Economic subjects | Management » Steven-Mirjam-Mandy - How to Study Consciousness in Consumer Research

1 How to Study Consciousness in Consumer Research Steven Sweldens RSM Erasmus University & INSEAD Mirjam A. Tuk Imperial College Business School & RSM Erasmus University Mandy Hütter Eberhard Karls Universität Tübingen Abstract Consumer research can benefit greatly from more insight in unconscious processes underlying behavior. Williams and Poehlman’s effort at more clearly conceptualizing consciousness and call for more research provides a welcome stimulus in this regard. At the same time, providing evidence for unconscious causation is fraught with methodological difficulties. We outline why it is vital to uphold standards of evidence for claims regarding unconscious processes, as it is precisely a lack of rigor on this front which has generated a countermovement by researchers sceptical of dual process models in general and unconscious processes in particular. We contend that the sceptics have offered valid causes for concern, which we leverage to formulate six

concrete recommendations for future research on consciousness. Researchers should (1) specify the process level at which they claim evidence for unconscious processes, (2) not confuse unconscious influences with unconscious processes, (3) carefully choose between different operational definitions of awareness, (4) maximally satisfy four criteria for awareness measures, and (5) complement measurement with experimental manipulations of awareness. Finally, we recommend to (6) refrain from hard claims about unconscious causation that transcend the limitations of the evidence, recognizing that consciousness is a continuous construct. Author Note Steven Sweldens is associate professor of marketing at the Rotterdam School of Management, Erasmus University, Burgemeester Oudlaan 50, 3000 DR Rotterdam, The Netherlands (sweldens@rsm.nl) and distinguished research fellow at INSEAD Mirjam A Tuk is assistant professor of marketing at Imperial College Business School, Imperial College London,

Exhibition Road, London, SW7 2AZ, UK (m.tuk@imperialacuk) and visiting professor at the Rotterdam School of Management, Erasmus University. Mandy Hütter is junior professor of social psychology at the Eberhard Karls Universität Tübingen, Fachbereich Psychologie, Schleichstr. 4, 72076 Tübingen, Germany (mandyhuetter@uni-tuebingende) 2 To better understand, aid and protect consumers, it is imperative to have an accurate understanding of unconscious drivers of behavior. We therefore welcome Williams and Poehlman’s (2016) effort to stimulate and more clearly conceptualize the study of consciousness in consumer research. Past research on consciousness has struggled with two major stumbling blocks. First, it is difficult to provide an accurate definition of consciousness, not least because we do not really know how the experience of consciousness originates. As a consequence, there has been much variation in how conscious versus unconscious processing have been defined and

operationalized in past research. We believe Williams and Poehlman (WP) have made important progress here by restricting the definition of consciousness to awareness, highlighting its functions and distinguishing it from other features of automaticity. Second, even when researchers agree on a definition (e.g, “awareness”), it remains difficult to measure accurately On both fronts there are elements we feel are still unsatisfactory in WP’s discussion of consciousness and some accents we would place differently. Mostly these relate to the practical question of how to study consciousness in consumer research. This comment will broadly develop the following three interrelated points. (1) It is vital to uphold and further improve our standards of evidence for unconscious causation. After all, critical evaluations by some of the most knowledgeable scholars in this field have revealed grave deficiencies in the evidence that has so far been presented for unconscious processes, to the

extent that they question whether unconscious processes play a significant role at all in many areas relevant for consumer research (Newell and Shanks 2014). (2) We believe these criticisms should not be regarded as a threat by researchers studying unconscious processes but as an opportunity. We outline six concrete recommendations to improve the quality and communication of evidence for unconscious processes. 3 (3) Development of better measures of awareness and other features of automaticity should be a research priority. However, we also illustrate how the pursuit of better measures of awareness can run into fundamental problems and a perfect measure of awareness is unlikely to be developed. Meanwhile, researchers should recognize the limits of the awareness measures they use, refraining from unwarranted claims given these limitations. THE NEED TO UPHOLD STANDARDS OF EVIDENCE Extraordinary Claims Require Extraordinary Evidence We agree with WP’s call to ‘rein in

