Firestone & Scholl (F&S) make a strong case against the effect of top-down beliefs on perception. The argument for the cognitive impenetrability and modular nature of (visual) perception is reminiscent of the historic debate between Fodor and Turvey (Fodor & Pylyshyn Reference Fodor and Pylyshyn1981; Turvey et al. Reference Turvey, Shaw, Reed and Mace1981), especially when most of the top-down modulation literature can find its roots in Gibson's (Reference Gibson1966; Reference Gibson1979) ecological psychology. However, several aspects make F&S's article theoretically unique and important: (1) taking a logical approach such as with the El Greco fallacy, (2) speaking to a wider range of researchers beyond the action or embodied cognition literature, and (3) perhaps most important, providing a checklist of criteria for future studies against the six pitfalls they have cogently identified.

One common pitfall among studies that mistake top-down effects in judgment for perceptual effect is the use of subjective report in measuring percepts (e.g., light and darkness, reachability, distance). Not only is subjective report highly susceptible to task demand (e.g., Durgin et al. Reference Durgin, DeWald, Lechich, Li and Ontiveros2011a), but also it is problematic because it provides no additional information that would enable researchers to trace the source of the top-down effect. Accordingly, in order to dissociate perception and judgment, it is advisable to use performance-based measures that supply additional information (e.g., spatial, temporal), thereby making it possible to infer the stage of processing over which top-down cognition exerts its influence. One of our previous studies (Tseng & Bridgeman Reference Tseng and Bridgeman2011) demonstrates this point: To test whether hands near a visual stimulus would enhance processing of the stimulus (as opposed to hands far; see Tseng et al. Reference Tseng, Bridgeman and Juan2012 for a review), participants performed a forced-choice visual memory change detection task that provides accuracy and reaction time data, as opposed to subjective report. The rationale was that if hand proximity could really change the way visual stimuli are processed in a positive way, then hand proximity would predict enhanced visual processing, which would lead to better change detection performance. This would effectively rule out the judgment component; the participants cannot fake better performance.

Here it is important to clarify F&S's conceptual distinction between “perception and judgment”: The two are not mutually exclusive, nor do they exhaust all alternatives (e.g., attention). Therefore, even if the judgment factor is accounted for by performance-based measures, such a result would not necessarily guarantee an effect in perception, especially because the effects on attention – which can modulate perception – can often disguise themselves as effects on perception (F&S's “periphery effect of attention”). To revisit the example above, although it is tempting to conclude that perception was directly modulated by hand proximity, it is equally plausible that the effect stemmed from biased attention near the hands. Indeed, analyzing participants' hit rates region by region on the screen showed a shift of correct responses toward the right-hand side, suggesting that the effect was mediated by biased spatial attention, not visual perception. This conclusion not only reemphasizes the importance of having a performance-based measure that can be analyzed differently to provide additional information, but also it is consistent with Pitfall 5, “peripheral attentional effects” (sect. 4.5), on F&S's checklist. The same rationale is also true for Pitfall 6, “memory and recognition” (sect. 4.6), and we attacked this problem by turning a potential artifact into an independent variable. Throwing a marble into a hole makes the thrower judge the hole as bigger following success than failure, but only if the hole is obscured after throwing. If the hole remains visible, the effect disappears (Blaesi & Bridgeman Reference Blaesi and Bridgeman2015; Cooper et al. Reference Cooper, Sterling, Bacon and Bridgeman2012). The logic of this experiment is analogous to many efforts to demonstrate effects of action on perception, and it shows those results to affect memory, not perception. Modifying memory on the basis of experience is useful; modifying perception is not. Taken together, we recommend that future studies should consider Pitfalls 2, 5, and 6 together by controlling for judgment and memory effects and then moving on to tease apart the effects in perception versus attention.

Lastly, it is intriguing to us that a majority of the studies reporting top-down effects on perception are related to action (e.g., affordance, reachability). Might action studies be more susceptible to misattributing attentional or memory effects to perception? We speculate four possible reasons unique to the action literature for why this may be the case:

1. 1. Universality: Due to motor action's depth in evolutionary time, action's effects on perception or attention are likely very widespread. This differs from the way in which ruminating about things, such as a sordid past, would make the room seem darker (e.g., Banerjee et al. Reference Banerjee, Chatterjee and Sinha2012; Meier et al. Reference Meier, Robinson, Crawford and Ahlvers2007). Because ruminations about the past involve parts of the cognitive economy that are evolutionarily recent, and because darkness metaphors of this type depend largely upon cultural interpretations that might be unique to humans, effects on perception are not as likely as actions and affordances.

2. 2. Implicitness: Unlike certain, consciously accessible, top-down beliefs, information regarding action possibilities, or affordances, is often implicit properties that subjects may not be consciously aware of. Thus, the implicit nature of affordance information is assumed to be processed below consciousness threshold, and likely at the perceptual stage.

3. 3. Well-established neurophysiology: The neuronal mechanisms for processing affordance or other action-relevant information (e.g., space, distance, graspability) have been well investigated in monkeys (e.g., Graziano & Botvinick Reference Graziano, Botvinick, Prinz and Hommel2002). Visual–tactile neurons in premotor and parietal cortices move their receptive fields with the hands instead of eyes, and they respond to objects that are within reach, even when “reachable” means “reachable with a tool.”

4. 4. Perception–action loop: The idea of perception–action coupling has been important in ecological psychology, and still is today in the embodied cognition literature. We suspect an overly literal interpretation of the idea can sometimes mislead researchers to mistake attentional effects as perceptual.

In summary, the effect of action on perception or attention is clearly quite different from other types of top-down beliefs. Although it is unfortunate that most action studies have mistaken attentional effects as perceptual, one can at least see why these studies may be more vulnerable to an inclination towards perceptual interpretations. Therefore, we recommend researchers in the field of perception and action and embodied cognition to especially consider F&S's arguments in the context of action when making conclusions.