|
Perceptual Learning for Discriminating Complex Gratings ((I.
Fine and R.A. Jacobs)) University of California, San Diego; University of
Rochester Purpose: The visual images of real objects contain a wide range of spatial frequencies and orientations. We used a discrimination task with complex gratings to examine what mechanisms underlie improvements in the ability of observers to combine information across wide ranges of spatial frequency and orientation with practice. Methods: Stimuli contained two orthogonal signal components of 3 and 9 c/deg. Observers discriminated a 15% spatial frequency shift in these components. The stimulus also contained four noise components, separated from the signal components by at least 45 degrees of orientation or ~2 octaves in spatial frequency. Naive observers were trained with a 4-alternative same-different forced choice judgment with feedback (Exp 1). We also examined perceptual learning without the noise components (Exp 2), transfer of learning across orientation (Exp 3), how performance was impaired by spatial frequency shifts in the noise components (Exp 4), and transfer of learning from a same-sign task where the two signal components both increased/decreased in spatial frequency to an opposite-sign task where both signal components shifted in opposite directions in frequency space (Olzak and Thomas, 1997) (Exp 5). Results and Conclusions: Observers showed significant learning: thresholds dropped to ~1/3 of their original value over 8 sessions (Exp 1). Observers showed far less learning when the noise components were not present (Exp 2). In addition, unlike other studies, we found almost complete transfer of learning across orientation (Exp 3). These last two findings suggest that most learning occurs at a mid-level processing stage as opposed to low-level analyzers tuned for spatial frequency and orientation. Performance was more severely impaired by shifts in noise components of the same spatial frequency or orientation as the signal components (though there was significant variability between observers), suggesting that after training observers based their responses on mechanisms tuned for selective regions of Fourier space (Exp 4). Transfer of learning from a same to an opposite-sign task (Exp 5) was complete: even after practice observers continued to combine information from the two signal components independently. Support: NIH grant R29-MH54770. None. |