studied by measuring the performance of the network on stimuli with theta' not equal to theta. The two-layer perceptron shows a partial transfer for angles that are within a distance a from theta, where a is the angular width of the input tuning curves. The change in performance due to learning is positive for angles close to theta, but for magnitude of theta-theta' approximately a it is negative, i.e., its performance after training is worse than before. In contrast, negative transfer can be avoided in the gating network by limiting the effects of learning to hidden units that are optimized for angles that are close to the trained angle.

Discrimination, Psychological↗
PubMed1996

Localization and discrimination of "pop-out" targets.

In parallel visual search, a target pattern "pops out" among distractors rapidly, requiring no effort, regardless of distractor numbers. The localization and discrimination of "pop-out" targets was investigated for this research note using similar multiple displays to those used in Sagi and Julesz's [(1985a) Science, 228, 1217-1219] and Folk and Egeth's [(1989) Journal of Experimental Psychology: Human Perception & Performance, 15, 97-110] studies. The stimulus display contained 2, 5 or 10 oblique target line segments embedded in vertical distractor lines. In the first localization task, the observer indicated whether all the targets were in other positions. The second localization task was otherwise identical to the first one, except that the number of critical inside-corner positions was four. In the discrimination task, the observer reported whether all the target lines had the same orientation, or whether one of them differed in orientation from the others. Reaction times for correct responses were measured in all three tasks. The results showed that target discrimination took place in parallel, but target localization was a "serial" process, i.e. the localization time depended on the number of targets and critical locations to be checked.

Discrimination, Psychological↗
PubMed1995

Parvocellular neurons limit motion acuity in human peripheral vision.

It is generally believed that the perception of moving targets is mediated by the magnocellular (M) pathway in primate vision, but evidence is emerging that the parvocellular (P) pathway may also play a role in motion perception. Human peripheral vision is susceptible to anomalous motion perception because of spatial aliasing, and in this study we used this fact to determine if the P pathway can mediate information about low- and high-velocity stimuli. Psychometric functions relating visual performance to stimulus spatial frequency were measured for the directional discrimination of drifting sinusoidal gratings presented at 40 degrees eccentricity. Applying the sampling theorem to our results, we estimated that the Nyquist frequency of the limiting sampling array for directional discrimination is 1.7 cycles per degree. This result was compared with the Nyquist limit and spatial filtering properties of M and P ganglion cells in the human peripheral retina, calculated from histological data on their density and dendritic field size. Our results provide evidence to suggest that the reversed motion illusion in human peripheral vision is due to spatial aliasing by the P ganglion cell mosaic. We conclude that the sampling density of P ganglion cells limits veridical motion acuity in human peripheral vision, even for high-velocity targets. This provides further evidence that the P pathway is involved in processing information about motion.

Discrimination, Psychological↗