consciousness’ and ‘consider consciousness second’ as a way to stimulate new theories and research ideas. WP achieve their second high-level aim to stimulate consumer researchers to build more conceptual integration into their models by more deeply considering how neural and physiological processes inaccessible to consciousness are implicated in consumer behavior (WP, p. 2) Their primary high-level aim, however, is to enhance conceptual rigor in consumer research. It is precisely on this front that we believe important qualifications need to be made. As we explain below, a few elements in their article could have opposite from intended effects if uncritically applied, leading to less-than-desired levels of rigor in future research. Threats to rigor could emerge if WP’s call to ‘rein in consciousness’ would be generalized from the theory generation stage to the theory testing and validation stages (see Baumeister et al., this issue, for other problems with this heuristic) To

their credit, WP “encourage researchers to rigorously examine the standards of evidence brought to bear when assessing whether mental processes have conscious or unconscious influences on behavior” (WP, p. 13) At the same time, however, they endorse the view propagated by Dijksterhuis and colleagues (2014) and Evans (2014) who argue that the standards of evidence required for 4 unconscious processes are often higher than those for conscious processes. Conscious causation, they claim, is often privileged as a null hypothesis and unconscious processes are unfairly ‘penalized’ as alternative hypothesis, requiring stringent evidence. In our opinion, in most situations the burden of proof should lie on claims for unconscious rather than conscious causation. Carl Sagan famously remarked that “extraordinary claims require extraordinary evidence,” an intuition formalized in Bayesian approaches to hypothesis testing where the empirical evidence for a hypothesis needs to be

weighted by that hypothesis’ prior level of likelihood. We believe that claims for unconscious causation are more extraordinary than claims for conscious causation. Consider the following examples No researcher interested in judgment and decision making would doubt the claim that consumers can consciously integrate information and choose between different alternatives by comparing the extent to which the options’ attributes satisfy the consumer’s purchasing goals. However, it is far less obvious that consumers would unconsciously continue to integrate information when their attention is directed elsewhere (see WP’s discussion of consciousness: its primary function is to integrate information; see Plassmann and Mormann (this issue) for more elaboration on the role of attention). Furthermore, it would be downright extraordinary if the choices people make when not even thinking about the alternatives turn out to be superior compared to when they devote their full attentional and

cognitive resources (Dijksterhuis 2004; Waroquier et al. 2009) Similarly, no researcher interested in persuasion processes would doubt the fact that consumers can be influenced by blatant persuasion attempts – a billboard for Coca-Cola® may feed consumers’ belief that Coca-Cola can satisfy their need for refreshment. However, when James Vicary claimed in 1957 that subliminal messages could influence movie-goers to consume 5 more Coca-Cola and popcorn, it caused an uproar and investigation by the CIA (which exposed the fraudulent nature of his claims). This illustrates that, from a consumer protection point of view, unconscious effects will always have graver implications than conscious ones. Finally, we could list many examples where evidence for unconscious causation is presented to rule out much less interesting (conscious) causes such as experimental demand effects. One example is early research on evaluative conditioning (EC) The basic demonstration of the EC effect – a

change in liking of a conditioned stimulus (e.g, brand logo) after repeated pairing with valenced unconditioned stimuli – would not be all too remarkable if respondents would be aware of the contingencies between the stimuli and of the researcher’s hypothesis. As a result, some of the earliest awareness measures were developed to guard precisely against such trivializing explanations (Allen and Janiszewski, 1989; Allen and Madden, 1985; Page 1974; Stuart, Shimp and Engle, 1987). Similar concerns could be leveled against demonstrations of prime-to-behavior effects, mindset effects or identity priming effects. In all such cases it is only appropriate that the burden of proof lies on the side of unconscious causation. The Danger in Considering Consciousness Second We are concerned that a “consider consciousness second” heuristic can lead to a tendency to uncritically categorize research findings as evidence of ‘unconscious processes.’ Some references in WP’s treatment show

