How does imagery in interactive consumption lead to false memory? A reconstructive memory perspective

Consumers are often exposed to information that may contradict their consumption experience, leading to “false memory” for aspects of the consumption. We develop a theory suggesting why false memory may be heightened due to imagery processing. We subject this theory to a rigorous test in three experiments. We find that post-consumption imagery makes it more difficult to discern between consumption and misleading information thereby boosting false recall. Yet, imagery that is more consumption-based helps reduce consumers' susceptibility to false memory. Consistent with this notion, we explicate the role that attention plays at different stages of consumption in affecting false memory. © 2009 Society for Consumer Psychology. Published by Elsevier Inc. All rights reserved. Consumers often encounter information that contradicts their consumption experience. Contradictory informationmay be seen while answering surveys or participating in message boards, online discussion groups and/or social networking sites. Even conversations with other consumers or salespeople may expose us to misleading information. Conventional wisdom suggests that consumption experiences lead to lasting impressions that are impervious to misinformation. Yet, anecdotal evidence offers evidence to the contrary. The commotion surrounding “unpatchable” flaws in the Firefox browser during an industry conference highlights this problem (Lyman, 2006). Although untrue, the speed with which news about this “problem” spread and the outpouring of concern among users—even after retraction of the report—underlines the potential effect of postconsumption misinformation on consumer memory. Academic research demonstrates that exposure to misleading information leads to a post-consumption misinformation effect (Braun, 1999). Findings suggest that while evaluations are susceptible to misinformation, facts—especially those central to the consumption—may be immune (Braun, 1999; Cowley & Janus, 2004). However, conclusions about the immunity of facts centrally related to consumption are based on two implicit ⁎ Corresponding author. Department of Marketing, University at Buffalo, State University of New York, Amherst, NY 14260, USA. E-mail addresses: [email protected] (A. Lakshmanan), [email protected] (H.S. Krishnan). 1057-7408/$ see front matter © 2009 Society for Consumer Psychology. Publish doi:10.1016/j.jcps.2009.04.005 premises that may not generalize to a significant chunk ofmodern consumption. The first premise is that consumption is largely underpinned by reasoning-based or “discursive” information processing (e.g., ELM, Petty & Cacioppo, 1986; Schlosser, 2003, p. 184). The second, stemming from the first premise, is that when facts are judgments about the product, the centrality of a fact might even sharpen memory—possibly due to greater thought (e.g., Cowley & Janus, 2004). Both these premises are subject to challenge when consumption-related information processing is seen from a broader, more inclusive viewpoint. Modern consumption spans both products that may be viewed from a discursive processing lens (such as food products, financial services, and consumer durables) and those that implicate imagery processing (e.g., Green & Brock, 2002; Wyer, Hung & Jiang, 2008) such as movies, video games, leisure parks and many forms of internet usage. With such products, imagery is not only an integral part of the consumption, but is also intertwined in various postconsumption behaviors. For instance, narrating the experience to a friend, posting on blogs, or responding to market research in these contexts would involve imagery. Indeed, previous research on web-based consumption suggests that inferences using a discursive lens may not readily generalize to imageryrich environments (Schlosser, 2003). When consumption involves imagery, the centrality of a fact may not accord it any particular benefit (in terms of greater thought). This raises the question of memory resilience in such environments. ed by Elsevier Inc. All rights reserved. 452 A. Lakshmanan, H.S. Krishnan / Journal of Consumer Psychology 19 (2009) 451–462 In this paper we investigate the false memory problem using the misinformation experimental paradigm with interactive consumption as the imagery-based consumption context. We find that key facets of consumption may not only be susceptible to false memory, but—when post-consumption imagery is explicitly encouraged—also be heightened because it is difficult to monitor the true memory of consumption versus the images generated via imagery. Yet, imagery that is more focused on the consumption itself may help reduce consumers' susceptibility to false memory due to additional opportunities to discern between competing sources of information. Consistent with this notion, we find that attention devoted to the consumption at retrieval may help reduce false memory. Fig. 1 summarizes our conceptual model. Reconstructive memory and the role of source monitoring The broad theoretical framework guiding false memory research is the idea of reconstructive memory (Bartlett, 1932). This framework suggests that we do not necessarily retrieve our memories about previous events in a manner similar to replaying a video recording. Rather, our seemingly veridical memories are reconstructed from different pieces of information accessed on the basis of retrieval cues. Thus