Return to search

Non-invasive dissection of the human visual system

A flickering stimulus can appear brighter or of different colour than a steady stimulus of the same spectral composition and equal time-averaged intensity. The change in appearance is consistent with the distortion product of a nonlinearity in the human visual system. This nonlinearity is used for non-invasive dissection of the human visual system into prenonlinearity and post-nonlinearity linear stages, which affect the amplitude of the sinusoidal signals that they process. A five-channel Maxwellian system was used to generate cosine-windowed, amplitude-modulated, sinusoidally-flickering stimuli. The subjects adjusted the maximum modulation of the signal in order to set the threshold for the perception of flicker and colour or brightness change. The pre-nonlinearity filter was tested by varying the carrier frequency between 5 Hz and 60 Hz and the post-nonlinearity filter was measured by varying the amplitude-modulation frequency between 0.25 Hz and 5 Hz. The L- and M-cone pathways are separately measured by using a new combination of the silent-cone-substitution technique and the sandwich model. The distortion product at the output of the nonlinearity is measured by using a new matching method for measuring the colour and brightness changes. The results suggest that the pre-nonlinearity filter is band-pass and shows substantial adaptation with light intensity that can explain most of the adaptation in the visual pathways. The post-nonlinearity filter is low-pass and shows little or no adaptation with light intensity. The input-output function of the nonlinearity can be described as an expansive, quadratic function that rapidly saturates at high input levels.

Identiferoai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:565296
Date January 2011
CreatorsPetrova, D.
PublisherUniversity College London (University of London)
Source SetsEthos UK
Detected LanguageEnglish
TypeElectronic Thesis or Dissertation
Sourcehttp://discovery.ucl.ac.uk/1310438/

Page generated in 0.0019 seconds