Abstract:What determines what we see? In contrast to the traditional "modular" understanding of perception, according to which visual processing is encapsulated from higher-level cognition, a tidal wave of recent research alleges that states such as beliefs, desires, emotions, motivations, intentions, and linguistic representations exert direct, top-down influences on what we see. There is a growing consensus that such effects are ubiquitous, and that the distinction between perception and cognition may itself be unsustainable. We argue otherwise: None of these hundreds of studies -either individually or collectively -provides compelling evidence for true top-down effects on perception, or "cognitive penetrability." In particular, and despite their variety, we suggest that these studies all fall prey to only a handful of pitfalls. And whereas abstract theoretical challenges have failed to resolve this debate in the past, our presentation of these pitfalls is empirically anchored: In each case, we show not only how certain studies could be susceptible to the pitfall (in principle), but also how several alleged top-down effects actually are explained by the pitfall (in practice). Moreover, these pitfalls are perfectly general, with each applying to dozens of other top-down effects. We conclude by extracting the lessons provided by these pitfalls into a checklist that future work could use to convincingly demonstrate top-down effects on visual perception. The discovery of substantive top-down effects of cognition on perception would revolutionize our understanding of how the mind is organized; but without addressing these pitfalls, no such empirical report will license such exciting conclusions.

A chief goal of perception is to help us navigate our environment. According to a rich and ambitious theory of spatial perception, the visual system achieves this goal not by aiming to accurately depict the external world, but instead by actively distorting the environment’s perceived spatial layout to bias action selection toward favorable outcomes. Scores of experimental results have supported this view—including, famously, a report that wearing a heavy backpack makes hills look steeper. This perspective portrays the visual system as unapologetically paternalistic: Backpacks make hills harder to climb, so vision steepens them to discourage ascent. The “paternalistic” theory of spatial perception has, understandably, attracted controversy; if true, it would radically revise our understanding of how and why we see. Here, this view is subjected to a kind and degree of scrutiny it has yet to face. After characterizing and motivating the case for paternalistic vision, I expose several unexplored defects in its theoretical framework, arguing that extant accounts of how and why spatial perception is ability-sensitive are deeply problematic and that perceptual phenomenology belies the view’s claims. The paternalistic account of spatial perception not only isn’t true—it couldn’t be true, even if its empirical findings were accepted at face value.

A tidal wave of recent research purports to have discovered that higher-level states such as moods, action capabilities, and categorical knowledge can literally and directly affect how things look. Are these truly effects on perception, or might some instead reflect influences on judgment, memory, or response bias? Here, we exploited an infamous art-historical reasoning error (the so-called "El Greco fallacy") to demonstrate that multiple alleged top-down effects (including effects of morality on lightness perception and effects of action capabilities on spatial perception) cannot truly be effects on perception. We suggest that this error may also contaminate several other varieties of top-down effects and that this discovery has implications for debates over the continuity (or lack thereof) of perception and cognition.

Does the human mind resemble the machine-learning systems that mirror its performance? Convolutional neural networks (CNNs) have achieved human-level benchmarks in classifying novel images. These advances support technologies such as autonomous vehicles and machine diagnosis; but beyond this, they serve as candidate models for human vision itself. However, unlike humans, CNNs are “fooled” by adversarial examples—nonsense patterns that machines recognize as familiar objects, or seemingly irrelevant image perturbations that nevertheless alter the machine’s classification. Such bizarre behaviors challenge the promise of these new advances; but do human and machine judgments fundamentally diverge? Here, we show that human and machine classification of adversarial images are robustly related: In 8 experiments on 5 prominent and diverse adversarial imagesets, human subjects correctly anticipated the machine’s preferred label over relevant foils—even for images described as “totally unrecognizable to human eyes”. Human intuition may be a surprisingly reliable guide to machine (mis)classification—with consequences for minds and machines alike.

No abstract