- Email: [email protected]
The impact of a relational mindset on information distortion
Journal of Experimental Social Psychology 60 (2015) 1–7
Contents lists available at ScienceDirect
Journal of Experimental Social Psychology journal homepage: www.elsevier.com/locate/jesp
The impact of a relational mindset on information distortion Anne-Sophie Chaxel ⁎ McGill University, Marketing Department, Canada
H I G H L I G H T S • • • •
Generating solutions to cross-domain analogies activates a relational mindset. Predecisional information distortion increases when a relational mindset is activated. Construal level mediates the relationship between relational mindset and distortion. Cognitive effort moderates the relationship between relational mindset and distortion.
a r t i c l e
i n f o
Article history: Received 16 December 2014 Revised 23 April 2015 Accepted 26 April 2015 Available online 29 April 2015 Keywords: Information distortion Relational thinking Construal level Analogical thinking
a b s t r a c t The preference-supporting bias in information evaluation, known as information distortion, is a ubiquitous phenomenon. The present work demonstrates that priming a relational mindset induces individuals to process independent units of information interdependently and therefore contributes to increasing distortion. In three studies, a relational mindset is activated by asking participants to generate solutions to cross-domain analogies. All three studies show that the activation of a relational mindset then carries over into a second, unrelated choice task and increases distortion. In addition, the present work shows that generating solutions to cross-domain analogies activates a high level of construal, which in turn mediates the effect of relational thinking on information distortion. Finally, the present work also demonstrates that imposing a cognitive load during the choice task reduces the impact of the relational mindset on distortion. In sum, this research demonstrates that the same mechanism that promotes creative thinking (i.e., seeing relationships across concepts) may also induce more biased information processing by prompting individuals to process independent units of information interdependently. © 2015 Elsevier Inc. All rights reserved.
Information distortion is a confirmation bias defined as the biased evaluation of new information to support an emerging preference or belief (Russo, Medvec, & Meloy, 1996). For instance, imagine that you are deciding between two vacation packages (Hotel A and Hotel B) and that you develop a tentative preference for Hotel A. Information distortion occurs if your emerging preference for Hotel A causes a shift in your evaluation of subsequent information toward favoring Hotel A. In other words, your evaluation of incoming information is biased toward your emerging preference. To empirically determine the presence of information distortion, a method known as the stepwise evolution of preference (SEP) was developed by Russo et al. (1996). As above, consider a choice between two hotels, which would be described by several product attributes presented sequentially. To track the progress of the choice, participants provide two responses after reading each product attribute. First, they rate the diagnosticity of the information on a scale from 1 to 9, with 1 ⁎ Corresponding author at: McGill University, 1001 Sherbrooke Street West, Montreal, Quebec H3A 1G5, Canada. E-mail address: [email protected].
http://dx.doi.org/10.1016/j.jesp.2015.04.007 0022-1031/© 2015 Elsevier Inc. All rights reserved.
being “strongly favors Hotel A,” 9 being “strongly favors Hotel B,” and the midpoint of 5 being “favors neither hotel.” Pretests are used to create neutral attributes that do not favor either product (i.e., 5 on the 9-point scale). As their second response, participants identify which is the leading alternative (Hotel A or Hotel B). To calculate distortion, the absolute difference between the perceived diagnosticity of the decision maker and the unbiased estimate of the control group is calculated. This absolute difference is signed positively if it is directed toward favoring the leading alternative and signed negatively if it favors the trailing alternative. The mean difference among all attributes yields a single value for each individual that is generally positive and thereby indicates the presence of distortion. Nearly two decades of research on information distortion have demonstrated that its effects hold across populations (Carlson & Russo, 2001; Kostopoulou, Russo, Keenan, Delaney, & Douiri, 2012; Russo, Meloy, & Wilks, 2000) and across target categories (Russo, Meloy, & Medvec, 1998). Distortion has also been observed in a wide array of domains, such as gambling (Brownstein, Read, & Simon, 2004), jurors' verdicts (Carlson & Russo, 2001), wine selection (Carlson & Pearo, 2004), and professional auditing (Russo et al., 2000). Although prior research
2
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7
has focused on the variety of contexts that favor the emergence of distortion, the question of the drivers of this bias has only recently arisen (Chaxel, Russo, & Wiggins, in press; Russo, Carlson, Meloy, & Yong, 2008). In the context of this literature, the general objective of this research is to provide insight into the cognitive mechanisms that are associated with predecisional information distortion.
1. The drivers of distortion Recent research (Chaxel et al., in press; Russo et al., 2008) has demonstrated that information distortion is driven by the goal of cognitive consistency, that is, the desire for two beliefs to be consistent with one another. In the case of information distortion, the first belief is the emerging preference for one option over the other (generated by one's evaluation of prior information), and the second belief is the evaluation of new information. Although this conceptualization of cognitive consistency as a goal and driver of distortion has recently received some attention, the present work takes a different path to studying the mechanisms underlying distortion, namely, examining the cognitive processes that are associated with the emergence of distortion. Prior research has demonstrated that “cognitive consistency” within belief systems is maintained through coherence-driven mechanisms of constraint satisfaction (e.g., Read & Miller, 1994; Simon & Holyoak, 2002; Spellman, Ullman, & Holyoak, 1993). These models predict how belief systems are modified by incoming information and how they eliminate potential inconsistencies. In this case, consistency is perceived not only as a desirable end-state (i.e., a goal) but also, even foremost, as a way to organize our thoughts and to integrate incoming information with our existing knowledge and preferences. In other words, cognitive consistency may not only be conceptualized as a goal but also refer to specific cognitive procedures that encourage decision makers to draw meaningful connections between beliefs (i.e., a “relational” procedure), such as between a preference and the evaluation of new information in a sequential choice task. In such a case, the relational procedure would encourage the presence of distortion because pieces of information would be evaluated in relation to other pieces of information (“interdependently”) during the act of decision making. To test the hypothesis that such a relational procedure may be associated with the emergence of distortion, the present work first relies on recent methodological work on mindset priming (for a review, see Wyer & Xu, 2010). In this study, we restrict the definition of the term “mindset” to a cognitive procedure that can be activated in one domain and can influence decision making in a different subsequent situation (Gollwitzer, 1990; Gollwitzer, Heckhausen, & Steller, 1990; Gollwitzer & Kinney, 1989; Xu & Wyer, 2012). For example, making comparative judgments in one domain can activate a “which-to-choose” mindset that disposes consumers to decide which of two products to buy in a subsequent situation without considering the possibility of buying neither product (Xu & Wyer, 2007, 2008). Similarly, we posit that it may be possible to activate a relational mindset in a first task and to observe whether this activation affects distortion in a subsequent unrelated task. Second, the present work also relies on recent empirical work on relational thinking (Vendetti, Wu, & Holyoak, 2014), which has shown that generating solutions for cross-domain analogies in a first task activates a relational mindset that induces decision makers to identify relationships between seemingly dissimilar items in a second task. In the work of Vendetti et al. (2014), the second task required participants to find more relational matches between two dissimilar pictures. In our work, we predict that solving crossdomain analogies in an initial task should contribute to increasing information distortion in a second, unrelated choice task by inducing participants to make connections between their emerging preference and their evaluation of new information in a sequential choice task.
