Implications of the Craik-O'Brien illusion for brightness perception.

Measurements are reported for detection thresholds of high and low-pass filtered squarewaves, and for brightness matches of those waveforms. The threshold measurements agree closely with those of Campbell, Howell and Johnstone [J. Physiol., Lond. 284, 193-201 (1978)]: contrasts at which a high-pass squarewave was indistinguishable from an unfiltered squarewave could be well predicted from detection thresholds for an appropriate low-pass squarewave. However, the brightness of high-pass squarewaves (the "Craik-O'Brien illusion") was not related to the threshold measurements. Brightness was virtually constant with spatial frequency, even at spatial frequencies 10 times higher than the region of the low frequency cut. Brightness depended strongly on contrast, being relatively greater at low than at high contrasts. The results can be well accounted for by a recent theory of edge detection, and the existence of parallel channels in vision.