Counterpoint

Schnall (2017, this issue) argues that theories of attribution can help explain why some labs report data that seem inconsistent with the energetics hypothesis of space perception (which supposes that changes in physiological state alter perceived slant). I argue that Schall’s position not only neglects the important role of control experiments that seek to test alternative hypotheses but that it also trivializes the energetics hypothesis by conflating explicit judgment with perceptual experience. Energetic considerations must affect choices, but they probably contribute directly rather than by affecting the perception of spatial layout.

Why do hills look so steep (Durgin & Li, 2013; Kammann, 1967; Ross, 1974)? A hill of 5° is normally perceived to be about 20°, even measured implicitly (Li & Durgin, 2010, 2013). Proffitt, Bhalla, Gossweiler, and Midgett (1995) proposed that the perceptual exaggeration of hills is the result of energetic considerations being embedded in perception so as to affect decisions about navigation. In contrast, Li and Durgin (2010) have proposed that perceptual slant exaggerations, which are experienced underfoot even by the congenitally blind (Hajnal, Abdul-Malak, & Durgin, 2011), represent perceptual scale expansion that contributes enhanced sensitivity to the immediate control of action (Durgin, 2014; Durgin & Li, 2011).

These two theories differ in how malleable they expect perception to be. The scale expansion hypothesis depends on a predictable (exaggerated) coding of slant so that action can be calibrated. The energetics hypothesis argues that changes in physiological state can immediately alter perceived slant, as illustrated by Bhalla and Proffitt’s (1999) famous study using heavy backpacks. So, do backpacks affect how steep things look?

Durgin et al. (2009) reported a novel control condition in a backpack study. The control showed that carrying a heavy backpack full of scientific equipment was insufficient to produce changes in estimated slant, whereas participant beliefs that the experimenter expected the heavy weight to affect estimates seemed to matter ¹ —a finding replicated later using multiple variations to control for various alternative hypotheses (e.g., Durgin, Klein, Spiegel, Strawser, & Williams, 2012; Durgin, Ruff, & Russell, 2012; Shaffer, McManama, Swank, & Durgin, 2013). These results are problematic for the energetics account of perceptual experience (Firestone, 2013).

Moreover, Shaffer et al. (2013) showed that when participants were insightful about the energetics hypothesis but misconstrued the experimental manipulation, they made the complete opposite judgments than those predicted by energetics. Specifically, participants who arrived in the lab after fasting and were administered a sweetened drink that did not contain sugar both (a) typically assumed that the drink had contained sugar (a misconstrual) and (b) sometimes (25%) believed that the sugar was supposed to affect their estimates of slant ² (were insightful despite an elaborate cover story). This insightful group of misconstruers gave lower estimates than everyone else.

Schnall suggests we can’t tell why people made these lowered judgment, but the insightful participants affected in this experiment were precisely the ones mistaken about the condition they were in (due to their misconstrual). If they had resisted cooperation (as Schnall proposes), they would have given higher estimates of the hill, but they didn’t. Their lower estimates, which cannot be predicted by energetics (they hadn’t been given sugar), nor by anti-cooperation (they thought they had been given sugar), are thus either a sign of cooperation with their insightful beliefs about the energetics theory (given the misconstrual of what condition they were in) or something else. Crucially, these estimation biases only occurred for those with low blood sugar (not for insightful participants who had actually received sugar), consistent with the previously hypothesized role of (low) sugar in increasing likelihood of cooperation with experimental demand (e.g., Durgin, Hajnal, Li, Tonge, & Stigliani, 2010, 2011).

Schnall mentions our subsequent work (Williams, Ciborowski, & Durgin, 2012) on attribution effects: Participants (all in a state of low blood sugar) who drank diet ginger ale ³ poured from a nondiet bottle behaved differently than those who drank the diet ginger ale from a diet bottle. The misconstrual condition led to judgmental bias: They gave higher estimates of the slant of stairs and gave higher estimates of the number of Stroop trials they had done between having the drink and making the estimate. Many studies of low blood sugar and cognition may inadvertently confound low blood sugar (the intended manipulation) with misconstrual by participants (who tend to assume that drinks they are given contain sugar). Construals matter, and they can sometimes be controlled implicitly.

But Schnall argues that attribution (when consistent with the energetics hypothesis) actually affects perception and points out that there are established theories that can predict why control experiments might conceal perceptual attributional differences that are revealed by prior experiments. This argument seems problematic for at least two reasons.

First, Schnall seems to have some predictions backward. Schnall discusses primarily the Durgin, Klein, et al. (2012) instructional manipulation, but Schnall’s conclusion requires that the use of a deceptive cover story in our many other studies also disrupts the normal attributional processes by making the backpack salient. (By Schnall’s hypothesis, this would allow people not to be affected by the backpack.) The weakness of Schnall’s framing, however, is that the original experimental procedure of Bhalla and Proffitt (1999)—asking people to wear a heavy backpack to estimate its weight, and then leaving it on—is already abnormal in comparison with the typical human purpose for wearing backpacks (to carry things). So how do you get people to wear a heavy backpack without calling attention to it? If people normally wear backpacks in order to carry things, then being asked to wear a backpack to carry equipment—as the various deceptions we have used typically do (e.g., Durgin et al., 2009; Durgin, Ruff, & Russell, 2012; Shaffer et al., 2013)—seems much more plausibly generalizable to the normal use of backpacks. Our three deception manipulations lend force to the conclusion of Durgin, Klein, et al. (2012) that backpacks, when explicitly worn for the purpose of carrying things, typically have no effect on estimates of slant, let alone perception.

The second fundamental problem with Schnall’s argument is that suggesting that perceptual experience is the same as attribution trivializes the energetics hypothesis. Should we be interested in this work if it is just about judgment? Isn’t it the purported effect on perception that made the theory interesting? Conflating (judgmental) attribution effects with perception has been a common artifact of the energetics approach, and this is why many psychologists have rightly lost interest in it. Perceptual experience can often be distinguished from judgmental bias (e.g., Dean et al., 2016; Durgin, Leonard-Solis, Masters, Schmelz & Li, 2012). To the extent that energetics theorists have stopped asking Koffka’s (1935) question, “Why do things look as they do?” their theories aren’t really about perceptual experience at all.