2. Relational reasoning, creativity, and construal level theory Prior research has demonstrated a close link between cross-domain analogical reasoning and creative thinking (Chan, Paletz, & Schunn, 2012; Dahl & Moreau, 2002; Holyoak & Thagard, 1995). The primary reason for such a close link is the property of cross-domain analogies that induces decision makers to make connections between ideas drawn from disparate domains, which subsequently fosters creative outputs. Therefore, the same relational mindset that may encourage creative thinking may also trigger information distortion, which requires making connections between one's prior preferences and the evaluation of new information. In addition, prior research has emphasized that cross-domain analogical reasoning activates abstract thinking (Gick & Holyoak, 1983; Knowlton, Morrison, Hummel, & Holyoak, 2012) because crossdomain analogies require participants to find abstract relationships between disparate domains. Consider a within-domain analogy such as nose:scent::tongue:taste. The solution to this analogy requires the participant to map an identical relation within each pair of items (the nose is the sense organ for scent, and the tongue is the sense organ for taste). Now consider a cross-domain analogy such as nose:scent::antenna:signal. In contrast to the solution to the withindomain analogy, this cross-domain analogy requires the participant to draw an abstract relation between both pairs of items that will bridge both domains (the nose detects scent as an antenna detects a signal). Therefore, cross-domain analogical reasoning yields not only relational reasoning but also, more precisely, abstract relational reasoning. Based on the relationships among cross-domain analogical thinking, abstract relational reasoning and creativity, this study proposes a link between the activation of a relational mindset and construal level theory (Liberman & Trope, 1998). Construal level theory posits that individuals represent psychologically distant events with abstract, general, high-level construals and represent psychologically near events with concrete, contextual, low-level construals (Liberman & Trope, 2008). In turn, a high level of construal activates holistic thinking (Smith & Trope, 2006) and fosters creativity by facilitating the association of distant concepts (Ward, 1995). Because solving cross-domain analogies requires “far-out thinking” (as named by Vendetti et al., 2014) by inducing people to draw abstract connections between semantically distant pairs of items, we propose that generating solutions to cross-domain analogies activates a high level of construal. As such, a relational mindset would induce more information distortion from participants in a choice task because of the adoption of a high-level, holistic, abstract type of reasoning. That is, the level of construal is proposed as a mediator of the relationship between the activation of a relational mindset and distortion. 3. Relational mindset and effort A third objective of this research is to investigate whether cognitive effort is necessary for a relational mindset to affect distortion. That is, we investigate whether cognitive effort is a moderator of the relationship between relational mindset and distortion. This investigation is motivated by earlier research by Waltz, Lau, Grewal, and Holyoak (2000), who showed that imposing a cognitive load on analogical reasoning results in fewer relational mappings between two visual scenes. In other words, relational thinking is impaired under conditions of low effort (in contrast to baseline conditions). If relational thinking is impaired by a cognitive load and if relational thinking is associated with greater distortion, then information distortion should be lower when effort is manipulated to be low (i.e., when a cognitive load is imposed). The unexpected consequence of such a result is that distortion as a bias may be less likely to appear when cognitive effort is low, as long as a relational mindset is activated. Interestingly, Polman and Russo (2012) showed that imposing a cognitive load leads to increased levels of distortion. However, this
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7
finding does not necessarily contradict the present hypothesis. Indeed, Polman and Russo (2012) observed the impact of a cognitive load under neutral, baseline conditions, whereas the present research predicts the impact of a cognitive load when a relational mindset is activated. That is, we do not predict a main effect of effort (as Polman and Russo did); rather, we expect a moderating role of effort when a relational mindset is activated. What is the theoretical consequence of such a distinction for the present research? Polman and Russo (2012) explained the effect of low effort on distortion by defending the idea that imposing a cognitive load activates a type of “System 1” reasoning (compared with System 2; for reviews see Evans, 2007; Kahneman & Frederick, 2002; Keren & Schul, 2009), which one of the features is to be associative. The authors explained as follows: “The predecisional distortion of information should be inhibited by a process that is more analytical and deliberative and, inversely, should be augmented by a process that is more associative” (p. 81; Polman & Russo, 2012). In other words, the researchers attributed their results to the associative nature of System 1, which should be activated when a cognitive load is imposed. In the present research, we activate a relational mindset; that is, we actually activate a cognitive procedure that is associative by nature. Impairing the action of the cognitive procedure (by imposing a cognitive load) should in turn reduce the associative action of the relational mindset and may thereby reduce distortion. On a larger scale, such prediction raises the question of whether all associative processes are necessarily System 1. The present research and hypotheses hint at the idea that it may not, and that different types of associative processes may exist and (at least in the case of distortion) trigger similar results. 4. Objectives and studies
3
domain (cross-domain) analogy required the participants to focus on matching identical relations within the same domain (across domains). For instance, an analogy such as nose:scent::tongue:taste would qualify as a within-domain analogy because the C:D pair is semantically close to the A:B pair. By contrast, nose:scent::antenna:signal is a crossdomain analogy because the two pairs belong to different semantic categories. In the second part of the study, the participants chose between two options (athletic shoes labeled as Product J and Product K) based on five attributes that were presented sequentially (e.g., Russo et al., 1998). After reading each attribute, the participants were asked two questions. The first involved a rating of the attribute on a 9-point scale in which 1 indicated “strongly favors” one product, 9 indicated “strongly favors” the other product, and 5 indicated “favors neither product.” This response formed the basis of the measurement of information distortion. Second, knowing that more information would subsequently be provided, the participants were asked which product they would choose if they were to select one product at this stage. This response identified the currently leading option. Notably, all attributes presented during the choice task were pretested in an independent study conducted with 50 participants. In this pre-test, all participants were asked to rate the same attributes on the same 9-point scale, but the five attributes were said to be describing 10 different products; that is, the labels used to describe the two products varied for each attribute. Varying labels prevented the development of a preference for one product and in turn prevented information distortion (which requires a prior preference). The pre-test confirmed that no attribute evaluation differed significantly from the scale mid-point, 5 (all ps N .15). In other words, all attributes were “neutral” when seen independently, i.e., they did not favor either product.
In sum, the present research has three objectives: 1. To show that distortion is elevated by the activation of a relational mindset, 2. To establish that the relational mindset affects distortion through the activation of a high construal level, and 3. To investigate whether low effort (by imposing a cognitive load) reduces the impact of a relational mindset on information distortion. Three studies were run. Study 1 establishes the main effect of the relational mindset on distortion (first objective). Study 2 investigates the mediation through construal level (second objective). Study 3 examines the moderating role of effort (third objective). The final sample size of minimum 60 participants in each condition of each study was consistent with the required sample size, which had been found in advance to be 58 participants for a power of .70, an expected .70 unit difference in distortion between the two main experimental conditions (control vs. relational mindset), and an expected standard deviation of 1.5. 5. Study 1 5.1. Method Participants were recruited online (N = 120, U.S. residents only) using Amazon's Mechanical Turk, and they were compensated 65 cents for their participation. All participants first generated solutions to 20 four-term verbal analogy problems, such as nose:scent::tongue:taste. All problems were presented in an A:B::C:D format with the D term left blank. Immediately afterward, all participants completed a binary choice task between two athletic shoes. The stimulus for the analogy task was identical to the stimulus used by Green, Kraemer, Fugelsang, Gray, and Dunbar (2010) and Vendetti et al. (2014). More specifically, two types of analogies were used: cross-domain or within-domain analogies. The type of analogy was a between-subjects factor, and the participants were randomly assigned to one of the two conditions. Generating a solution to a within-
5.2. Results The calculation of information distortion required two steps (Russo et al., 1998). First, using the 9-point scale, a participant's estimate of an attribute's value was compared to the unbiased evaluation of identical information, which had been pretested to be 5. Second, the absolute difference between those two values was given a positive sign if the attribute's rating favored the leading alternative and a negative sign if the rating favored the trailing alternative. An average distortion score was then computed across the attributes. We expected participants who generated solutions to crossdomain analogies to distort information more than those who solved within-domain analogies. Confirming this expectation, we found that the mean information distortion score was significantly higher in the cross-domain analogies condition (M = 1.52, SD = 1.52) than in the within-domain analogies condition (M = .81, SD = 1.46; t(118) = − 2.63, p b .01). A MANOVA of the distortion scores following the presentation each attribute confirmed the main effect of the between-subjects factor (F(1, 118) = 6.95; p b .01) and revealed a main effect of the within-subjects factor: distortion tended to increase over the course of the choice process (F(3, 116) = 2.89; p b .05).
6. Study 2 The results of Study 1 confirm that activating a relational mindset increases distortion. Study 2 investigated whether this result would be mediated by construal level. Specifically, we predicted that activating a relational mindset with cross-domain analogies would induce to participants to think at a higher level of construal, which would in turn contribute to increased distortion.
4
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7
6.1. Method A total of 210 U.S. residents participated in a 20-minute study with a between-subjects design and three conditions: within-domain analogies, cross-domain analogies, and a control group. The participants were recruited on Amazon Mechanical Turk in exchange for 65 cents. The two analogy conditions were identical to those used in Study 1. The additional control group wrote two short essays on neutral topics (the Statue of Liberty and the Great Depression of 1929). After the priming task, all participants completed the Behavioral Identification Form (BIF; Vallacher & Wegner, 1989), which asked them to identify 25 actions (e.g., I am talking to a child) with either a low-level statement (e.g., I am using simple words) or a high-level statement (I am teaching a child something). The average number of high-level statements made by participants was used as a measure of construal level (with higher scores indicating higher construal levels). Finally, all participants completed a choice task similar to the task in Study 1, but in this case, the choice was between two pairs of headphones described by five product attributes: technical performance, music trials, pros and cons, design, and editor review. As in Study 1, each attribute had been pre-tested to be neutral (i.e., to not be significantly different from 5 on the relative information evaluation scale; all ps N .15) such that the attributes would not favor either of the two products.
relational thinking and distortion, we conducted a mediation analysis with 5,000 bootstrap samples. The control group was omitted, and the mediation analysis was conducted with the within vs. cross-domain analogies as a dummy predictor variable. The indirect effect reached a bootstrap confidence interval of [.0179; .4013], which does not contain zero and therefore indicates significance. The results of the mediation analysis are presented in Fig. 1. 7. Study 3 The objective of Study 3 was to investigate the moderating role of effort in the relationship between relational mindset and distortion. Specifically, we hypothesized that the effect of the relational mindset on distortion is reduced when effort is low, as cross-domain analogical thinking has been shown to be effortful (Waltz et al., 2000). When a relational mindset is activated, distortion should therefore be lower when effort is manipulated to be low. To manipulate effort, we imposed a cognitive load (low versus high) on the participants during the choice task (e.g., Epley & Gilovich, 2006; Polman & Russo, 2012). Notably, in the control condition in which a relational mindset is not activated, imposing a cognitive load should increase distortion if our results are consistent with those of Polman and Russo (2012). That is, imposing a cognitive load should activate a System 1 type of reasoning, which may contribute to increased distortion. Hence, a cross-over interaction effect of relational mindset and effort on distortion was predicted.
6.2. Results 7.1. Method Information distortion was computed following the procedures in Study 1, resulting in one aggregated distortion score for each participant. The results of this study replicated the results of Study 1. Overall, participants who completed cross-domain analogies tended to distort information more (M = 1.23, SD = 1.21) than participants who completed within-domain analogies (M = .72, SD = 1.59). Even when we controlled for multiple comparisons (using Hsu's multiple comparisons with the best method — Hsu MCB), the difference between the two conditions reached significance (t(137) = 2.17, p b .05). The score for the control group was between the scores for the two experimental conditions but was not significantly different from the score for the cross-domain analogies condition (M = .83, SD = 1.28). Second, the participants in the cross-domain analogies condition tended to show a higher level of construal (M = .70, SD = .22) than the participants in the two other conditions (M = .61 and SD = .25 for the control group; M = .54 and SD = .24 for the within-domain analogies condition). Even when we applied Hsu's MCB method, significant differences were observed between the cross-domain analogies condition and the within-domain analogies condition (t(137) = 3.90, p b .0001) and between the control condition and the cross-domain analogies condition (t(139) = −2.2, p b .0001). Finally, we observed a positive relationship between the level of construal and information distortion: a higher construal level induced more distortion (b = 1.14 t = 3.04, p b .005). To more closely examine the role of construal level as a mediator of the relationship between
A total of 360 U.S. residents participated in a 20-minute study on Amazon Mechanical Turk in exchange for 75 cents. The experiment was a 3 (priming: cross-domain analogies vs. within-domain analogies vs. control) × 2 (load: high vs. low) between-subjects design. The analogies differed from those used in Studies 1 and 2 in order to replicate the effect with different sets. The control group wrote two short essays on neutral topics (the Statue of Liberty and the Great Depression of 1929) rather than solving analogies. The cognitive load manipulation was then administered. Immediately after solving the analogies (or writing the essays), the participants in the high-load condition were asked to memorize a string of eight letters (POQWYTNM) while completing the choice task (Epley & Gilovich, 2006). The participants in the low-load condition were also asked to memorize a string of letters, but each string contained only two letters (YU). The instructions clearly indicated that the letter string would need to be recalled after the choice task. The instructions also noted that the letter string should not be written down, as doing so would compromise the purpose of the experiment. The last part of the study consisted of a choice task (similar to Studies 1 and 2) between two vacation resorts. Each resort was described by five attributes: atmosphere, amenities, location, travel guide descriptions, and club activities. As in Studies 1 and 2, each attribute had been pre-tested with 50 participants to ensure that no attribute evaluation differed significantly from the scale mid-point, 5 (all ps N .15). At
Level of Construal b = .16***
b = 1.27 **
b = .50* (.33) Relational Mindset
Information Distortion
Fig. 1. Mediation analysis in Study 2 (N = 68 in the cross-domain analogies condition and N = 70 in the within-domain analogies condition). The values in parentheses indicate the direct effect of the interaction on information distortion when controlling for the level of construal. *p b .05, **p b .01, ***p b .0001.
