Construal and Metacognition 1 Running Head: CONSTRUAL AND METACOGNITION Missing the Trees for the Forest: A Construal Level Account of the Illusion of Explanatory Depth

An illusion of explanatory depth (IOED) occurs when people believe they understand a concept more deeply than they actually do. To date, IOEDs have been identified only in mechanical and natural domains, constraining our understanding of why they occur and suggesting that their implications are quite limited. We suggest that IOEDs occur because people adopt an inappropriately abstract construal style when they assess how well they understand concrete concepts. As this mechanism predicts, participants who naturally adopted concrete construal styles (Study 1) or were induced to adopt a concrete construal style (Studies 2-4 and 6), experienced diminished IOEDs. Demonstrating the generality of the construal mechanism, we also extended the presumed boundary conditions of the effect beyond mechanical and natural domains, documenting a novel IOED in the social psychological domain of electoral voting (Studies 5 and 6). These findings suggest a novel factor that might contribute to such diverse social-cognitive shortcomings as stereotyping, egocentrism, and the planning fallacy, where people adopt abstract representations of concepts that should be represented concretely. Construal and Metacognition 3 Missing the Trees for the Forest: Abstract Construal Fosters the Illusion of Explanatory Depth For over thirty years, research on human metacognition—the process of evaluating one’s own thoughts and abilities—has demonstrated the fallibility of human introspection (e.g., Alicke, 1985; Castel, McCabe, & Roediger, 2007; Kruger, 1999; Kruger & Dunning, 1999; Nisbett & Wilson, 1977; Rozenblit & Keil, 2002). People generally tend to believe they are more competent than they actually are, and this above-average effect is particularly pronounced amongst poor performers. For example, participants in one paper who demonstrated logical reasoning and grammatical skills in the bottom quartile of their classmates believed they actually performed better than 58% of their classmates (Kruger & Dunning, 1999). Though the poorest performers in the population tend to overestimate their capabilities most dramatically, more competent members of the population sometimes make similar errors in a variety of domains. One striking demonstration, the illusion of explanatory depth (IOED), arises when people overestimate their ability to explain mechanical and natural processes (e.g., Keil, 2003; Rozenblit & Keil, 2002). For example, people know that a zipper closes because it has teeth that somehow interlock, but they know very little about how the teeth actually interlock to enable the bridging mechanism. Similarly, many people know vaguely that an earthquake occurs because two geological plates collide and move relative to one another, but again they know little about the mechanism that initially produces these collisions. Nonetheless, people believe they understand these concepts quite deeply, and are surprised by the shallowness of their own explanations when prompted to describe the concepts thoroughly. To date, researchers have only shown IOEDs when people assess their understanding of manmade devices and natural processes (like zippers and earthquakes). In contrast, people Construal and Metacognition 4 appear to be better calibrated when assessing their understanding of semantic concepts, like declarative facts and trivia (e.g., country capitals; Rozenblit & Keil, 2002). As Rozenblit and Keil noted (2002), one important characteristic that distinguishes manmade devices and natural processes from trivia, is that people do not need to understand manmade devices deeply to appreciate their function. For example, whereas people either do or do not know that Ouagadougou is the capital of Burkina Faso, they might understand why zippers exist, how they look, and identify their parts without understanding how zippers function. Similarly, people need not understand how the earth quakes to recognize the consequences of an earthquake and the basic fact that they cause land to shift violently. In other words, people can understand certain concepts at an abstract level quite well, while only superficially understanding their more concrete characteristics. IOEDs are likely to emerge when people mistake their mastery of the abstract characteristics of the concept for a belief that they understand the concrete aspects of the concept much more deeply than they actually do. On the surface this appears similar to the mechanism that Pronin and her colleagues ascribe to the so-called introspection illusion—the tendency for perceivers to believe they understand other people more deeply than others understand them (e.g., Pronin, Gilovich, & Ross, 2004). The introspection illusion occurs because people focus on how others behave and ignore others’ thoughts, which are largely inaccessible. However, these mechanisms are distinct in one critical sense: people experience the IOED despite having access to both abstract and concrete information, whereas the introspection illusion occurs precisely because people have access to a limited field of information. The IOED is arguably easier to manipulate, then, because people are capable of assessing their understanding of the target mechanism—they just fail to focus on the information that illuminates the depth of their understanding and instead Construal and Metacognition 5 focus on relatively uninformative cues. Specifically, the IOED should diminish if people are induced to focus on a target’s concrete properties instead of its abstract features. To examine this question, we adopted the Construal Level Theory (CLT: for reviews, see Liberman & Trope, 2008; Trope & Liberman, 2003) framework. Construal Level Theory and the Illusion of Explanatory Depth According to CLT, people construe or represent the environment along a continuum from abstractly to concretely. Abstract representations tend to capture the essence of a target, focusing broadly on its superordinate or general features. In contrast, concrete representations focus on the target’s narrow or specific features. For example, in the context of serial