An Orientation Dependent Size Illusion Is Underpinned by Processing in the Extrastriate Visual Area, LO1

Abstract

We use the simple, but prominent Helmholtz's squares illusion in which a vertically striped square appears wider than a horizontally striped square of identical physical dimensions to determine whether functional magnetic resonance imaging (fMRI) BOLD responses in V1 underpin illusions of size. We report that these simple stimuli which differ in only one parameter, orientation, to which V1 neurons are highly selective elicited activity in V1 that followed their physical, not perceived size. To further probe the role of V1 in the illusion and investigate plausible extrastriate visual areas responsible for eliciting the Helmholtz squares illusion, we performed a follow-up transcranial magnetic stimulation (TMS) experiment in which we compared perceptual judgments about the aspect ratio of perceptually identical Helmholtz squares when no TMS was applied against selective stimulation of V1, LO1, or LO2. In agreement with fMRI results, we report that TMS of area V1 does not compromise the strength of the illusion. Only stimulation of area LO1, and not LO2, compromised significantly the strength of the illusion, consistent with previous research that LO1 plays a role in the processing of orientation information. These results demonstrate the involvement of a specific extrastriate area in an illusory percept of size.

Keywords: TMS; fMRI; perception; spatial vision.

Figure 1.

The Helmholtz squares illusion and the BOLD responses it induces in primary visual cortex. (a) On the left, a vertically lined square appears above a physically matched, horizontally lined square—note the lower square appears narrower than the upper square as first described by Helmholtz (Helmholtz & Southall, 1925). On the right, the horizontally lined stimulus has been widened so it is no longer physically square, but now appears to match the dimensions of the vertically lined square above it. The width of the perceptually matched horizontally lined stimulus was increased (by 8–18%) according to each individual’s psychophysics results. (b) BOLD response phase to rotating checkerboard wedges superimposed in false color on an inflated representation of the occipital lobe. Data have been restricted to V1. BOLD responses to squares and rectangles in (a) were analyzed over a line ROI, illustrated as a dark blue outline. This ROI extended along the representation of the horizontal meridian, where the inner and outer stimulus edges are represented. (c) The response (t-statistic) to all stimuli compared against a uniform gray background. (d) The BOLD response profile (t-statistic) as a function of cortical distance along the ROI, with negative t-values set to zero. The measures, d, for different stimulus conditions were computed by taking the width of the function at 25% of its peak value and were used to compute a contrast measure D for each hemisphere. Note. ROI = region of interest.

Figure 2.

Predictions and results for BOLD responses elicited in primary visual cortex by the Helmholtz squares illusion. (a) The pattern of results based on the prediction that the extent of BOLD responses in V1 will reflect the perceived (top) or physical (bottom) width of the stimulus. The horizontal axis is labeled with the lined-stimuli that elicited the extent of BOLD responses d₁ (top row) and d₂ (bottom row), which were used to compute the contrast measure, Blue arrows indicate physically wider squares. (b) Individual data (circles) (c) Group mean data (bars) with error bars indicating standard error of the mean. The results fit with the prediction that V1 maps physical, not perceived width.

Figure 3.

(a) Overview of experimental procedure. (b) Individual data (circles). (b) Group mean data (bars). (b) and (c) data are plotted in terms of proportion correct responses: 0.50 indicates participant guessing as expected for PSE-matched, perceptually square stimuli; 1.00 indicates correct reporting of the physically elongated stimulus dimension on all trials. Results for horizontally and vertically striped stimuli are pooled together, except in the case of S2 and S5 (see Procedure section for the TMS study examining the causal roles of V1, LO1, and LO2 in size perception). Error bars denote ± SEM. Asterisks denote a significant difference between conditions (*p < .05, **p < .01; Bonferroni-corrected).

Figure 4.

Plots of potentially confounding variables for the PSE-matched Helmholtz stimuli. (a) Reaction time. (b) Coil-target distance. (c) Coil-target offset. Error bars represent the standard error of the mean. No significant differences were found between any of the conditions reported here.