evidence of such a mindset. For example, WP refer (among others) to the research by Van den Bergh, Dewitte and Warlop (2008) and by Genevsky and Knutson (2015) as examples of research showing the effect of low-level biological processes and claim that “processes that occur at low-level, neural, or physiological levels (outside of consciousness) can account for unique variance in predictive models of behavior” (WP, p. 7) Whereas we certainly agree that this research is inspiring, effects of low-level biological or 6 physiological factors do not necessarily occur outside of consciousness. For example, Van den Bergh and colleagues demonstrate that men seek more immediate rewards after confrontation with sexually arousing cues. Participants were certainly very aware of the arousing stimuli, might have been aware of their own elevated arousal levels, and maybe even of the fact that physiological arousal could influence their decision making. Similarly, Genevsky and Knutson (2015)

demonstrated that consumers’ affective responses to photographs accompanying microloan requests are an important predictor of market-level microlending decisions. Their research showed that neural imaging data of brain areas involved in affective responses improve prediction accuracy over and above participants’ self-reported affective responses – interpreted by WP as evidence for an unconscious effect of affective responses. However, it is unclear whether the additional variance explained by brain activity data reflects an unconscious process. It could possibly reflect another consciously accessible feature of the photographs that was not measured (e.g, the extent to which they are considered self-relevant). Importantly, neither Genevsky and Knutson (2015) nor Van den Bergh and colleagues (2008) made claims about the (un)conscious nature of the effects they identified. To assume all too easily that these effects are unconsciously generated, simply because physiological factors

are involved, demonstrates the dangers of taking the “consider consciousness second” adage too far. In their treatment of these articles, WP allowed unconscious causation to take the spot of the ‘null hypothesis,’ that is, a hypothesis for which no evidence is required. Taking evidence for unconscious processes for granted will not improve research in this domain. Instead, as we illustrate below, such lack of rigor is precisely what fueled researchers’ skepticism about unconscious influences. Why the Case against Unconscious Processes Merits Attention 7 While there is no doubt that vast amounts of human mental processing occur unconsciously (e.g, the integration of sensory information), there are also a few areas in human cognition where there is much more debate and disagreement about whether unconscious processes play a significant role. Specifically, in recent years a few influential articles were published questioning seemingly abundant evidence for unconscious

processes as (1) underlying human associative learning (Mitchell, De Houwer, and Lovibond 2009; Shanks 2010), and (2) playing a significant role as direct, independent and proximal causes of human judgment and decision making (Kruglanski and Gigerenzer 2011; Newell and Shanks 2014). How can these scholars arrive at such sweeping conclusions in the face of hundreds of scientific articles claiming evidence for unconscious processes in learning, behavior and decision making? As we shall see below, they identified a number of serious limitations common to much of the evidence for unconscious processes, based on which they put into doubt the entire literature. Proponents of unconscious processes have argued that Shanks and colleagues are overly radical and dismissive of large bodies of literature (Dijksterhuis et al. 2014), or apply unfair standards of evidence to unconscious processes (Evans 2014). As a consequence, it can be tempting to dismiss Shanks’ criteria for good awareness

measures as arising from a perspective that puts consciousness central (WP, p. 13) Shanks and colleagues might have been overly radical in their conclusions, yet we also believe it would be a mistake not to take their substantive concerns seriously. Instead, we argue that for our science to progress, we need to regard their (and other) criticisms as an opportunity to, first, improve the design of studies and measures of unconscious processes, and, second, guide authors and reviewers in more carefully interpreting empirical evidence. FROM CRITICISM TO PROGRESS 8 We believe that criticisms raised against evidence for unconscious processes usually fall into one of the following four brackets: not specifying the part of a process assumed to operate unconsciously, confusing distal causes with unconscious processes, failing to clearly conceptualize and operationalize awareness, and using inappropriate measures of awareness. Each type warrants consideration and implies concrete

recommendations to improve consumer research on consciousness. Awareness of What? The Importance of Specifying Awareness at the Process Level Awareness (or lack of it) can take place at different processing levels. These levels can be stimuli, cognitive processes, behaviors, and relations between all of these. For example, Chartrand (2005) describes the case of unconscious goal priming where one could be unaware of the goal prime itself, of the thought process generated by the goal prime, or of the behavior displayed as a result. Similarly, for judgment and decision tasks, Newell and Shanks (2014) use the lens model (Brunswik 1952) to illustrate that one needs to distinguish awareness of decision criteria, of cues and their validities, as well as of cue utilization and the resulting judgments. Specifying the process level where an unconscious effect takes place is highly important for both practical and theoretical reasons. Practically, it is crucial to know which part of a process

