TEMPORAL CHARACTERISTICS OF FAST CONTRAST ADAPTATION IN HIGH LEVEL OBJECT IDENTIFICATION TASKS
((J. Fiser and I. Fine)) Dept. of Brain and Cognitive Sci., University of Rochester, 14627; Dept. of Psychology, University of California, San Diego, 92093.
Purpose: To compare the time course of fast contrast adaptation in grating and object identification tasks. Methods: Gray-scale images of everyday objects were contrast normalized using histogram equalization based on square wave histograms to low (29.25-39.75 cd/m2) or high (13-56 cd/m2) luminance contrast. An RSVP paradigm was used each trial consisted of either 72 gratings or object images presented sequentially at a rate of 50 msec/image. With object sequences, observers had to report whether a target object, verbally specified before each trial, had been presented. With grating sequences, subjects had to report whether a 0.45 cpd horizontal grating appeared among distracter gratings of random spatial frequency and phase but orientated at least 45˚ from horizontal. Grating or object sequences were modulated between low and high contrast according to a temporal square wave varying from 0.4 to 3.6 Hz. In each trial a single object or grating image (the target in half of the trials) was of the opposite contrast than the surrounding images. Results: In the case of grating sequences, identification of a high contrast grating among low contrast distracters was significantly better than identification of a low contrast grating among high contrast distracters, and was independent of the frequency of the temporal contrast modulation. Performance with object sequences was identical to that with gratings at the lowest temporal frequency. But as the temporal frequency increased to 1.2 Hz, differences in performance between high and low contrast targets gradually disappeared. With even higher temporal frequencies differences between high and low contrast targets reappeared, ruling out simple explanations based on masking by images closely preceding or following the target image. Conclusions: Identifying simple suprathreshold gratings does not require complete contrast adaptation. In comparison, complex visual tasks require observers to be more fully adapted to contrast. This adaptation requires more than 300 msec, and when it is incomplete performance is equally hindered for both low and high contrast images.
None. Support: JSMF 96-32 and
NIH grant R29-MH54770