Illusory motion reversal does not imply discrete processing: Reply to Rojas et al.

Illusory motion reversal does not imply discrete processing: Reply to Rojas et al. Under certain conditions a smoothly moving pattern viewed under continuous light is sporadically perceived to The existence of this surprising illusion has been taken as evidence that the visual system takes discrete snapshots of the visual scene In our previous work we have provided evidence against this supposition (Kline, Holcombe, & Eagleman, 2004). Our key observation was that two identical rotating drums do not appear to reverse simultaneously , although simultaneous reversal is predicted by the hypothesis of discrete sampling of the scene. As a result, we proposed that illusory motion reversal results from rivalry between motion detectors tuned for the direction of the moving pattern in competition with spuriously activated detectors tuned for the opposite direction (Kline et al., 2004). In an effort to strike a middle ground between the snapshot theory of perception and our proposal of rivalry among Reichardt-like motion detectors (RLMDs), Rojas, Carmona-Fontaine, Ló pez-Calderón, and Aboitiz (2006) suggest that if RLMDs are themselves discrete samplers, then the snapshot hypothesis is still valid—but at the level of the individual motion detector rather than the whole visual field (Rojas et al., 2006). In other words, Rojas et al. attempt to salvage a theory of temporally quantized visual processing by suggesting it may happen at an extremely local level. The main problem with RojasÕ claim is that it is based on the idea that RLMDs are temporally discrete, and therefore subject to temporal aliasing. However, the motion detector model proposed by Reichardt (1961) is not temporally discrete and hence is not susceptible to temporal alias-ing. A subunit of the RLMD receives asymmetrically delayed input from two visual receptive fields, such that sequential stimulation of the receptive fields (e.g., by a stimulus moving in the preferred direction) will result in the excitation of the motion detector. This produces a response that is continuous in time. RLMDs can be susceptible to spatial aliasing if they are not preceded by the appropriate pre-filters that Reichardt (1961) also proposed. That is, a periodic pattern moving in the anti-preferred direction may spuriously activate the detector (see Fig. 1 in Kline et al., 2004). This type of aliasing, which was proposed by Kline et al. (2004) to underlie illusory motion reversal, is very different from temporally-discrete processing , both at the level of the motion detector as described by Rojas et …