it might include pieces of the picture that have been inferred from the rest of the information available to us. The false memory effect is one specific manifestation of this process whereby post-consumption stimulus information is integrated into the memory for the consumption event during (a later) reconstructive process. Source monitoring theory (Johnson, Hashtroudi, & Lindsay, 1993) may help explain how misinformation seen after consumption becomes integrated with consumption memory. This framework has been used to explain misidentification of event sponsors or brands in advertisements (Johar & Pham, 1999; Pham & Johar, 1997). Johnson et al. (1993) define source monitoring as the set of processes involved in making attributions about the origins of memories, knowledge and beliefs. A failure in source monitoring leads to misattribution between information remembered from a consumption episode versus post-consumption information (such as ads, word of mouth or even erroneous questions). This, in turn leads to the merging of post-consumption information into consumption memory leading to the misinformation effect. Thus, the mix-up between post-consumption information and memory for the Fig. 1. Conceptual Model. consumption occurs because of failure to accurately monitor the source of each piece of information. Experiment 1: Imagery processing and reconstructive memory Imagery-based consumption and source monitoring Mental imagery has been the subject of inquiry in numerous consumer contexts. A core finding is that imagery enables consumers to generate, interpret and manipulate information through spatial representation (Dahl, Chattopadhyay, & Gorn, 1999). According to Dahl et al. (1999), mental images can stem from past experience or from our imagination. Given that the experienced and the imagined share high levels of perceptual detail, imagined experiences may appear as real as true ones (Dahl et al., 1999; Mitchell & Johnson, 2000). Further, the presence of mental images sharing a high level of similar perceptual detail should enhance the likelihood of source misattribution (Schacter, 1996). In most consumption, discursive and imagery processing occur in parallel (Epstein, 1991). However, in non-interactive consumption, discursive processing may be expected to dominate (Schlosser, 2003). In such situations, a central either/or fact—for example, whether a juice is grapefruit or orange—is salient and easy to monitor while less tangible observations (such as how sweet or “orangey” a juice was) are more difficult to pin down. When customers encounter contradicting information after consumption, source misattribution is less likely for the fact. Indeed, presentation of misinformation may in fact, lead to a sharpening of memory—a pattern found by Cowley and Janus (2004). This situation, however, changes quite dramatically when processing is imagery-based as in many interactive contexts (e.g., Holbrook & Hirschmann, 1982) or evaluation of new products (e.g., Zhao, Hoeffler, & Dahl, 2009). In such contexts, consumers generate vivid mental images during consumption. When new target relevant (albeit misleading) information is presented afterwards, the availability of a vivid and embellished mental record of the consumption episode allows for easy integration of this new information with the experience (Jacoby, Hessels, & Bopp, 2001). This integration is the basis for possible source attribution errors during recall (see Fig. 2). Further, the eyewitness testimony literature suggests that in situations where the memory of an imagined event contains a wealth of detail about the context, individuals are more likely to believe that it is a real memory of a true event (Schacter, 1996, p. 116). Thus, source monitoring errors are more likely in richer, more embellished, rather than impoverished contexts since it is easier to “fill in”missing pieces (Bartlett, 1932). Combining the above: H1a. In imagery-rich consumption contexts, exposure (versus no exposure) to misinformation leads to higher false recall of consumption-related facts. Extended imagery and source monitoring Consider now when imagery processing is explicitly encouraged after consumption. The process of generating Fig. 2. Conceptual overview of process. 453 A. Lakshmanan, H.S. Krishnan / Journal of Consumer Psychology 19 (2009) 451–462 imagery after consumption involves the use of a consumptionrelated locus that is accessed from short-term memory. Further, the type of imagery encouraged after consumption may influence the locus chosen. When imagery engenders building a new story involving the product, it is likely that the most recently encountered product-relevant information forms the locus for imagery. Given that the misinformation was most recently seen, it is likely to be more accessible and thus, form the basis for imagery generation. With misinformation as the locus for imagery, the consumer is focused towards building a newmental narrative that is not as well connected with the initial consumption experience. Greater rehearsal of the misinformation during imagery generation leads to better availability of images associated with it and higher likelihood of these images being used for reconstruction later. We know from past research that imagery draws upon cognitive resources and diminishes their availability for other purposes (Unnava & Burnkrant, 1991; Unnava, Agarwal, & Haugtvedt, 1996). Since source memory is more difficult to maintain and retrieve than