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7
the end of the choice task, all participants were asked to recall the letter string that that they had been told to memorize. 7.2. Results First, we examined the results when cognitive load was low (i.e., in baseline conditions of effort) to confirm the results of Studies 1 and 2. When cognitive load was low, information distortion was lower in the within-domain analogies condition (M = .70, SD = 1.42) than in the cross-domain analogies condition (M = 1.42, SD = 1.67). The difference reached significance (correcting for multiple comparisons using Hsu's MCB method), t(177) = −2.46, p b .05. Again, the score for the control group was between the scores for the experimental conditions (M = .99, SD = 1.80) but did not significantly differ from the score for the cross-domain analogy condition (p N .05). Therefore, the results of Studies 1 and 2 were replicated. Second, we examined the effect of the interaction between cognitive load and condition on distortion. As predicted, a full-factorial ANOVA on information distortion with cognitive load and conditions (withindomain analogies vs. control vs. cross-domain analogies) as the main factors yielded a significant interaction effect (F(2, 354) = 4.43, p b .05). When a relational mindset was induced (i.e., in the crossdomain analogy condition), the participants in the low-load condition distorted information more than the participants in the high-load condition (Mlow = 1.42, SDlow = 1.67 vs. Mhigh = .81, SDhigh = 1.33; t(122) = 2.26, p b .05). In other words, reducing effort weakened the effect of the relational mindset on distortion. Conversely, when no relational mindset was induced (i.e., in the control and within-domain analogy conditions), the participants in the low-load condition distorted information less than the participants in the high-load condition (Mlow = .84, SDlow = 1.62 vs. Mhigh = 1.25, SDhigh = 1.56; t(234) = − 2.01, p b .05). This result replicated Polman and Russo's findings: in the absence of the activation of a relational mindset, imposing a cognitive load increases the extent to which decision makers distort information. The results of Study 3 confirm that impairing the activation of the relational mindset (with a cognitive load) lessens its impact on distortion. However, when no relational mindset is activated, less effort increases distortion, possibly because of the automatic activation of System 1 (Polman & Russo, 2012). 8. Discussion 8.1. Summary of results This research shows that priming a relational type of reasoning increases distortion through the activation of a higher level of construal. As such, the same mechanism that has been shown to promote creative thinking (i.e., recognizing relationships across concepts) may also induce more biased information processing by prompting individuals to process independent units of information interdependently. The present research also shows that this process of information integration is relatively effortful, as imposing a cognitive load during the choice task impairs the impact of the relational mindset on distortion. Across our studies, the effect size of the relational mindset on distortion ranged from .36 to .48 (see Table 1A). Overall, the demonstrated
5
effect on distortion is moderate, and we recognize that the power of relational thinking to drive the same effects in a natural setting may be weaker. The results prompt the question of naturally occurring primes, those that are conceptually similar to the relational mindset but more likely to occur spontaneously. In parallel, the demonstrated effect on construal level is relatively high (.69) but is comparable to seminal papers using the BIF as a dependent variable with more traditional construal level priming (see Table 1B). Without a complete meta-analysis, it is difficult to determine the relative effectiveness of cross-domain analogical thinking for construal level with other methods, such as psychological distance (Liberman & Trope, 1998) or why/how reasoning (Freitas, Gollwitzer, & Trope, 2004). However, we can reasonably state that relational thinking activated by crossdomain analogies appears to yield results that are sufficiently strong to provide a meaningful way to yield a high level of construal. 8.2. Discussion of findings in relation to prior research The present paper does not claim that a relational mindset must be activated to increase distortion. Rather, this study presents one way in which information distortion can be increased. For instance, Study 3 shows that imposing a cognitive load under neutral conditions also increases information distortion (see Polman & Russo, 2012). As noted in the introduction, imposing a cognitive load in the absence of the activation of a relational mindset may activate a more “System 1” type of reasoning, which is associative by nature and therefore encourages distortion. Thus, different contexts or mechanisms that encourage relational thinking may result in the same outcome (i.e., more distortion). Some mechanisms may be effortful (as in our case), and some may be more automatic (as is assumed with System 1). Additionally, it is important to distinguish the present results with findings obtained in the self-regulatory literature (Freitas et al., 2004; Fujita, Gollwitzer, & Oettingen, 2007; Taylor & Gollwitzer, 1995) regarding the effect of deliberative and implemental mindsets on judgmental biases and, in particular, on selective information processing. For instance, Fujita et al. (2007) showed that a deliberative mindset (compared with an implemental mindset) increases open-mindedness to processing incidental information, thereby leading decision makers to develop more accurate judgments. In our work, we do not directly test whether participants are in an implemental or deliberative mindset. On the one hand, activating a relational mindset may increase the tendency of participants to compare information and therefore to engage in an even more deliberative type of reasoning. On the other hand, priming a relational mindset could actually induce participants to reach a decision sooner because they are more prone to regard units of information as consistent with one another. Thereby, subjects may switch to a more implemental mindset (i.e., more selective information processing) earlier in the choice process to defend their decisions. Without further testing, it is therefore difficult to predict which of the two mindsets (deliberative or implemental) is more closely associated with a relational mindset and at which stages of the decision-making process. Interestingly, the current work may help explain two findings from previous studies. Because positive affect has recently been shown to promote a high level of construal (Pyone & Isen, 2011), the present results may help explain why positive affect increases distortion
Table 1A Effect sizes in Studies 1, 2 and 3. Study
Dependent variable
M1 within-domain analogies
M2 cross-domain analogies
SD1 within-domain analogies
SD2 cross-domain analogies
Pooled SD
Effect size
1 2 2 3
Distortion Distortion BIF — average high-level items out of 25 items Distortion
0.81 0.72 0.54 0.7
1.52 1.23 0.70 1.42
1.46 1.59 0.24 1.42
1.52 1.21 0.22 1.66
1.49 1.41 0.23 1.54
0.48 0.36 0.69 0.47
6
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7
Table 1B Effect sizes in articles using the Behavioral Identification Form. Source
Manipulation
Dependent variable
M1 low construal
M2 high construal
Pooled SD
Effect size
Liberman & Trope (1998) (Study 1, N = 32) Fujita, Henderson, Eng, Trope, & Liberman (2006) (Study 1, N = 68) Liviatan, Trope, & Liberman (2008) (Study 1, N = 34) Pyone & Isen (2011) (Study 1, N = 41) Alter, Oppenheimer, & Zemla (2010) (Study 3A, N = 80)
Psychological distance Psychological distance
BIF — count of high-level items out of 19 items BIF — count of high-level items out of 13 items
10.19 8.47
13.44 9.88
3.59 2.58
0.90 0.55
Interpersonal similarity
BIF — count of high-level items out of 19 items
11.59
14.65
4.36
0.70
Positive affect Why vs. how
BIF — count of high-level items out of 26 items BIF — preference for high-level items on a 7-point scale
12.85 3.49
16.19 4.67
4.55 0.71
0.73 1.66