processes like getting dressed in the morning, an abstract construal might emphasize why the process is important, whereas a concrete construal might emphasize how and in what order people complete the process (Trope & Liberman, 2003; Vallacher & Wegner, 1987). Construal style influences a vast array of cognitive processes, including attitude formation (e.g., Fujita, Eyal, Chaiken, Trope, & Liberman, 2008), likelihood estimates (e.g., Todorov, Goren, & Trope, 2007), and feature processing at large (e.g., Alter & Oppenheimer, 2008; Henderson, Fujita, Trope, & Liberman, 2006). Construal Level Theory has important implications for IOEDs. When asked whether they understand how a ballpoint pen works, for example, people are likely to reach very different conclusions depending on whether they adopt an abstract or concrete construal of the pen. Abstractly, ballpoint pens enable people to write information by accurately applying ink to a sheet of paper. This abstract construal is superficially compelling, and people in an abstract mindset are likely to interpret their understanding of the pen’s general, abstract properties as a satisfying, concise explanation of how the pen works. In fact, this abstract construal of the pen Construal and Metacognition 6 ignores the mechanism that enables the pen to perform that function in the first place. In contrast, a more concrete representation might focus on the parts that constitute the pen (the barrel, the ball, the ink reservoir) and how they work in concert to release the ink. Consider the following description of how a ball point pen works (from the materials used by Rozenblit & Keil, 2002): The key to a ballpoint pen is the ball. This ball acts as a buffer between the material you're writing on and the quick-drying ink inside the pen. The ball rotates freely and rolls out the ink as it is continuously fed from the ink reservoir (usually a narrow plastic tube filled with ink).The ball is kept in place -between the ink reservoir and the paper -by a socket; and while it is in tight, it still has enough room to roll around as you write. As the pen moves across the paper, the ball turns and gravity forces the ink down the reservoir and onto the ball, where it is transferred onto the paper. It's this rolling mechanism that allows the ink to flow onto the top of the ball and roll onto the paper you're writing on, while at the same time sealing the ink from the air so it does not dry in the reservoir. People who adopt a concrete representation of a ballpoint pen are therefore better equipped to assess how well they understand the role of each component of the pen than are people who adopt a broad, abstract construal of the process. IOEDs might arise because people confuse the metacognitive experience of understanding an abstract concept with the more relevant metacognitive experience of understanding the concept’s concrete details. These two metacognitive experiences arise from two distinct sets of cognitive content. For example, people who construe a ballpoint pen abstractly are more likely to focus on the pen’s function and perhaps its global appearance. In contrast, people who construe the pen concretely are more likely to focus on how well theyly are more likely to focus on the pen’s function and perhaps its global appearance. In contrast, people who construe the pen concretely are more likely to focus on how well they Construal and Metacognition 7 understand how its parts work together to enable the pen to function—in this case, the appropriate metacognition. Accordingly, people are less likely to overestimate their understanding of how the pen works when their introspections focus appropriately on the pen’s concrete features rather than its abstract features. Importantly, in both cases people rely on their metacognitions to assess their understanding of a pen. At the same time, however, they fail to recognize that construing the pen abstractly leads them to focus primarily on cognitive content that cannot illuminate how deeply they understand how the pen functions concretely. As such, IOEDs should become more pronounced as people adopt increasingly abstract mindsets, and construal style should mediate the strength of the illusion. Overview of Present Research We conducted six studies designed to show that the IOED emerges because people adopt an abstract construal style and inappropriately rely on the ensuing metacognitions when assessing their understanding of concrete concepts. Accordingly, we began by examining whether participants who naturally adopted an abstract construal style would experience heightened IOEDs (Study 1). Having explored the naturally arising relationship between construal and the IOED, we examined whether inducing a broad (Study 2) or abstract (Studies 3 and 4) construal style would exaggerate the IOED. We examined this effect using two distinct construal manipulations, and also tested directly whether participants who transiently adopted an abstract construal style experienced exaggerated IOEDs. Since this construal mechanism is domain-general, it suggests that the boundary conditions identified in earlier research (e.g., Rozenblit & Keil, 2002) may be too stringent. We therefore tested whether these boundary conditions should be relaxed by seeking evidence for an IOED in the novel domain of electoral Construal and Metacognition 8 voting, where voters typically construe candidates as abstract representatives of their chosen party’s ideals (Studies 5 and 6). Study 1: The Relationship between Abstraction and the IOED We began by examining whether people who naturally perceive the world more abstractly experience heightened IOEDs. Participants completed two tasks: first, a task designed to assess whether they were experiencing the IOED; and, second, a task that assessed how abstractly or concretely they construed a series of everyday behaviors. We calculated the relationship between these measures to determine the relationship between participants’ construal style and their tendency to experience the IOED. Method Participants Sixty adult participants (35 females; Mage = 36.22 years, SD = 13.22; 36 White, 22 Asian, 2 Black) 1 completed a brief two-part questionnaire on Amazon.com’s Mechanical Turk online survey