Abstract
Some models of adaptation assume that aftereffects represent a subtractive inhibition of a sensory signal that improves the dynamic range of sensory coding (Barlow & Földiék, 1989). In contrast, texture density aftereffects have been described as multiplicative, rather than additive, changes in perceived density (Durgin, 1995). Here we use precise quantification of one contingent density aftereffect to suggest that, rather than being additive or multiplicative with respect to density, these effects are additive with respect to perceived inter-dot distance. Texture displays were presented in gray boxes to the left and right of fixation, with a frame surrounding the boxes. Subjects were adapted to a correlation between surround color and relative density (e.g., dense on right when frame was green and on left when red) and then tested for frame-color contingent adaptation at several levels of density using a staircase procedure. Gaze was tracked to prevent artifacts of refixation. Points of subjective equality (PSE) were computed for each standard density and frame color. If the aftereffects resulted from a simple subtraction, then the difference between PSE and standard density should have been constant for all levels of density. Alternatively, if the aftereffects involved multiplicative changes (as if operating on log density), then the deviations should have been constant across standard density in logarithmic space. In our data, however, both linear and logarithmic deviations increased with standard density. A better fit for our data was provided by a subtractive model applied to average inter-dot distance. Only in this model were the deviations constant across standard density. Thus, it appears that the frame-contingent density aftereffect (and perhaps density aftereffects more generally) may operate on neural codes representing linear spacing rather than density itself. Density aftereffects may be a form of tuning to the locally prevailing spatial scale.