(Meloy, 2000). In addition, decisions made for others (as opposed to decisions made for oneself) have been shown to be more creative because of the activation of a high level of construal (Polman & Emich, 2011). Therefore, the present results may also help explain why decisions made for others (as opposed to decisions made for oneself) trigger more predecisional distortion (Polman, 2010). Although this conjecture remains speculative, the present research appears to corroborate the findings of previous studies in this regard. References Alter, A.L., Oppenheimer, D.M., & Zemla, J.C. (2010). Missing the trees for the forest: A construal level account of the illusion of explanatory depth. Journal of Personality and Social Psychology, 99(3), 436–451. http://dx.doi.org/10.1037/a0020218. Brownstein, A.L., Read, S.J., & Simon, D. (2004). Bias at the racetrack: Effects of individual expertise and task importance on predecision reevaluation of alternatives. Personality and Social Psychology Bulletin, 30(7), 891–904. http://dx.doi.org/10.1177/ 0146167204264083. Carlson, K.A., & Pearo, L.K. (2004). Limiting predecisional distortion by prior valuation of attribute components. Organizational Behavior and Human Decision Processes, 94(1), 48–59. http://dx.doi.org/10.1016/j.obhdp.2004.02.001. Carlson, K.A., & Russo, J.E. (2001). Biased interpretation of evidence by mock jurors. Journal of Experimental Psychology. Applied, 7, 91–103. Chan, J., Paletz, S.B., & Schunn, C.D. (2012). Analogy as a strategy for supporting complex problem solving under uncertainty. Memory & Cognition, 40, 1352–1365. http://dx. doi.org/10.3758/s13421-012-0227-z. Chaxel, A. S., Russo, J. E., & Wiggins, C. E. (2015). A Goal-Priming Approach to Cognitive Consistency: Applications to Judgment, Journal of Behavioral Decision Making. . http://dx. doi.org/10.1002/bdm.1874 (in press). Dahl, D.W., & Moreau, P. (2002). The influence and value of analogical thinking during new product ideation. Journal of Marketing Research, 39, 47–60. http://dx.doi.org/10. 1509/jmkr.39.1.47.18930. Epley, N., & Gilovich, T. (2006). The anchoring-and-adjustment heuristic: Why the adjustments are insufficient. Psychological Science, 17, 311–318. http://dx.doi.org/10.1111/j. 1467-9280.2006.01704.x. Evans, J.S.B.T. (2007). On the resolution of conflict in dual process theories of reasoning. Thinking and Reasoning, 13(4), 321–339. http://dx.doi.org/10.1080/ 13546780601008825. Freitas, A.L., Gollwitzer, P., & Trope, Y. (2004). The influence of abstract and concrete mindsets on anticipating and guiding others' self-regulatory efforts. Journal of Experimental Social Psychology, 40(6), 739–752. http://dx.doi.org/10.1016/j.jesp. 2004.04.003. Fujita, K., Gollwitzer, P.M., & Oettingen, G. (2007). Mindsets and pre-conscious openmindedness to incidental information. Journal of Experimental Social Psychology, 43, 48–61. http://dx.doi.org/10.1016/j.jesp.2005.12.004. Fujita, K., Henderson, M.D., Eng, J., Trope, Y., & Liberman, N. (2006). Spatial distance and mental construal of social events. Psychological Science, 17(4), 278–282. http://dx. doi.org/10.1111/j.1467-9280.2006.01698.x. Gick, M.L., & Holyoak, K.J. (1983). Schema induction and analogical transfer. Cognitive Psychology, 15, 1–38. http://dx.doi.org/10.1016/0010-0285(83)90002-6. Gollwitzer, P.M. (1990). Action phases and mind-sets. In E.T. Higgins, & R.M. Sorrentino (Eds.), Handbook of Motivation and Cognition: Foundations of Social Behavior, vol. 2. (pp. 53–92). New York: Guilford Press. Gollwitzer, P.M., Heckhausen, H., & Steller, B. (1990). Deliberative versus implemental mind-sets: Cognitive tuning toward congruous thoughts and information. Journal of Personality and Social Psychology, 59, 1119–1127. http://dx.doi.org/10.1037/00223514.59.6.1119. Gollwitzer, P.M., & Kinney, R.F. (1989). Effects of deliberative and implemental mind-sets on illusion of control. Journal of Personality and Social Psychology, 56, 531–554. http:// dx.doi.org/10.1037/0022-3514.56.4.531. Green, A.E., Kraemer, D.J., Fugelsang, J.A., Gray, J.R., & Dunbar, K.N. (2010). Connecting long distance: Semantic distance in analogical reasoning modulates frontopolar cortex activity. Cerebral Cortex, 20(1), 70–76. http://dx.doi.org/10.1093/cercor/bhp081. Holyoak, K.J., & Thagard, P. (1995). Mental Leaps: Analogy in Creative Thought. Cambridge, MA: MIT Press.
Kahneman, D., & Frederick, S. (2002). Representativeness revisited: Attribute substitution in intuitive judgment. Heuristics and Biases: The Psychology of Intuitive Judgment, 49. Keren, G., & Schul, Y. (2009). Two is not always better than one. A critical evaluation of two-system theories. Perspectives on Psychological Science, 4(6), 533–550. http://dx. doi.org/10.1111/j.1745-6924.2009.01164.x. Knowlton, B.J., Morrison, R.G., Hummel, J.E., & Holyoak, K.J. (2012). A neurocomputational system for relational reasoning. Trends in Cognitive Sciences, 16, 373–381. http://dx. doi.org/10.1016/j.tics.2012.06.002. Kostopoulou, O., Russo, J.E., Keenan, G., Delaney, B.C., & Douiri, A. (2012). Information distortion in physicians' diagnostic judgments. Medical Decision Making, 32(6), 831–839. http://dx.doi.org/10.1177/0272989X12447241. Liberman, N., & Trope, Y. (1998). The role of feasibility and desirability considerations in near and distant future decisions: A test of temporal construal theory. Journal of Personality and Social Psychology, 75(1), 5–18. http://dx.doi.org/10.1037//00223514.75.1.5. Liberman, N., & Trope, Y. (2008). The psychology of transcending the here and now. Science, 322, 1201–1205. http://dx.doi.org/10.1126/science.1161958. Liviatan, I., Trope, Y., & Liberman, N. (2008). Interpersonal similarity as a social distance dimension: Implications for perception of others' actions. Journal of Experimental Social Psychology, 44(5), 1256–1269. http://dx.doi.org/10.1016/j.jesp.2008.04.007. Meloy, M. (2000). Mood‐driven distortion of product information. Journal of Consumer Research, 27, 345–359. http://dx.doi.org/10.1086/317589. Polman, E. (2010). Information distortion in self-other decision making. Journal of Experimental Social Psychology, 46, 432–435. http://dx.doi.org/10.1016/j.jesp.2009. 11.003. Polman, E., & Emich, K.J. (2011). Decisions for others are more creative than decisions for the self. Personality and Social Psychology Bulletin, 37, 492–501. http://dx.doi.org/10. 1177/0146167211398362. Polman, E., & Russo, J.E. (2012). Commitment to a developing preference and predecisional distortion of information. Organizational Behavior and Human Decision Processes, 119, 78–88. http://dx.doi.org/10.1016/j.obhdp.2012.05.004. Pyone, J.S., & Isen, A.M. (2011). Positive affect, intertemporal choice, and levels of thinking: Increasing consumers' willingness to wait. Journal of Marketing Research, 48, 532–543. http://dx.doi.org/10.1509/jmkr.48.3.532. Read, S.J., & Miller, L.C. (1994). Dissonance and balance in belief systems: The promise of parallel constraint satisfaction processes and connectionist modeling approaches. In R.C. Schank, & E.J. Langer (Eds.), Beliefs, Reasoning, and Decision Making: Psycho-logic in Honor of Bob Abelson. Hillsdale, NJ: Erlbaum. Russo, J.E., Carlson, K.A., Meloy, M.G., & Yong, K. (2008). The goal of consistency as a cause of information distortion. Journal of Experimental Psychology. General, 137, 456–470. http://dx.doi.org/10.1037/a0012786. Russo, J.E., Medvec, V.H., & Meloy, M.G. (1996). The distortion of information during decisions. Organizational Behavior and Human Decision Processes, 66, 102–111. http://dx.doi.org/10.1006/obhd.1996.0041. Russo, J.E., Meloy, M.G., & Medvec, V.H. (1998). Predecisional distortion of product information. Journal of Marketing Research, 35, 438–452. http://dx.doi.org/10.2307/ 3152163. Russo, J.E., Meloy, M.G., & Wilks, T.J. (2000). Predecisional distortion of information by auditors and salespersons. Management Science, 46(1), 13–27. http://dx.doi.org/10. 1287/mnsc.46.1.13.15127. Simon, D., & Holyoak, K.J. (2002). Structural dynamics of cognition: From consistency theories to constraint satisfaction. Personality and Social Psychology Review, 6, 283–294. http://dx.doi.org/10.1207/S15327957PSPR0604_03. Smith, P.K., & Trope, Y. (2006). You focus on the forest when you're in charge of the trees: Power priming and abstract information processing. Journal of Personality and Social Psychology, 90, 578–596. http://dx.doi.org/10.1037/0022-3514.90.4.578. Spellman, B.A., Ullman, J.B., & Holyoak, K.J. (1993). A coherence model of cognitive consistency: Dynamics of attitude change during the Persian Gulf War. Journal of Social Issues, 49, 147–165. http://dx.doi.org/10.1111/j.1540-4560.1993.tb01185.x. Taylor, S.E., & Gollwitzer, P.M. (1995). Effects of mindset on positive illusions. Journal of Personality and Social Psychology, 69(2), 213–226. http://dx.doi.org/10.1037/00223514.69.2.213. Vallacher, R.R., & Wegner, D.M. (1989). Levels of personal agency: Individual variation in action identification. Journal of Personality and Social Psychology, 57, 660–671. http:// dx.doi.org/10.1037/0022-3514.57.4.660. Vendetti, M.S., Wu, A., & Holyoak, K.J. (2014). Far-out thinking: Generating solutions to distant analogies promotes relational thinking. Psychological Science, 25, 928–933. http://dx.doi.org/10.1177/0956797613518079.
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7 Waltz, J.A., Lau, A., Grewal, S.K., & Holyoak, K.J. (2000). The role of working memory in analogical mapping. Memory and Cognition, 28, 1205–1212. http://dx.doi.org/10.3758/ BF03211821. Ward, T.B. (1995). What's old about new ideas? In S.M. Smith, T.B. Ward, & R.A. Finke (Eds.), The Creative Cognition Approach (pp. 157–178). Cambridge, MA: MIT Press. Wyer, R.S., & Xu, A.J. (2010). The role of behavioral mind-sets in goal-directed activity: Conceptual underpinnings and empirical evidence. Journal of Consumer Psychology, 20(2), 107–125. http://dx.doi.org/10.1016/j.jcps.2010.01.003.
7
Xu, A., & Wyer, R., Jr. (2007). The effect of mind‐sets on consumer decision strategies. Journal of Consumer Research, 34(4), 556–566. http://dx.doi.org/10.1086/519293. Xu, A.J., & Wyer, R.S. (2008). The comparative mind-set: From animal comparisons to increased purchase intentions. Psychological Science, 19, 859–864. http://dx.doi.org/ 10.1111/j.1467-9280.2008.02169.x. Xu, A.J., & Wyer, R.S., Jr. (2012). The role of bolstering and counterarguing mind-sets in persuasion. Journal of Consumer Research, 38(5), 920–932. http://dx.doi.org/10. 1086/661112.
