Date on Master's Thesis/Doctoral Dissertation

12-2004

Document Type

Doctoral Dissertation

Degree Name

Ph. D.

Department

Psychological and Brain Sciences

Committee Chair

Essock, Edward A.

Author's Keywords

Visual processing; Orientation; Broadband stimuli; Contrast normalization; Amplitude spectrum

Subject

Visual perception--Ability testing; Visual perception--Experiments

Abstract

Recently our lab has shown that with broadband stimuli (either visual noise or natural scenes), performance for detecting oriented content is worst at horizontal, best at the obliques, and intermediate at vertical orientations--an anisotropy (termed the "horizontal effect") quite different from the well-known "oblique effect" (worst performance obliques) obtained with simple line or grating stimuli. This horizontal effect can be explained by a proposed anisotropic normalization model that operates at the level of striate cortex by implementing the known numerical biases of striate neurons preferring different orientations as well as the strength of those responses from neurons tuned to similar orientations and spatial frequencies (with that strength being dependent on the spatial relationships between different scales and orientations present in the stimuli). To assess how the proposed striate normalization mechanism might operate when the visual system is presented with broadband stimuli containing different amounts of spatial frequency and orientation content, two suprathreshold matching experiments were conducted. Additionally, to provide an estimate of how broadband stimuli might modulate the weights of the proposed model, a series of neural response simulations were carried out on different types of broadband natural scene imagery. The stimuli for the psychophysical experiments were generated by making broadband isotropic visual noise patterns and filtering their amplitude spectra to contain a test increment across a specified range of orientations and spatial frequencies. The extent of the test increment's orientation and frequency bandwidth was systematically varied. A standard psychophysical matching paradigm was used to assess the perceived strength of the oriented structure in a test pattern relative to the oriented structure in a comparison pattern. The results yielded the traditional oblique effect when a fairly small range of orientations and high spatial frequencies were incremented and the horizontal effect was observed for broadband increments of about 20 degrees and 1-octave in frequency and larger. A blend of the two anisotropies was observed at intermediate increment bandwidth. The results of the psychophysical experiments were discussed in the context of the proposed striate normalization model with added insight from the results of the neural response simulations.

Download

Share

COinS