Ahumada & Beard (Perception 26, ECVP Suppl., 34) calibrated an image sequence discrimination model to masking conditions where the target and masker were the same spatially. Masking was based on local contrast energy so that compound grating masking is greater than that of either component alone, and masking is nearly independent of mask frequency.

Detection thresholds were measured for a 12 msec target in the presence of a 700 msec mask. Targets were 1, 3 or 9 c/deg sine wave gratings. Masks were one grating or two gratings combined in peaks-subtract phase. The target was presented in 17 temporal positions, before, during and after the mask. Peak masking was found near mask onset and offset for 1 and 3 c/deg targets, while masking was more uniform during the mask for the 9 c/deg target. In general, combination maskers were not significantly more effective than the more effective single component mask. Also there was much less masking when the mask frequency differed from that of the target.

By altering the model so that much of the masking is caused by a point nonlinearity rather than local contrast energy, the model can predict the results without the complexity of spatial frequency channels.

This image sequence discrimination model features different masking rules in its sustained and transient systems.

The model differs from that in Ahumada, Beard, and Eriksson (1998 SPIE/IS&T) only in that the combination rule for the transient and sustained systems is additive in difference contrast energy.

Here we compare predictions of this model for grating masking experiments where the test spatial frequency can be the same or different from that of the mask.

 Test stimulus:
    Sinusoidal grating
    Duration: 12 msec (single frame)
    Spatial frequency: 1, 3, or 9 c/deg
    Sine phase
    Circular window diameter: 5 deg
 Mask stimulus:
    Single or compound grating
    Component contrast amplitude: 15%
    Duration: 700 msec
    Abrupt onset and offset
    Spatial frequency: 1, 3, or 9 c/deg
    Sine phase
    Circular window diameter: 5 deg
 Mean luminance: 4.2 cd/m2
 16 SOA values: -300 to 1200 msec
 3 observers (only BLB data shown)
 2IFC staircase

The following graphs show the measured thresholds and the model predictions using the following model parameters.

 spatial blur spreads, arc min
                 cone,  1.7
           horizontal,  8.5
                magno,  4.8
 magno masking energy, 25.

    integration times, msec
                 cone,  4.8
           horizontal, 32.
                parvo, 27.
 magno masking energy, 27.

 masking thresholds, % contrast
      parvo masking, 8.
      magno masking, 0.8


contrast energy thresholds, deg^2 sec
                     parvo, 0.008
                     magno, 0.0007

Model can predict masking by different frequencies without cortical filters.

Model cannot predict masking which could be duration discrimination.

Foley's result that masking orientation bandwidth is narrower for 500 msec targets than for 100 msec targets is predicted from increased spatio-temporal masking spread in the transient system.

Perceived brightness is spatially integrated at a transient system scale and then applied to sustained system spatial details. The same may be true of perceived contrast.