Contents lists available at ScienceDirect
Journal of Experimental Social Psychology journal homepage: www.elsevier.com/locate/jesp
The impact of a relational mindset on information distortion Anne-Sophie Chaxel ⁎ McGill University, Marketing Department, Canada
H I G H L I G H T S • • • •
Generating solutions to cross-domain analogies activates a relational mindset. Predecisional information distortion increases when a relational mindset is activated. Construal level mediates the relationship between relational mindset and distortion. Cognitive effort moderates the relationship between relational mindset and distortion.
a r t i c l e
i n f o
Article history: Received 16 December 2014 Revised 23 April 2015 Accepted 26 April 2015 Available online 29 April 2015 Keywords: Information distortion Relational thinking Construal level Analogical thinking
a b s t r a c t The preference-supporting bias in information evaluation, known as information distortion, is a ubiquitous phenomenon. The present work demonstrates that priming a relational mindset induces individuals to process independent units of information interdependently and therefore contributes to increasing distortion. In three studies, a relational mindset is activated by asking participants to generate solutions to cross-domain analogies. All three studies show that the activation of a relational mindset then carries over into a second, unrelated choice task and increases distortion. In addition, the present work shows that generating solutions to cross-domain analogies activates a high level of construal, which in turn mediates the effect of relational thinking on information distortion. Finally, the present work also demonstrates that imposing a cognitive load during the choice task reduces the impact of the relational mindset on distortion. In sum, this research demonstrates that the same mechanism that promotes creative thinking (i.e., seeing relationships across concepts) may also induce more biased information processing by prompting individuals to process independent units of information interdependently. © 2015 Elsevier Inc. All rights reserved.
Information distortion is a confirmation bias defined as the biased evaluation of new information to support an emerging preference or belief (Russo, Medvec, & Meloy, 1996). For instance, imagine that you are deciding between two vacation packages (Hotel A and Hotel B) and that you develop a tentative preference for Hotel A. Information distortion occurs if your emerging preference for Hotel A causes a shift in your evaluation of subsequent information toward favoring Hotel A. In other words, your evaluation of incoming information is biased toward your emerging preference. To empirically determine the presence of information distortion, a method known as the stepwise evolution of preference (SEP) was developed by Russo et al. (1996). As above, consider a choice between two hotels, which would be described by several product attributes presented sequentially. To track the progress of the choice, participants provide two responses after reading each product attribute. First, they rate the diagnosticity of the information on a scale from 1 to 9, with 1 ⁎ Corresponding author at: McGill University, 1001 Sherbrooke Street West, Montreal, Quebec H3A 1G5, Canada. E-mail address: [email protected].
http://dx.doi.org/10.1016/j.jesp.2015.04.007 0022-1031/© 2015 Elsevier Inc. All rights reserved.
being “strongly favors Hotel A,” 9 being “strongly favors Hotel B,” and the midpoint of 5 being “favors neither hotel.” Pretests are used to create neutral attributes that do not favor either product (i.e., 5 on the 9-point scale). As their second response, participants identify which is the leading alternative (Hotel A or Hotel B). To calculate distortion, the absolute difference between the perceived diagnosticity of the decision maker and the unbiased estimate of the control group is calculated. This absolute difference is signed positively if it is directed toward favoring the leading alternative and signed negatively if it favors the trailing alternative. The mean difference among all attributes yields a single value for each individual that is generally positive and thereby indicates the presence of distortion. Nearly two decades of research on information distortion have demonstrated that its effects hold across populations (Carlson & Russo, 2001; Kostopoulou, Russo, Keenan, Delaney, & Douiri, 2012; Russo, Meloy, & Wilks, 2000) and across target categories (Russo, Meloy, & Medvec, 1998). Distortion has also been observed in a wide array of domains, such as gambling (Brownstein, Read, & Simon, 2004), jurors' verdicts (Carlson & Russo, 2001), wine selection (Carlson & Pearo, 2004), and professional auditing (Russo et al., 2000). Although prior research
2
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7
has focused on the variety of contexts that favor the emergence of distortion, the question of the drivers of this bias has only recently arisen (Chaxel, Russo, & Wiggins, in press; Russo, Carlson, Meloy, & Yong, 2008). In the context of this literature, the general objective of this research is to provide insight into the cognitive mechanisms that are associated with predecisional information distortion.
1. The drivers of distortion Recent research (Chaxel et al., in press; Russo et al., 2008) has demonstrated that information distortion is driven by the goal of cognitive consistency, that is, the desire for two beliefs to be consistent with one another. In the case of information distortion, the first belief is the emerging preference for one option over the other (generated by one's evaluation of prior information), and the second belief is the evaluation of new information. Although this conceptualization of cognitive consistency as a goal and driver of distortion has recently received some attention, the present work takes a different path to studying the mechanisms underlying distortion, namely, examining the cognitive processes that are associated with the emergence of distortion. Prior research has demonstrated that “cognitive consistency” within belief systems is maintained through coherence-driven mechanisms of constraint satisfaction (e.g., Read & Miller, 1994; Simon & Holyoak, 2002; Spellman, Ullman, & Holyoak, 1993). These models predict how belief systems are modified by incoming information and how they eliminate potential inconsistencies. In this case, consistency is perceived not only as a desirable end-state (i.e., a goal) but also, even foremost, as a way to organize our thoughts and to integrate incoming information with our existing knowledge and preferences. In other words, cognitive consistency may not only be conceptualized as a goal but also refer to specific cognitive procedures that encourage decision makers to draw meaningful connections between beliefs (i.e., a “relational” procedure), such as between a preference and the evaluation of new information in a sequential choice task. In such a case, the relational procedure would encourage the presence of distortion because pieces of information would be evaluated in relation to other pieces of information (“interdependently”) during the act of decision making. To test the hypothesis that such a relational procedure may be associated with the emergence of distortion, the present work first relies on recent methodological work on mindset priming (for a review, see Wyer & Xu, 2010). In this study, we restrict the definition of the term “mindset” to a cognitive procedure that can be activated in one domain and can influence decision making in a different subsequent situation (Gollwitzer, 1990; Gollwitzer, Heckhausen, & Steller, 1990; Gollwitzer & Kinney, 1989; Xu & Wyer, 2012). For example, making comparative judgments in one domain can activate a “which-to-choose” mindset that disposes consumers to decide which of two products to buy in a subsequent situation without considering the possibility of buying neither product (Xu & Wyer, 2007, 2008). Similarly, we posit that it may be possible to activate a relational mindset in a first task and to observe whether this activation affects distortion in a subsequent unrelated task. Second, the present work also relies on recent empirical work on relational thinking (Vendetti, Wu, & Holyoak, 2014), which has shown that generating solutions for cross-domain analogies in a first task activates a relational mindset that induces decision makers to identify relationships between seemingly dissimilar items in a second task. In the work of Vendetti et al. (2014), the second task required participants to find more relational matches between two dissimilar pictures. In our work, we predict that solving crossdomain analogies in an initial task should contribute to increasing information distortion in a second, unrelated choice task by inducing participants to make connections between their emerging preference and their evaluation of new information in a sequential choice task.