operates outside of awareness to design effective countermeasures to protect consumers (Chartrand 2005). Theoretically, demonstrations of unconscious influences can range from trivial to crucial in their theoretical contribution, depending on the process level. In the realm of EC, for example, the debate on ‘awareness’ has been raging for decades (for a review, see Sweldens, Corneille, and Yzerbyt 2014). One reason why the subject generates so much attention is due to its theoretical relevance: since researchers concluded that classical (Pavlovian) conditioning effects in humans are generally not established without participants’ conscious knowledge of 9 contingencies between conditioned and unconditioned stimuli (Brewer 1974; Lovibond and Shanks 2002), demonstrations of unaware associative learning were almost exclusively restricted to EC effects. As a result, the whole premise that humans would be able to learn associations between stimuli unconsciously came to rest on EC

demonstrations. However, not all demonstrations of “unaware EC effects” should be considered equally important. Consider, for example, a researcher who first runs an EC procedure in which a conditioned stimulus (e.g, a brand logo) is paired with positive affective stimuli (eg, gorgeous visuals). Next, the researcher runs an indirect measure of attitudes (eg, an evaluative priming task), demonstrating that the brand logo has acquired positive valence, without participants’ awareness that their attitude towards the brand is being assessed. While such a demonstration of “unaware EC” is useful to dispel experimental demand explanations for the phenomenon, it would do little to convince scholars that unaware association formation between stimuli occurred. The relevant question is rather whether the EC effect was established without participants’ awareness of the contingency between conditioned stimulus and the valence of the unconditioned stimulus at the time of learning. To

this date, such demonstrations remain elusive (Hütter et al. 2012; Stahl, Haaf, and Corneille 2016; Sweldens et al 2014) The point is that evidence for unawareness at one level (e.g, of one’s behavioral or attitudinal response) would be evaluated very differently from evidence for unawareness at another level (e.g, of the relation between stimuli). Hence, our first recommendation is the following: R1: Analyze the process of the phenomenon of interest and be specific about the process level(s) at which you claim an unconscious influence is taking place. Confusing Distal with Proximal Causes or Unconscious Influences with Unconscious Thought Processes 10 There is no doubt that consumers are often not aware of all the factors that contributed to their choices or behavior. However, there lies a danger in confusing distal causal factors of which consumers might not be aware with the operation of unconscious thought processes influencing behavior or decision making. As an example,

Newell and Shanks (2014, p 5) discuss an experiment by Nisbett and Wilson (1977) which formed an early cornerstone in the evidence for unconscious influences on behavior. The experiment shows that consumers prefer the rightmost option in a list of identical consumer products (eg, socks), while when probed for their reasoning, they do not mention or even flatly deny being influenced by the position of the items. Newell and Shanks argue that this finding can be mediated by an entirely conscious decision making strategy during which options are sequentially sampled and consumers apply a heuristic like “if the current item is no worse than the previous item, I’ll prefer the current item.” As long as identical items are sampled from left to right, consumers will end up with the right-most option. Newell and Shanks argue that Nisbett and Wilson confused a distal cause (serial position) with a proximal one (consumers’ decision strategy), so that they argued in favor of unconscious

processes on false premises. Hence, there is a distinction between (distal) factors that have an influence outside of awareness versus unconscious thought processes as conceptualized in dual process theories (e.g, System 1; Kahneman and Frederick 2002) There is no debate on the existence of the former, but much more uncertainty and difficulty in demonstrating the latter. It might seem obvious that factors influencing our decisions outside of awareness are a different matter altogether from unconscious processing. And yet the two are easily confused In WP’s article, for example, the difference is never made explicit. WP refer to dozens of unconscious influences to make the case that consciousness should be nudged toward the background, “allowing low-level neural, physiological, and other unconscious influences on 11 behavior to share the stage” (p. 8, italics added) One could wonder if unconscious influences per se really need more of a stage in consumer research. Is not most

