Purchase this article with an account.
Yeon Jin Kim, Andrew M. Haun, Edward A. Essock; The effect of sustained/transient temporal modulation on the horizontal effect of contrast masking. Journal of Vision 2008;8(6):276. doi: https://doi.org/10.1167/8.6.276.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We have shown previously that contrast masking by broadband noise is greatest for horizontal orientations and least for vertical orientations (Kim, Haun and Essock, VSS 2007), which fits well with the earlier finding of greater salience and sensitivity for oblique content (and least salience for horizontal) in broadband patterns (Essock et al, Vis. Res., 2003). Here we examine this “horizontal effect” of broadband contrast masking with respect to sustained and transient temporal presentations. Contrast thresholds for sinewave gratings of either 1 or 8cpd were measured in the presence of oriented broadband (1/f) noise masks. The temporal presentation of the test grating and of the mask was either ‘transient’ or ‘sustained’ in nature. With no masking, threshold is poorest at oblique orientations and lowest at cardinal orientations (the “Class 1 oblique effect”). With a broadband mask present we find that when a brief (25ms squarewave pulse) or flickered (16Hz) test is combined with a brief or flickered mask, a horizontal effect of contrast sensitivity is obtained. When a sustained (560 ms, gradual ramp) test is combined with a sustained mask, we also obtain a horizontal effect. However, when test and mask have distinct temporal waveforms, an oblique effect is obtained along with a loss of masking. Since masking is typically accounted for in terms of a pooled response from other filters that is used to normalize output, these results suggest that there are separate normalization pools for the sustained and transient temporal conditions and that both are anisotropic (and in the pattern of a horizontal effect). It appears that in a divisive normalization model, only filters with temporal properties similar to that of the ‘test’ filter contribute to its normalization pool.
This PDF is available to Subscribers Only