2. Relational reasoning, creativity, and construal level theory Prior research has demonstrated a close link between cross-domain analogical reasoning and creative thinking (Chan, Paletz, & Schunn, 2012; Dahl & Moreau, 2002; Holyoak & Thagard, 1995). The primary reason for such a close link is the property of cross-domain analogies that induces decision makers to make connections between ideas drawn from disparate domains, which subsequently fosters creative outputs. Therefore, the same relational mindset that may encourage creative thinking may also trigger information distortion, which requires making connections between one's prior preferences and the evaluation of new information. In addition, prior research has emphasized that cross-domain analogical reasoning activates abstract thinking (Gick & Holyoak, 1983; Knowlton, Morrison, Hummel, & Holyoak, 2012) because crossdomain analogies require participants to find abstract relationships between disparate domains. Consider a within-domain analogy such as nose:scent::tongue:taste. The solution to this analogy requires the participant to map an identical relation within each pair of items (the nose is the sense organ for scent, and the tongue is the sense organ for taste). Now consider a cross-domain analogy such as nose:scent::antenna:signal. In contrast to the solution to the withindomain analogy, this cross-domain analogy requires the participant to draw an abstract relation between both pairs of items that will bridge both domains (the nose detects scent as an antenna detects a signal). Therefore, cross-domain analogical reasoning yields not only relational reasoning but also, more precisely, abstract relational reasoning. Based on the relationships among cross-domain analogical thinking, abstract relational reasoning and creativity, this study proposes a link between the activation of a relational mindset and construal level theory (Liberman & Trope, 1998). Construal level theory posits that individuals represent psychologically distant events with abstract, general, high-level construals and represent psychologically near events with concrete, contextual, low-level construals (Liberman & Trope, 2008). In turn, a high level of construal activates holistic thinking (Smith & Trope, 2006) and fosters creativity by facilitating the association of distant concepts (Ward, 1995). Because solving cross-domain analogies requires “far-out thinking” (as named by Vendetti et al., 2014) by inducing people to draw abstract connections between semantically distant pairs of items, we propose that generating solutions to cross-domain analogies activates a high level of construal. As such, a relational mindset would induce more information distortion from participants in a choice task because of the adoption of a high-level, holistic, abstract type of reasoning. That is, the level of construal is proposed as a mediator of the relationship between the activation of a relational mindset and distortion. 3. Relational mindset and effort A third objective of this research is to investigate whether cognitive effort is necessary for a relational mindset to affect distortion. That is, we investigate whether cognitive effort is a moderator of the relationship between relational mindset and distortion. This investigation is motivated by earlier research by Waltz, Lau, Grewal, and Holyoak (2000), who showed that imposing a cognitive load on analogical reasoning results in fewer relational mappings between two visual scenes. In other words, relational thinking is impaired under conditions of low effort (in contrast to baseline conditions). If relational thinking is impaired by a cognitive load and if relational thinking is associated with greater distortion, then information distortion should be lower when effort is manipulated to be low (i.e., when a cognitive load is imposed). The unexpected consequence of such a result is that distortion as a bias may be less likely to appear when cognitive effort is low, as long as a relational mindset is activated. Interestingly, Polman and Russo (2012) showed that imposing a cognitive load leads to increased levels of distortion. However, this
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7
finding does not necessarily contradict the present hypothesis. Indeed, Polman and Russo (2012) observed the impact of a cognitive load under neutral, baseline conditions, whereas the present research predicts the impact of a cognitive load when a relational mindset is activated. That is, we do not predict a main effect of effort (as Polman and Russo did); rather, we expect a moderating role of effort when a relational mindset is activated. What is the theoretical consequence of such a distinction for the present research? Polman and Russo (2012) explained the effect of low effort on distortion by defending the idea that imposing a cognitive load activates a type of “System 1” reasoning (compared with System 2; for reviews see Evans, 2007; Kahneman & Frederick, 2002; Keren & Schul, 2009), which one of the features is to be associative. The authors explained as follows: “The predecisional distortion of information should be inhibited by a process that is more analytical and deliberative and, inversely, should be augmented by a process that is more associative” (p. 81; Polman & Russo, 2012). In other words, the researchers attributed their results to the associative nature of System 1, which should be activated when a cognitive load is imposed. In the present research, we activate a relational mindset; that is, we actually activate a cognitive procedure that is associative by nature. Impairing the action of the cognitive procedure (by imposing a cognitive load) should in turn reduce the associative action of the relational mindset and may thereby reduce distortion. On a larger scale, such prediction raises the question of whether all associative processes are necessarily System 1. The present research and hypotheses hint at the idea that it may not, and that different types of associative processes may exist and (at least in the case of distortion) trigger similar results. 4. Objectives and studies
3
domain (cross-domain) analogy required the participants to focus on matching identical relations within the same domain (across domains). For instance, an analogy such as nose:scent::tongue:taste would qualify as a within-domain analogy because the C:D pair is semantically close to the A:B pair. By contrast, nose:scent::antenna:signal is a crossdomain analogy because the two pairs belong to different semantic categories. In the second part of the study, the participants chose between two options (athletic shoes labeled as Product J and Product K) based on five attributes that were presented sequentially (e.g., Russo et al., 1998). After reading each attribute, the participants were asked two questions. The first involved a rating of the attribute on a 9-point scale in which 1 indicated “strongly favors” one product, 9 indicated “strongly favors” the other product, and 5 indicated “favors neither product.” This response formed the basis of the measurement of information distortion. Second, knowing that more information would subsequently be provided, the participants were asked which product they would choose if they were to select one product at this stage. This response identified the currently leading option. Notably, all attributes presented during the choice task were pretested in an independent study conducted with 50 participants. In this pre-test, all participants were asked to rate the same attributes on the same 9-point scale, but the five attributes were said to be describing 10 different products; that is, the labels used to describe the two products varied for each attribute. Varying labels prevented the development of a preference for one product and in turn prevented information distortion (which requires a prior preference). The pre-test confirmed that no attribute evaluation differed significantly from the scale mid-point, 5 (all ps N .15). In other words, all attributes were “neutral” when seen independently, i.e., they did not favor either product.
In sum, the present research has three objectives: 1. To show that distortion is elevated by the activation of a relational mindset, 2. To establish that the relational mindset affects distortion through the activation of a high construal level, and 3. To investigate whether low effort (by imposing a cognitive load) reduces the impact of a relational mindset on information distortion. Three studies were run. Study 1 establishes the main effect of the relational mindset on distortion (first objective). Study 2 investigates the mediation through construal level (second objective). Study 3 examines the moderating role of effort (third objective). The final sample size of minimum 60 participants in each condition of each study was consistent with the required sample size, which had been found in advance to be 58 participants for a power of .70, an expected .70 unit difference in distortion between the two main experimental conditions (control vs. relational mindset), and an expected standard deviation of 1.5. 5. Study 1 5.1. Method Participants were recruited online (N = 120, U.S. residents only) using Amazon's Mechanical Turk, and they were compensated 65 cents for their participation. All participants first generated solutions to 20 four-term verbal analogy problems, such as nose:scent::tongue:taste. All problems were presented in an A:B::C:D format with the D term left blank. Immediately afterward, all participants completed a binary choice task between two athletic shoes. The stimulus for the analogy task was identical to the stimulus used by Green, Kraemer, Fugelsang, Gray, and Dunbar (2010) and Vendetti et al. (2014). More specifically, two types of analogies were used: cross-domain or within-domain analogies. The type of analogy was a between-subjects factor, and the participants were randomly assigned to one of the two conditions. Generating a solution to a within-
5.2. Results The calculation of information distortion required two steps (Russo et al., 1998). First, using the 9-point scale, a participant's estimate of an attribute's value was compared to the unbiased evaluation of identical information, which had been pretested to be 5. Second, the absolute difference between those two values was given a positive sign if the attribute's rating favored the leading alternative and a negative sign if the rating favored the trailing alternative. An average distortion score was then computed across the attributes. We expected participants who generated solutions to crossdomain analogies to distort information more than those who solved within-domain analogies. Confirming this expectation, we found that the mean information distortion score was significantly higher in the cross-domain analogies condition (M = 1.52, SD = 1.52) than in the within-domain analogies condition (M = .81, SD = 1.46; t(118) = − 2.63, p b .01). A MANOVA of the distortion scores following the presentation each attribute confirmed the main effect of the between-subjects factor (F(1, 118) = 6.95; p b .01) and revealed a main effect of the within-subjects factor: distortion tended to increase over the course of the choice process (F(3, 116) = 2.89; p b .05).
6. Study 2 The results of Study 1 confirm that activating a relational mindset increases distortion. Study 2 investigated whether this result would be mediated by construal level. Specifically, we predicted that activating a relational mindset with cross-domain analogies would induce to participants to think at a higher level of construal, which would in turn contribute to increased distortion.
4
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7
6.1. Method A total of 210 U.S. residents participated in a 20-minute study with a between-subjects design and three conditions: within-domain analogies, cross-domain analogies, and a control group. The participants were recruited on Amazon Mechanical Turk in exchange for 65 cents. The two analogy conditions were identical to those used in Study 1. The additional control group wrote two short essays on neutral topics (the Statue of Liberty and the Great Depression of 1929). After the priming task, all participants completed the Behavioral Identification Form (BIF; Vallacher & Wegner, 1989), which asked them to identify 25 actions (e.g., I am talking to a child) with either a low-level statement (e.g., I am using simple words) or a high-level statement (I am teaching a child something). The average number of high-level statements made by participants was used as a measure of construal level (with higher scores indicating higher construal levels). Finally, all participants completed a choice task similar to the task in Study 1, but in this case, the choice was between two pairs of headphones described by five product attributes: technical performance, music trials, pros and cons, design, and editor review. As in Study 1, each attribute had been pre-tested to be neutral (i.e., to not be significantly different from 5 on the relative information evaluation scale; all ps N .15) such that the attributes would not favor either of the two products.
relational thinking and distortion, we conducted a mediation analysis with 5,000 bootstrap samples. The control group was omitted, and the mediation analysis was conducted with the within vs. cross-domain analogies as a dummy predictor variable. The indirect effect reached a bootstrap confidence interval of [.0179; .4013], which does not contain zero and therefore indicates significance. The results of the mediation analysis are presented in Fig. 1. 7. Study 3 The objective of Study 3 was to investigate the moderating role of effort in the relationship between relational mindset and distortion. Specifically, we hypothesized that the effect of the relational mindset on distortion is reduced when effort is low, as cross-domain analogical thinking has been shown to be effortful (Waltz et al., 2000). When a relational mindset is activated, distortion should therefore be lower when effort is manipulated to be low. To manipulate effort, we imposed a cognitive load (low versus high) on the participants during the choice task (e.g., Epley & Gilovich, 2006; Polman & Russo, 2012). Notably, in the control condition in which a relational mindset is not activated, imposing a cognitive load should increase distortion if our results are consistent with those of Polman and Russo (2012). That is, imposing a cognitive load should activate a System 1 type of reasoning, which may contribute to increased distortion. Hence, a cross-over interaction effect of relational mindset and effort on distortion was predicted.
6.2. Results 7.1. Method Information distortion was computed following the procedures in Study 1, resulting in one aggregated distortion score for each participant. The results of this study replicated the results of Study 1. Overall, participants who completed cross-domain analogies tended to distort information more (M = 1.23, SD = 1.21) than participants who completed within-domain analogies (M = .72, SD = 1.59). Even when we controlled for multiple comparisons (using Hsu's multiple comparisons with the best method — Hsu MCB), the difference between the two conditions reached significance (t(137) = 2.17, p b .05). The score for the control group was between the scores for the two experimental conditions but was not significantly different from the score for the cross-domain analogies condition (M = .83, SD = 1.28). Second, the participants in the cross-domain analogies condition tended to show a higher level of construal (M = .70, SD = .22) than the participants in the two other conditions (M = .61 and SD = .25 for the control group; M = .54 and SD = .24 for the within-domain analogies condition). Even when we applied Hsu's MCB method, significant differences were observed between the cross-domain analogies condition and the within-domain analogies condition (t(137) = 3.90, p b .0001) and between the control condition and the cross-domain analogies condition (t(139) = −2.2, p b .0001). Finally, we observed a positive relationship between the level of construal and information distortion: a higher construal level induced more distortion (b = 1.14 t = 3.04, p b .005). To more closely examine the role of construal level as a mediator of the relationship between
A total of 360 U.S. residents participated in a 20-minute study on Amazon Mechanical Turk in exchange for 75 cents. The experiment was a 3 (priming: cross-domain analogies vs. within-domain analogies vs. control) × 2 (load: high vs. low) between-subjects design. The analogies differed from those used in Studies 1 and 2 in order to replicate the effect with different sets. The control group wrote two short essays on neutral topics (the Statue of Liberty and the Great Depression of 1929) rather than solving analogies. The cognitive load manipulation was then administered. Immediately after solving the analogies (or writing the essays), the participants in the high-load condition were asked to memorize a string of eight letters (POQWYTNM) while completing the choice task (Epley & Gilovich, 2006). The participants in the low-load condition were also asked to memorize a string of letters, but each string contained only two letters (YU). The instructions clearly indicated that the letter string would need to be recalled after the choice task. The instructions also noted that the letter string should not be written down, as doing so would compromise the purpose of the experiment. The last part of the study consisted of a choice task (similar to Studies 1 and 2) between two vacation resorts. Each resort was described by five attributes: atmosphere, amenities, location, travel guide descriptions, and club activities. As in Studies 1 and 2, each attribute had been pre-tested with 50 participants to ensure that no attribute evaluation differed significantly from the scale mid-point, 5 (all ps N .15). At
Level of Construal b = .16***
b = 1.27 **
b = .50* (.33) Relational Mindset
Information Distortion
Fig. 1. Mediation analysis in Study 2 (N = 68 in the cross-domain analogies condition and N = 70 in the within-domain analogies condition). The values in parentheses indicate the direct effect of the interaction on information distortion when controlling for the level of construal. *p b .05, **p b .01, ***p b .0001.