program. Mechanical Turk allows researchers to post questionnaires that are completed by Amazon.com users who participate in exchange for small contributions towards an Amazon.com gift voucher. The platform records each participant’s IP address to prevent participants from completing the same questionnaire more than once. Since neither gender nor ethnicity interacted with the results in this or the following studies, we refrain from discussing gender and ethnicity effects further. Materials, Design, and Procedure IOED task. Participants were asked to complete a questionnaire titled “How do different things work?” The instructions explained that participants would be asked to rate their Construal and Metacognition 9 understanding of three different mechanical processes, each on a 7-point scale (from 1 = know nothing about how this object works to 7 = know everything about how this object works). To give participants a sense of how to use the scale, we provided the sample paragraph describing how a ballpoint pen works (from Rozenblit & Keil, 2002). A paragraph following the description explained that a score of 7 would require all the elements of the description (e.g., what the parts are, their function, how they interact), a 4 would require knowledge of some of the basics, but not all the intricacies of the description, and a score of 1 would reflect an absence of knowledge about how the object worked at all. Having read the sample paragraph, participants assessed their understanding of how three target devices (bicycle lock, sewing machine, and zipper) worked using the 7-point scale. After making these initial ratings, participants were asked to imagine they had “just met a person who did not understand how these three items work” and to “write as complete an explanation of how each item works” as they could manage. Participants typed open-ended descriptions of how each device worked. Finally, participants reassessed their knowledge of how each device worked again on the same 7-point scale as they used previously. We calculated the magnitude of each participants’ IOED by subtracting their post-expression knowledge estimate from their initial knowledge estimate (both averaged across the three devices). A larger positive score therefore represents a larger IOED. Construal measure. To assess participants’ prevailing construal style, we asked participants to indicate whether they preferred a concrete description or an abstract description for 13 everyday behaviors taken from the Behavioral Identification Form (BIF: Vallacher & Wegner, 1989; for a similar approach to ours, see Fujita et al., 2006). For example, participants decided whether they preferred the description of eating as “chewing and swallowing” (concrete Construal and Metacognition 10 construal) or “getting nutrition” (abstract construal). The two descriptions were assigned the labels of Description A and Description B, and we counterbalanced which label referred to the concrete and abstract descriptions. Participants indicated their relative preference for the two descriptions on a 7-point scale (from 1 = strongly prefer Description A to 7 = strongly prefer Description B). We reverse-scored the ratings when Description A was the abstract description and Description B was the concrete description, and averaged the resulting 13 ratings to measure participants’ relative preference for the abstract descriptions. Results and Discussion We found basic evidence for an IOED across the sample, as participants tended to significantly overestimate their ability to explain how the three devices worked before attempting to express those explanations in writing, M = .42, SD = .82, one-sample t (58) = 3.89, p < .0001, ηp 2 = .21. Participants also expressed a significant preference for the abstract descriptions of the 13 items in the BIF. Specifically, their responses tended to be significantly higher than the midpoint of 4 on the 7-point scale, where higher numbers represented a preference for abstract construal, M = 4.39, t (58) = 2.65, p = .01, ηp 2 = .11. This latter result suggests that participants tended to naturally prefer the descriptions consistent with an abstract construal mindset. Most importantly, consistent with our expectations, participants tended to report higher IOEDs when they adopted a more abstract mindset according to the BIF measure, r (58) = .26, p < .05. This result provides initial evidence that people who construe the world relatively abstractly experience more pronounced IOEDs.ly experience more pronounced IOEDs. Construal and Metacognition 11 Study 2: Focusing on Parts vs. Focusing on the Whole Study 1 suggested a naturally-occurring relationship between construal and the IOED. In the remaining studies, we sought to demonstrate experimentally that people experience greater IOEDs when they are induced to adopt abstract or broad mindsets rather than concrete or narrow mindsets. Assuming that IOEDs occur because people inappropriately consult abstract knowledge when assessing their understanding of concrete mechanical concepts, forcing people into a narrower concrete mindset should diminish the IOED. Study 2 examined whether people who were induced to think more narrowly about mechanical devices were more calibrated when assessing their understanding of those devices. Participants were asked to estimate how well they could explain the operation of three mechanical devices. In one condition, participants were asked to rate how well they could explain how the mechanical devices worked (the broad construal condition), and in the other condition they were asked to rate how well they could explain how the parts of the mechanical devices enabled them to work (the narrow construal condition). After attempting to explain how the three devices worked, participants reassessed their knowledge on the rating scale they had used previously. Although both conditions prompted a similar introspective process, the narrow construal condition induced a more concrete construal of each process by explicitly asking participants to consider its component parts. If the mechanism we proposed indeed drives the IOED, participants in the narrow construal condition should provide better-calibrated knowledge estimates. 2