of the research we do (or publish) highlighting an influence that was not obvious to us as consumer researchers at the outset, and therefore probably even less transparent to consumers? Consumers are likely not aware of how the decision context affects their choices (Simonson 1989), of how mood influences their reasoning (Labroo and Patrick 2009), or of how merely deliberating an option already increases loss aversion (Carmon, Wertenbroch, and Zeelenberg 2003). Or consider some of the endocrinological research highlighted by WP, for example, the finding that consumers’ risk-taking can be predicted by their prenatal testosterone levels (Stenstrom et al. 2011) Although consumers could hardly be aware of this influence, this is an indication of a cause so distal (strictly, only a correlation) that it has little bearing on unconscious processing, but is likely mediated by personality development. In sum, we need to realize that demonstrations of unconscious influences (even

endocrinological or neurophysiological ones) do not necessarily offer evidence for the existence of unconscious processing as conceptualized in dual process theories. To pretend that they do will only generate more confusion. Hence, our second recommendation is straightforward: R2: Do not confuse unconscious influences with unconscious processing. What is Awareness? From Definition to Operationalization. WP equate consciousness with awareness, thereby distinguishing it from other features of automaticity. Yet, defining consciousness as awareness presumes we have a clear understanding of what awareness means without referring to consciousness. Unfortunately, consciousness and awareness are often equated and used interchangeably (Moors and De Houwer 2006; Reingold and Merikle 1988). Consumer researchers often speak of “conscious awareness” as if one term 12 clarifies or qualifies the other (e.g, Chartrand 2005; Chartrand et al 2008; Dalton and Huang 2014; Forehand and Perkins

2005). WP similarly use awareness to define consciousness on the one hand, but regularly mention “conscious awareness,” or relate the concept back to consciousness when discussing “consciously accessible thoughts,” highlighting the circularity in definitions (Fiedler and Hütter 2014). Awareness is often referred to as a state of subjective experience, defined for example as “introspective access to mental processes or mental contents” (Gawronski and Bodenhausen 2014, p. 194) Yet, operational definitions which link the construct to measurement are more useful than (rather philosophical because empirically inaccessible) definitions in terms of subjective experiences. When awareness is operationalized via measurement, one can distinguish between subjective, objective and metacognitive operational definitions of awareness (Timmermans and Cleeremans 2015). Subjective operational definitions depend on participants reporting the contents of their thought processes in self-report

measures. The limitation of subjective operationalizations is that self-reports are potentially influenced by other processes, such as consumers’ verbal skills, their interpretation of the question, and compliance. Objective operational definitions depend on participants utilizing their internal knowledge in performance measures (e.g, tasks that require participants to select a stimulus previously seen out of an array of stimuli). A drawback to such measures is that performance may also reflect familiarity or implicit memory (Hütter et al. 2012; Jacoby 1991) A third operational definition of awareness draws on the metacognitive insight in the accuracy of the verbal report or one’s performance. This definition is implemented in measures of confidence or betting tasks that assess the degree to which consumers trust their knowledge (Persaud, McLeod, and Cowey 2007). The drawback of this definition is that metacognitive acuity may be fuelled by both explicit knowledge and 13

intuitive feelings, of which consumers might be aware or unaware, and that consumers differ in their tendency to rely on these signals (Epstein et al. 1996) It is also possible to operationalize awareness via experimental manipulation (e.g, via subliminal presentations or with secondary tasks distracting attention). As we shall see below, both measurement and experimental approaches come with different limitations. Given this conceptual confusion and the multitude of operational definitions, we recommend to start with a theoretically motivated, operational definition of awareness. Researchers tackling awareness should be asking themselves whether they want to investigate whether participants know (measured subjectively or objectively) or whether they know that they know (metacognitive acuity). Awareness measures or manipulations should maximally correspond to this definition Hence: R3: Consider the different types of operational definitions when choosing your awareness measure or