episodicmemory, sourcemisattribution ismore likely under conditions of diminished cognitive resources (e.g., Johnson et al. 1993; Pham & Johar, 1997). This in turn, should heighten false memory. Thus, explicitly engaging imagery processing via extended imagery should lead to more false memory. In contrast to imagery processing, encouraging discursive processing (say, via cognitive elaboration) has a lower propensity to generate a vivid memory trace of the experience. The relative paucity of vivid images makes the incorporation of misinformation less likely compared to when imagery processing is encouraged. Further, since there is no additional imagery being generated around misinformation, there are fewer misinformation-consistent images to monitor during elaboration and recall. However, since false memory in imagery-rich contexts stems from source confusion between images, encouraging cognitive elaboration after consumption does not by itself lead to better discernment between images. Hence, we expect explicit cognitive elaboration to lead to baseline levels of false memory comparable to groups where such elaboration is not experimentally manipulated. We posit the following hypotheses: H1b. False recall will be higher for the imagery group compared to both the cognitive elaboration and control (noelaboration) groups. H1c. False recall will be equally likely for cognitive elaboration and the control group. These effects appear similar to imagination inflation (e.g., Garry, Manning, Loftus, & Sherman, 1996) however the process differs in two important ways from the traditional views. First, in the imagination inflation paradigm, imagination is used to create a fictitious event which forms the misinformation. In our studies, imagination revolves around consumption and is not involved in creating the misinformation. This separates imagery as a style of processing versus as a creator of misinformation. The second difference pertains to false memory inflation due to imagination focused upon the misinformation (e.g. Braun, Ellis, & Loftus, 2002) versus imagery as a style of processing that underlies consumption. Braun et al. explicitly instruct participants to engage in imagination focused on the target (misleading) ad during ad exposure. In our procedure, consumption occurs without any such instruction. Imagery is engaged after consumption and incidental exposure to misinformation with no explicit focus on the misinformation. These differences reflect our focus on first, separating imagery from the creation and elaboration on misinformation and second, on mapping onto real-world situations. For instance, in product testing consumers are often asked to draw upon their imagination to identify new uses or features. Alternatively, they may be encouraged to use imagery before product use with consequent effects on evaluations (Zhao et al., 2009). Thus, in practice, imagery processing is more likely to be temporally and conceptually dissociated from the creation of misinformation. Our procedure maps on to this practice as a way of building ecological validity. 454 A. Lakshmanan, H.S. Krishnan / Journal of Consumer Psychology 19 (2009) 451–462 Imagery in interactive consumption—process hypothesis In addition to objective recall measures, we adopt subjective judgments of memory—the Remember/Know/Guess measures adapted from Tulving (1985) and used by Braun (1999) and Cowley and Janus (2004)—to interpret the process. These memory judgments provide clues with regard to the phenomenological aspects of remembering (Tulving, 1985). A “remember” judgment indicates a conscious re-experiencing of the original experience. It indicates that the memory is vivid in detail and extremely close towhat the participant believes is her experience. In contrast, the “know” judgment relies on general knowledge gained possibly without consciously accessing the past. It relies more upon the conceptual information gleaned from the experience than the characteristics of the consumption episode. The “guess” option filters out random responses and indicates that participants have little access to the consumption memory. In imagery-rich environments, cognitive elaboration should lead to more remember judgments compared to no-elaboration. This expectation stems from the notion that remember judgments are based on a re-experiencing of the original experience. Further, since imagery-based elaboration involves more vivid and embellished memory, remember judgments are even more likely (compared to cognitive elaboration). When consumers are subject to misinformation, greater familiarity of the misleading theme emerges due to its relative recency compared to the original theme encountered during consumption. This subjective feeling of familiarity is exacerbated due to elaboration and imagery which should lead to more remember judgments. In particular, the remember judgments should be the highest for imagery-based elaboration, followed by cognitive elaboration and then the no-elaboration condition. Thus, ironically, consumers are expected to report more veridical memory when they are most in error. Formally, H2a. The likelihood of remember judgments will be highest when imagery-based elaboration is engaged and least when there is no-elaboration. H2b. These likelihoods will be further heightened when consumers are exposed to misinformation (than when not) such that the greatest inflation will be when imagery is engaged and least when there is no-elaboration.