Header logo is ei

Pedestal effects with periodic pulse trains

1999

Poster

ei


It is important to know for theoretical reasons how performance varies with stimulus contrast. But, for objects on CRT displays, retinal contrast is limited by the linear range of the display and the modulation transfer function of the eye. For example, with an 8 c/deg sinusoidal grating at 90% contrast, the contrast of the retinal image is barely 45%; more retinal contrast is required, however, to discriminate among theories of contrast discrimination (Wichmann, Henning and Ploghaus, 1998). The stimulus with the greatest contrast at any spatial-frequency component is a periodic pulse train which has 200% contrast at every harmonic. Such a waveform cannot, of course, be produced; the best we can do with our Mitsubishi display provides a contrast of 150% at an 8-c/deg fundamental thus producing a retinal image with about 75% contrast. The penalty of using this stimulus is that the 2nd harmonic of the retinal image also has high contrast (with an emmetropic eye, more than 60% of the contrast of the 8-c/deg fundamental ) and the mean luminance is not large (24.5 cd/m2 on our display). We have used standard 2-AFC experiments to measure the detectability of an 8-c/deg pulse train against the background of an identical pulse train of different contrasts. An unusually large improvement in detetectability was measured, the pedestal effect or "dipper," and the dipper was unusually broad. The implications of these results will be discussed.

Author(s): Henning, GB. and Wichmann, FA.
Journal: Perception
Volume: 28
Pages: S137
Year: 1999
Day: 0

Department(s): Empirical Inference
Bibtex Type: Poster (poster)

Digital: 0
Organization: Max-Planck-Gesellschaft
School: Biologische Kybernetik

BibTex

@poster{1149,
  title = {Pedestal effects with periodic pulse trains},
  author = {Henning, GB. and Wichmann, FA.},
  journal = {Perception},
  volume = {28},
  pages = {S137},
  organization = {Max-Planck-Gesellschaft},
  school = {Biologische Kybernetik},
  year = {1999}
}