manipulation. Recognize the constraints of each type of operationalization. Operationalization via Measurement: Four Criteria for Measures of Awareness Shanks and colleagues have proposed a set of four criteria for measures of awareness: reliability, relevance, immediacy, and sensitivity (Lovibond and Shanks 2002; Newell and Shanks 2014; Shanks and St John 1994). Together they form perhaps the greatest reason for these authors’ skepticism of the literature on unconscious processes, as few if any articles have used awareness measures satisfying all of these criteria. We recommend researchers interested in studying awareness to consult the original publications for more in-depth discussion of the different criteria. Here we will exemplify these criteria by considering the use of funneled debriefing protocols, one of the most frequently used methods to assess awareness of a priming 14 procedure (e.g, Chartrand et al 2008; Dalton and Huang 2014; Fitzsimons, Chartrand, and Fitzsimons

2008; Laran, Janiszewski, and Salerno 2016; Sweldens, van Osselaer, and Janiszewski 2010; Tuk et al. 2009; Wheeler and Berger 2007) Typically, a series of questions of increasing specificity is presented, ranging, for example, from “please guess the real purpose of the study,” over “did you see a connection between the first and second part of this session,” to “did you see a connection between the words in the first task? If so, which one?” Despite their intuitive appeal, funneled debriefing procedures often fail on multiple criteria for awareness measures. The first issue to consider is that of reliability, which concerns the reproducibility and independence of noise in the measure. Open-ended questions typically score low on reliability as participants differ widely in their eloquence and motivation to answer truthfully and thoughtfully. Reliability (how the question is answered) ties immediately into relevance, or what information is being probed by the question. The

relevance criterion dictates that awareness measures should test participants’ knowledge of precisely the information that an aware participant would rely on when responding to the key behavioral (or attitudinal) measure of the study. Many of the questions in a funneled debriefing procedure are often not well targeted at the most relevant dimension(s). Note that to achieve “relevance” in the question format, the researcher should first consider the operational definition (R3) and the process level at which s/he aims to demonstrate unawareness (R1, R2). To demonstrate, for example, that a goal priming effect occurs without awareness of the primed construct, questions should be targeted on that construct. Say the priming manipulation consists of a lexical decision task featuring selfcontrol related words (eg, Laran et al, 2016, Study 1) The awareness check should then assess whether participants were aware that the task contained self-control related words. Instead, a 15

funneled debriefing procedure is usually restricted to vague questions like “did you notice the words were related?” or “did your responses in one task influence those in another task?” The third issue is that of immediacy, which in Newell and Shanks’ (2014) description prescribes that the awareness check should happen as closely as possible to the target behavior. We believe this should be qualified: again researchers need to consider R1 to R3 before determining what would be the appropriate level of ‘immediacy.’ In a goal priming setting, for example, if one aims to demonstrate that the effects are caused without awareness of the primed construct, the awareness measure should follow the primed construct immediately. If, on the other hand, one aims to demonstrate that the behavior occurs without awareness of the thought process, the measure can occur closer to the behavior. The problem with funneled debriefing procedures is that they are typically collected at the very

end of an experiment. This is problematic because the human brain has an unparalleled ability to forget: according to some estimates, 90% of the information that is not transmitted to long-term memory disappears from short-term memory within 20 seconds (Rubin and Wenzel 1996). Hence, measures collected at the end of an experiment are at risk of severely underestimating actual levels of awareness and could thus easily provide spurious evidence for unconscious effects. The final issue is that of sensitivity, which specifies that awareness measures should be at least of equal sensitivity as the behavioral measures they speak to. This too can be a problem with funneled debriefing procedures which normally consist of roughly coded open-ended questions, while behavior is often measured to the millisecond (e.g, in an evaluative priming measure) or consists of forced choice measures (e.g, when participants choose between a healthy and unhealthy option). Now imagine that research participants

would like to minimize the effort they spend on the experiment (not too far-fetched an assumption). Every participant of this kind 16 would provide completely registered data on the key dependent measures (e.g, evaluative priming, forced choice, etc.), but would be free to skip through the funneled debriefing part as fast as s/he could. Disturbingly, the less effort participants decide to invest on the debriefing procedure, the greater their chances of being classified as ‘unaware.’ The problems can be further aggravated when researchers do not properly account for the consequences of measurement error in the way they combine measures of awareness and performance on a different criterion (e.g, attitudes, purchase intentions, reaction times) to draw inferences about unconscious processes. For example, when participants are selected based on an extreme score on one measure (i.e, selecting those scoring very low on an awareness measure), it is a statistical regularity that they

