Abstract
The process through which the outputs of V1 cells are combined at later stages in the form pathway remains largely unknown. As a step toward understanding this aspect of visual processing, we have shown the presence of mechanisms responding selectively to particular conjunctions of Gabor patterns (localised Fourier energy). To study these we used a compound adaptation technique, whereby the effect of adapting to a particular combination of Gabor elements is compared directly with equivalent adaptation to the same elements presented in isolation. When the Gabor elements fully overlap, to form a plaid, there is a greater contrast adaptation to this compound stimulus than predicted by equivalent adaptation to the components (Peirce & Taylor, 2006; Neuroscience 141:15 18). When the components abut but do not overlap, so as to form a concave contour, there is greater adaptation to the curvature of this contour than predicted by local tilt aftereffects (a curvature aftereffect, CAE) (Hancock & Peirce, 2008; Journal of Vision, 8(7):1 11).
Here, we aimed to determine whether the selective plaid adaptation effect results from a similar mechanism to the CAE by measuring the tuning of both effects to the spatial characteristics of the adaptors and probes. Both effects were clearly tuned to spatial frequency (SF) of the probe relative to the adaptor. However, when the SF of the component gratings differed relative to each other, the CAE was maintained, whereas plaid adaptation was markedly reduced for even moderate differences in the SF of the components.
We conclude that the effects are subserved by separate mechanisms, lying beyond the simple oriented filters of V1, yet before the shape-processing mechanisms that give rise to effects such as the shape frequency aftereffect (Gheorghiu and Kingdom, 2007; Vision Research 47:834 844) or radial frequency adaptation (Bell et al, 2008; Vision Research 48:2293 2301).
This work was supported by grants from the BBSRC (BB/C50289X/1) and The Wellcome Trust (WT085444).