Opposing views on orthogonal adaptation: a response to Clifford, Arnold, Smith, and Pianta (2003)

Clifford et al., 2003 point out that there is indeed a subtle difference between their experiment and ours (Westheimer & Gee, 2002). Because it is generally better to provide the observer with an explicit standard when measuring difference thresholds we had interposed a brief presentation of the comparison before showing the test pattern for which the orientation judgment had to be made. Clifford et al. now question whether this could undo the adapting effect of orthogonal gratings. Accordingly we modified the procedure used in our original experiments by blanking out the frames which at that time carried the comparison information; now the orthogonal adaptation was interrupted only to show the test pattern, suitably buffered by a blank screen. 1 The results for two observers are given in Table 1. With a foveal line stimulus, adaptation to a set of orthogonal stripes still does not change orientation discrimination thresholds. For the pattern duplicating Clifford et al. s, where two peripheral circular grating patches are subjected to orthogonal adaptation with similar parameters, one of our observers again had no adaptation effect. The other had a 15% reduction in threshold in the adapted compared to the non-adapted situation, just missing the 5% level of significance. This result is based on five sessions, spread over three experimental days always with properly counterbalanced sequences of adaptated/nonadapted conditions. In the original experiment with this pattern, this observer had an adapted threshold 4% higher than the one without adaptation. Thus, in this case, the presence of a comparison stimulus could have gone same way towards blocking the ability of orthogonal adaptation to improve orientation discrimination thresholds. However, this effect is not manifested by all observers and certainly not for all patterns. If it turns out that a brief presentation of a test pattern nulls out some adaptation phenomena, it is interesting to speculate on the origin of such an effect. One possibility is that the adapting pattern, although it had already been extinguished, provided some sort of an orientation framework that becomes diluted when there is an intercalated reference. Another possibility would involve transient changes in the orientation tuning of cortical neurons immediately subsequent to some specific stimuli, a phenomenon that has been demonstrated in a related context in the cat visual cortex (Yao & Dan, 2001). The challenge of Clifford et al. s original aim remains: the quest for a neural substrate for the opponency mechanism underlying the tilt illusion s reversal when the inducing pattern approaches orthogonality. The patterns used in this research all evoke both a direct and indirect tilt illusion. The predominant absence of an influence of orthogonal adaptation implies that orientation discrimination thresholds are not tightly enough coupled to the apparatus generating the mean orientation shifts to assign them a secure role as an analytical tool in this quest. Corresponding author. Tel.: +1-510-642-4828; fax: +1-510-6436791. E-mail address: [email protected] (G. Westheimer). 1 This modification in the procedure is feasible only for orientation discrimination thresholds in the vertical or horizontal meridians, where a comparison with a standard can be dispensed with because observers have a strong internal representation of the reference orientation. The extension of the experiment to oblique meridians, as was done in obtaining the results in Fig. 5 of our original communication, would not be possible without providing the observer an explicit reference.