Visual Motion Perception and Visual Information Processing

Abstract

This final progress report summaries the results of a study that successfully determined the functional architecture of visual motion perception, in the sense of defining the mechanisms involved and the relations between them. It was proved that visual motion is computed by two neural systems: primitive motion-energy extraction (e.g., Reichardt detector) and higher-level feature tracking. A psychophysical pedestal paradigm was used to exclude the feature- tracking process and thereby to obtain pure measures of motion-energy extraction. Motion energy extraction was found to be exclusively monocular, fast (cutoff frequency is 12 Hz) and sensitive (can utilize 0.2% contrast), 'bottom- up', and, to operate on both luminance (first-order) and contrast (second-order) motion stimuli. Motion feature tracking was found to operate interocularly as well as monocularly, have a cutoff frequency of 3 Hz, and to be both bottom up (it computes motion from luminance, contrast, depth, motion-motion, flicker and other type of stimuli) and top-down (e.g., attentional states influence what appears to move). The full report is appended.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Feb 28, 1994
Accession Number
ADA282928

Entities

People

  • George Sperling

Organizations

  • University of California, Irvine

Tags

DTIC Thesaurus Topics

  • Abstracts
  • Artificial Intelligence
  • Cognition
  • Cognitive Science
  • Data Displays
  • Detection
  • Detectors
  • Discrimination
  • Information Processing
  • Judgment
  • Motion Detectors
  • Pattern Recognition
  • Perception
  • Preprocessing
  • Psychology
  • Social Sciences
  • Stationary

Readers

  • Computer Vision.
  • Vision Science/Vision Psychology/Cognitive Neuroscience.