OBSERVERS SHOW DIFFERENTIAL SENSITIVITY TO CHANGES IN SPATIAL SCALE AND RELATIVE FREQUENCY IN A COMPLEX GRATING DISCRIMINATION TASK.
((I. Fine and R. A. Jacobs)) Center for Visual Science, University of Rochester, Meliora Hall, Rochester NY 14627-0268 USA
Purpose: Using complex
gratings, Olzak and Thomas (1997) showed that observers are more sensitive to
shifts in spatial scale of the underlying components (consistent with an object
changing distance or size) than they are to changes in relative frequency
(consistent with a change in object).
To explain these results, they proposed that the visual system sums
information over spatial frequency and orientation channels. However the identification task that they
used did not allow them to interleave trials in which shifts in spatial scale
and shifts in relative frequency occurred. Using discrimination tasks with
blocked as well as interleaved trials, we examined sensitivity to changes in
spatial frequency, and obtained results differing significantly from those
predicted by Olzak and Thomas. Methods: Observers discriminated
complex stimuli formed by summing two simple gratings drawn from disparate
regions of the Fourier plane. The
underlying components were of the same or orthogonal orientations. Observers made a forced-choice
discrimination decision based on small changes (10-20%) in the spatial
frequency of both the underlying gratings.
Task difficulty was governed using a QUEST procedure that modified the
spatial frequencies of the underlying gratings. Tasks with interleaved and with blocked trials were tested. Results
and Conclusions: When trials were interleaved observers were more
sensitive to frequency differences in the underlying components that produced a
change in relative frequency than they were to differences that produced a
shift in spatial scale when both components were of the same orientation. This effect disappeared when the trials were
blocked, consistent with the findings of Olzak and Thomas. These results suggest that sensitivity to
shifts in spatial scale or to changes in relative frequency is
task-dependent. Using interleaved
trials, observers seemed biased towards using cues consistent with a change in
object (changes in relative spatial frequency) rather than cues consistent with
changes in size or distance (shifts in spatial scale). Supported by NIH grant R29-MH54770. None.