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7
the end of the choice task, all participants were asked to recall the letter string that that they had been told to memorize. 7.2. Results First, we examined the results when cognitive load was low (i.e., in baseline conditions of effort) to confirm the results of Studies 1 and 2. When cognitive load was low, information distortion was lower in the within-domain analogies condition (M = .70, SD = 1.42) than in the cross-domain analogies condition (M = 1.42, SD = 1.67). The difference reached significance (correcting for multiple comparisons using Hsu's MCB method), t(177) = −2.46, p b .05. Again, the score for the control group was between the scores for the experimental conditions (M = .99, SD = 1.80) but did not significantly differ from the score for the cross-domain analogy condition (p N .05). Therefore, the results of Studies 1 and 2 were replicated. Second, we examined the effect of the interaction between cognitive load and condition on distortion. As predicted, a full-factorial ANOVA on information distortion with cognitive load and conditions (withindomain analogies vs. control vs. cross-domain analogies) as the main factors yielded a significant interaction effect (F(2, 354) = 4.43, p b .05). When a relational mindset was induced (i.e., in the crossdomain analogy condition), the participants in the low-load condition distorted information more than the participants in the high-load condition (Mlow = 1.42, SDlow = 1.67 vs. Mhigh = .81, SDhigh = 1.33; t(122) = 2.26, p b .05). In other words, reducing effort weakened the effect of the relational mindset on distortion. Conversely, when no relational mindset was induced (i.e., in the control and within-domain analogy conditions), the participants in the low-load condition distorted information less than the participants in the high-load condition (Mlow = .84, SDlow = 1.62 vs. Mhigh = 1.25, SDhigh = 1.56; t(234) = − 2.01, p b .05). This result replicated Polman and Russo's findings: in the absence of the activation of a relational mindset, imposing a cognitive load increases the extent to which decision makers distort information. The results of Study 3 confirm that impairing the activation of the relational mindset (with a cognitive load) lessens its impact on distortion. However, when no relational mindset is activated, less effort increases distortion, possibly because of the automatic activation of System 1 (Polman & Russo, 2012). 8. Discussion 8.1. Summary of results This research shows that priming a relational type of reasoning increases distortion through the activation of a higher level of construal. As such, the same mechanism that has been shown to promote creative thinking (i.e., recognizing relationships across concepts) may also induce more biased information processing by prompting individuals to process independent units of information interdependently. The present research also shows that this process of information integration is relatively effortful, as imposing a cognitive load during the choice task impairs the impact of the relational mindset on distortion. Across our studies, the effect size of the relational mindset on distortion ranged from .36 to .48 (see Table 1A). Overall, the demonstrated
5
effect on distortion is moderate, and we recognize that the power of relational thinking to drive the same effects in a natural setting may be weaker. The results prompt the question of naturally occurring primes, those that are conceptually similar to the relational mindset but more likely to occur spontaneously. In parallel, the demonstrated effect on construal level is relatively high (.69) but is comparable to seminal papers using the BIF as a dependent variable with more traditional construal level priming (see Table 1B). Without a complete meta-analysis, it is difficult to determine the relative effectiveness of cross-domain analogical thinking for construal level with other methods, such as psychological distance (Liberman & Trope, 1998) or why/how reasoning (Freitas, Gollwitzer, & Trope, 2004). However, we can reasonably state that relational thinking activated by crossdomain analogies appears to yield results that are sufficiently strong to provide a meaningful way to yield a high level of construal. 8.2. Discussion of findings in relation to prior research The present paper does not claim that a relational mindset must be activated to increase distortion. Rather, this study presents one way in which information distortion can be increased. For instance, Study 3 shows that imposing a cognitive load under neutral conditions also increases information distortion (see Polman & Russo, 2012). As noted in the introduction, imposing a cognitive load in the absence of the activation of a relational mindset may activate a more “System 1” type of reasoning, which is associative by nature and therefore encourages distortion. Thus, different contexts or mechanisms that encourage relational thinking may result in the same outcome (i.e., more distortion). Some mechanisms may be effortful (as in our case), and some may be more automatic (as is assumed with System 1). Additionally, it is important to distinguish the present results with findings obtained in the self-regulatory literature (Freitas et al., 2004; Fujita, Gollwitzer, & Oettingen, 2007; Taylor & Gollwitzer, 1995) regarding the effect of deliberative and implemental mindsets on judgmental biases and, in particular, on selective information processing. For instance, Fujita et al. (2007) showed that a deliberative mindset (compared with an implemental mindset) increases open-mindedness to processing incidental information, thereby leading decision makers to develop more accurate judgments. In our work, we do not directly test whether participants are in an implemental or deliberative mindset. On the one hand, activating a relational mindset may increase the tendency of participants to compare information and therefore to engage in an even more deliberative type of reasoning. On the other hand, priming a relational mindset could actually induce participants to reach a decision sooner because they are more prone to regard units of information as consistent with one another. Thereby, subjects may switch to a more implemental mindset (i.e., more selective information processing) earlier in the choice process to defend their decisions. Without further testing, it is therefore difficult to predict which of the two mindsets (deliberative or implemental) is more closely associated with a relational mindset and at which stages of the decision-making process. Interestingly, the current work may help explain two findings from previous studies. Because positive affect has recently been shown to promote a high level of construal (Pyone & Isen, 2011), the present results may help explain why positive affect increases distortion
Table 1A Effect sizes in Studies 1, 2 and 3. Study
Dependent variable
M1 within-domain analogies
M2 cross-domain analogies
SD1 within-domain analogies
SD2 cross-domain analogies
Pooled SD
Effect size
1 2 2 3
Distortion Distortion BIF — average high-level items out of 25 items Distortion
0.81 0.72 0.54 0.7
1.52 1.23 0.70 1.42
1.46 1.59 0.24 1.42
1.52 1.21 0.22 1.66
1.49 1.41 0.23 1.54
0.48 0.36 0.69 0.47
6
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7
Table 1B Effect sizes in articles using the Behavioral Identification Form. Source
Manipulation
Dependent variable
M1 low construal
M2 high construal
Pooled SD
Effect size
Liberman & Trope (1998) (Study 1, N = 32) Fujita, Henderson, Eng, Trope, & Liberman (2006) (Study 1, N = 68) Liviatan, Trope, & Liberman (2008) (Study 1, N = 34) Pyone & Isen (2011) (Study 1, N = 41) Alter, Oppenheimer, & Zemla (2010) (Study 3A, N = 80)
Psychological distance Psychological distance
BIF — count of high-level items out of 19 items BIF — count of high-level items out of 13 items
10.19 8.47
13.44 9.88
3.59 2.58
0.90 0.55
Interpersonal similarity
BIF — count of high-level items out of 19 items
11.59
14.65
4.36
0.70
Positive affect Why vs. how
BIF — count of high-level items out of 26 items BIF — preference for high-level items on a 7-point scale
12.85 3.49
16.19 4.67
4.55 0.71
0.73 1.66
