**2. Contrast Polarity**

The role of contrast polarity in human visual perception is a long-standing research topic (see, e.g., in [3–7]). As a consequence, the phenomena presented by Pinna and Conti are not really as novel or surprising as they suggested them to be, but they are indeed illustrative of the effects of, in particular, contrast polarity reversals. For example, Figure 1 shows, in the style of Pinna and Conti, a stimulus in which such a reversal triggers a substantial change in the way in which it is perceptually organized.

Depending on stimulus type, contrast polarity also affects visual regularity detection. For instance, the evident symmetry in the checkerboard pattern in Figure 2a is perceptually destroyed by the contrast polarity reversal in its right-hand half in Figure 2b (see, e.g., in [9]). In dot patterns, however, such a reversal does not seem to do much harm (see, e.g., in [10–12]). Furthermore, Figure 2c depicts a rotational Glass pattern with dipoles consisting of either two black dots or two white dots. It exhibits a moiré effect that is perceptually about as strong as when all dipoles consist of black dots (personal observation). However, when all dipoles are identical with one black dot and one white dot, as depicted in Figure 2d, the moiré effect disappears (see, e.g., [4,13,14]). The moiré effect does not disappear, by the way, when every dipole consists of differently shaped elements [15].

**Figure 1.** Contrast polarity and perceptual organization. Both compound stimuli can be described (i.e., interpreted) as consisting of, for instance, two triangles, two diabolos, or two parallelograms. The parallelogram interpretation is probably the weakest one in panel (**a**); but due to contrast polarity reversals, it is definitely the strongest one in panel (**b**) (after [8]).

**Figure 2.** Contrast polarity and regularity detection. The symmetry in the checkerboard pattern in panel (**a**) is perceptually destroyed by the contrast polarity reversal in its right-hand half in panel (**b**); The Glass pattern in panel (**c**), with dipoles consisting of either two black or two white dots, exhibits a clear moiré effect, which disappears in panel (**d**), where all dipoles are identical with one black dot and one white dot.

Therefore, I agree with Pinna and Conti that contrast polarity is a factor to reckon with. However, I think that, in their first claim, they concluded too easily—in fact, based on incorrect assumptions—that contrast polarity triggers local groupings that precede the global groupings allegedly predicted by simplicity and likelihood approaches. Next, this is discussed in more detail.