(Meloy, 2000). In addition, decisions made for others (as opposed to decisions made for oneself) have been shown to be more creative because of the activation of a high level of construal (Polman & Emich, 2011). Therefore, the present results may also help explain why decisions made for others (as opposed to decisions made for oneself) trigger more predecisional distortion (Polman, 2010). Although this conjecture remains speculative, the present research appears to corroborate the findings of previous studies in this regard. References Alter, A.L., Oppenheimer, D.M., & Zemla, J.C. (2010). Missing the trees for the forest: A construal level account of the illusion of explanatory depth. Journal of Personality and Social Psychology, 99(3), 436–451. http://dx.doi.org/10.1037/a0020218. Brownstein, A.L., Read, S.J., & Simon, D. (2004). Bias at the racetrack: Effects of individual expertise and task importance on predecision reevaluation of alternatives. Personality and Social Psychology Bulletin, 30(7), 891–904. http://dx.doi.org/10.1177/ 0146167204264083. Carlson, K.A., & Pearo, L.K. (2004). Limiting predecisional distortion by prior valuation of attribute components. Organizational Behavior and Human Decision Processes, 94(1), 48–59. http://dx.doi.org/10.1016/j.obhdp.2004.02.001. Carlson, K.A., & Russo, J.E. (2001). Biased interpretation of evidence by mock jurors. Journal of Experimental Psychology. Applied, 7, 91–103. Chan, J., Paletz, S.B., & Schunn, C.D. (2012). Analogy as a strategy for supporting complex problem solving under uncertainty. Memory & Cognition, 40, 1352–1365. http://dx. doi.org/10.3758/s13421-012-0227-z. Chaxel, A. S., Russo, J. E., & Wiggins, C. E. (2015). A Goal-Priming Approach to Cognitive Consistency: Applications to Judgment, Journal of Behavioral Decision Making. . http://dx. doi.org/10.1002/bdm.1874 (in press). Dahl, D.W., & Moreau, P. (2002). The influence and value of analogical thinking during new product ideation. Journal of Marketing Research, 39, 47–60. http://dx.doi.org/10. 1509/jmkr.39.1.47.18930. Epley, N., & Gilovich, T. (2006). The anchoring-and-adjustment heuristic: Why the adjustments are insufficient. Psychological Science, 17, 311–318. http://dx.doi.org/10.1111/j. 1467-9280.2006.01704.x. Evans, J.S.B.T. (2007). On the resolution of conflict in dual process theories of reasoning. Thinking and Reasoning, 13(4), 321–339. http://dx.doi.org/10.1080/ 13546780601008825. Freitas, A.L., Gollwitzer, P., & Trope, Y. (2004). The influence of abstract and concrete mindsets on anticipating and guiding others' self-regulatory efforts. Journal of Experimental Social Psychology, 40(6), 739–752. http://dx.doi.org/10.1016/j.jesp. 2004.04.003. Fujita, K., Gollwitzer, P.M., & Oettingen, G. (2007). Mindsets and pre-conscious openmindedness to incidental information. Journal of Experimental Social Psychology, 43, 48–61. http://dx.doi.org/10.1016/j.jesp.2005.12.004. Fujita, K., Henderson, M.D., Eng, J., Trope, Y., & Liberman, N. (2006). Spatial distance and mental construal of social events. Psychological Science, 17(4), 278–282. http://dx. doi.org/10.1111/j.1467-9280.2006.01698.x. Gick, M.L., & Holyoak, K.J. (1983). Schema induction and analogical transfer. Cognitive Psychology, 15, 1–38. http://dx.doi.org/10.1016/0010-0285(83)90002-6. Gollwitzer, P.M. (1990). Action phases and mind-sets. In E.T. Higgins, & R.M. Sorrentino (Eds.), Handbook of Motivation and Cognition: Foundations of Social Behavior, vol. 2. (pp. 53–92). New York: Guilford Press. Gollwitzer, P.M., Heckhausen, H., & Steller, B. (1990). Deliberative versus implemental mind-sets: Cognitive tuning toward congruous thoughts and information. Journal of Personality and Social Psychology, 59, 1119–1127. http://dx.doi.org/10.1037/00223514.59.6.1119. Gollwitzer, P.M., & Kinney, R.F. (1989). Effects of deliberative and implemental mind-sets on illusion of control. Journal of Personality and Social Psychology, 56, 531–554. http:// dx.doi.org/10.1037/0022-3514.56.4.531. Green, A.E., Kraemer, D.J., Fugelsang, J.A., Gray, J.R., & Dunbar, K.N. (2010). Connecting long distance: Semantic distance in analogical reasoning modulates frontopolar cortex activity. Cerebral Cortex, 20(1), 70–76. http://dx.doi.org/10.1093/cercor/bhp081. Holyoak, K.J., & Thagard, P. (1995). Mental Leaps: Analogy in Creative Thought. Cambridge, MA: MIT Press.
Kahneman, D., & Frederick, S. (2002). Representativeness revisited: Attribute substitution in intuitive judgment. Heuristics and Biases: The Psychology of Intuitive Judgment, 49. Keren, G., & Schul, Y. (2009). Two is not always better than one. A critical evaluation of two-system theories. Perspectives on Psychological Science, 4(6), 533–550. http://dx. doi.org/10.1111/j.1745-6924.2009.01164.x. Knowlton, B.J., Morrison, R.G., Hummel, J.E., & Holyoak, K.J. (2012). A neurocomputational system for relational reasoning. Trends in Cognitive Sciences, 16, 373–381. http://dx. doi.org/10.1016/j.tics.2012.06.002. Kostopoulou, O., Russo, J.E., Keenan, G., Delaney, B.C., & Douiri, A. (2012). Information distortion in physicians' diagnostic judgments. Medical Decision Making, 32(6), 831–839. http://dx.doi.org/10.1177/0272989X12447241. Liberman, N., & Trope, Y. (1998). The role of feasibility and desirability considerations in near and distant future decisions: A test of temporal construal theory. Journal of Personality and Social Psychology, 75(1), 5–18. http://dx.doi.org/10.1037//00223514.75.1.5. Liberman, N., & Trope, Y. (2008). The psychology of transcending the here and now. Science, 322, 1201–1205. http://dx.doi.org/10.1126/science.1161958. Liviatan, I., Trope, Y., & Liberman, N. (2008). Interpersonal similarity as a social distance dimension: Implications for perception of others' actions. Journal of Experimental Social Psychology, 44(5), 1256–1269. http://dx.doi.org/10.1016/j.jesp.2008.04.007. Meloy, M. (2000). Mood‐driven distortion of product information. Journal of Consumer Research, 27, 345–359. http://dx.doi.org/10.1086/317589. Polman, E. (2010). Information distortion in self-other decision making. Journal of Experimental Social Psychology, 46, 432–435. http://dx.doi.org/10.1016/j.jesp.2009. 11.003. Polman, E., & Emich, K.J. (2011). Decisions for others are more creative than decisions for the self. Personality and Social Psychology Bulletin, 37, 492–501. http://dx.doi.org/10. 1177/0146167211398362. Polman, E., & Russo, J.E. (2012). Commitment to a developing preference and predecisional distortion of information. Organizational Behavior and Human Decision Processes, 119, 78–88. http://dx.doi.org/10.1016/j.obhdp.2012.05.004. Pyone, J.S., & Isen, A.M. (2011). Positive affect, intertemporal choice, and levels of thinking: Increasing consumers' willingness to wait. Journal of Marketing Research, 48, 532–543. http://dx.doi.org/10.1509/jmkr.48.3.532. Read, S.J., & Miller, L.C. (1994). Dissonance and balance in belief systems: The promise of parallel constraint satisfaction processes and connectionist modeling approaches. In R.C. Schank, & E.J. Langer (Eds.), Beliefs, Reasoning, and Decision Making: Psycho-logic in Honor of Bob Abelson. Hillsdale, NJ: Erlbaum. Russo, J.E., Carlson, K.A., Meloy, M.G., & Yong, K. (2008). The goal of consistency as a cause of information distortion. Journal of Experimental Psychology. General, 137, 456–470. http://dx.doi.org/10.1037/a0012786. Russo, J.E., Medvec, V.H., & Meloy, M.G. (1996). The distortion of information during decisions. Organizational Behavior and Human Decision Processes, 66, 102–111. http://dx.doi.org/10.1006/obhd.1996.0041. Russo, J.E., Meloy, M.G., & Medvec, V.H. (1998). Predecisional distortion of product information. Journal of Marketing Research, 35, 438–452. http://dx.doi.org/10.2307/ 3152163. Russo, J.E., Meloy, M.G., & Wilks, T.J. (2000). Predecisional distortion of information by auditors and salespersons. Management Science, 46(1), 13–27. http://dx.doi.org/10. 1287/mnsc.46.1.13.15127. Simon, D., & Holyoak, K.J. (2002). Structural dynamics of cognition: From consistency theories to constraint satisfaction. Personality and Social Psychology Review, 6, 283–294. http://dx.doi.org/10.1207/S15327957PSPR0604_03. Smith, P.K., & Trope, Y. (2006). You focus on the forest when you're in charge of the trees: Power priming and abstract information processing. Journal of Personality and Social Psychology, 90, 578–596. http://dx.doi.org/10.1037/0022-3514.90.4.578. Spellman, B.A., Ullman, J.B., & Holyoak, K.J. (1993). A coherence model of cognitive consistency: Dynamics of attitude change during the Persian Gulf War. Journal of Social Issues, 49, 147–165. http://dx.doi.org/10.1111/j.1540-4560.1993.tb01185.x. Taylor, S.E., & Gollwitzer, P.M. (1995). Effects of mindset on positive illusions. Journal of Personality and Social Psychology, 69(2), 213–226. http://dx.doi.org/10.1037/00223514.69.2.213. Vallacher, R.R., & Wegner, D.M. (1989). Levels of personal agency: Individual variation in action identification. Journal of Personality and Social Psychology, 57, 660–671. http:// dx.doi.org/10.1037/0022-3514.57.4.660. Vendetti, M.S., Wu, A., & Holyoak, K.J. (2014). Far-out thinking: Generating solutions to distant analogies promotes relational thinking. Psychological Science, 25, 928–933. http://dx.doi.org/10.1177/0956797613518079.
A.-S. Chaxel / Journal of Experimental Social Psychology 60 (2015) 1–7 Waltz, J.A., Lau, A., Grewal, S.K., & Holyoak, K.J. (2000). The role of working memory in analogical mapping. Memory and Cognition, 28, 1205–1212. http://dx.doi.org/10.3758/ BF03211821. Ward, T.B. (1995). What's old about new ideas? In S.M. Smith, T.B. Ward, & R.A. Finke (Eds.), The Creative Cognition Approach (pp. 157–178). Cambridge, MA: MIT Press. Wyer, R.S., & Xu, A.J. (2010). The role of behavioral mind-sets in goal-directed activity: Conceptual underpinnings and empirical evidence. Journal of Consumer Psychology, 20(2), 107–125. http://dx.doi.org/10.1016/j.jcps.2010.01.003.
7
Xu, A., & Wyer, R., Jr. (2007). The effect of mind‐sets on consumer decision strategies. Journal of Consumer Research, 34(4), 556–566. http://dx.doi.org/10.1086/519293. Xu, A.J., & Wyer, R.S. (2008). The comparative mind-set: From animal comparisons to increased purchase intentions. Psychological Science, 19, 859–864. http://dx.doi.org/ 10.1111/j.1467-9280.2008.02169.x. Xu, A.J., & Wyer, R.S., Jr. (2012). The role of bolstering and counterarguing mind-sets in persuasion. Journal of Consumer Research, 38(5), 920–932. http://dx.doi.org/10. 1086/661112.