will score closer to the average on a different measure (i.e, a performance measure). Such a “regression to the mean” bias is sufficient to generate spurious evidence for unconscious processes (Shanks 2016). Considering the various ways in which measures of awareness have fallen short of their target, it is understandable why some scholars doubt whether evidence presented for unconscious processes stands up to closer scrutiny. Hence: R4: Make sure the awareness measure satisfies the criteria of reliability, sensitivity, relevance, and immediacy to the best of your ability. Beware of spurious inferences due to measurement error. Operationalization via Experimental Manipulation The manifold problems associated with measurement of awareness have prompted calls to rely more on operationalization of awareness via experimental manipulation (Gawronski and Walther 2012; Shanks 2016). Popular approaches to ensure information has been presented ‘without awareness’ include presenting

critical information in a hidden format, for example by mixing target words in between filler items in scrambled sentence tasks or word search puzzles 17 (e.g, Laran et al 2016; Tuk et al 2009) or by presenting information subliminally (eg, Chartrand et al. 2008; Dedonder et al, 2014; Fitzsimons et al 2008; Galli and Gorn 2011; Stahl et al. 2016) Experimental approaches can have important advantages over pure measurement approaches. Notably, random allocation of participants to ‘unaware’ versus ‘aware’ conditions prevents regression to the mean effects on other variables, which occur when this allocation is based on measurement (Shanks 2016). Nevertheless, manipulations of awareness still need to be accompanied by sensitive measures to offer convincing evidence for unconscious processes. Otherwise, it could not be excluded that (a subset of) participants can somehow still detect the presented information and drive a spurious ‘unconscious’ effect. For example, one

problem with subliminal presentations is the inter-individual variation in detection thresholds. Hence, subliminal presentations should always be accompanied by sensitive measures of detection thresholds, so potentially aware participants can be excluded or the presentation times can be individually adjusted (Holender 1986). Such combined approaches are still rare in consumer research; we would like to highlight Galli and Gorn (2011) as a commendable example in this regard. Note that one additional advantage of combining experimental manipulation with measurement of awareness is that the manipulation provides a direct test of the sensitivity of the measurement. Hence: R5: The strongest approaches combine experimental manipulation with sensitive measurement of awareness. While a combined approach can in principle offer strong evidence for unconscious processes, it should be noted that failures to find evidence (i.e, null effects) for unconscious processes via subliminal presentations

need to be interpreted with caution. We agree with WP 18 and Evans (2014) that this might be a case where unfair standards of evidence are imposed on demonstrations of unconscious thought processes. Bargh and Morsella (2008, p 74) expressed this concern most clearly: “We [] oppose the cognitive psychology equation of the unconscious with subliminal information processing []. Subliminal stimuli do not occur naturallythey are by definition too weak or brief to enter conscious awareness. Thus, it is unfair to measure the capability of the unconscious in terms of how well it processes subliminal stimuli because unconscious (like conscious) processes evolved to deal and respond to naturally occurring (regular strength) stimuli; assessing the unconscious in terms of processing subliminal stimuli is analogous to evaluating the intelligence of a fish based on its behavior out of water. And as one might expect, the operational definition of the unconscious in terms of subliminal

information processing has in fact led to the conclusion of the field that the unconscious is, well, rather dumb.” THE WAY AHEAD Perfection is not attainable, but if we chase perfection, we can catch excellence - Vince Lombardi Perfect Measures or Manipulations of Awareness Do Not Exist We hope our recommendations can be helpful to researchers studying consciousness. At the same time, it should be noted that none of these recommendations is easily satisfied, let alone all of them together. At the risk of discouraging researchers, we note that the criteria by Shanks and colleagues are not even the only ones to be satisfied by awareness measures. A particularly difficult issue is the extent to which the measures exhaustively and exclusively measure awareness, in the sense that the measure should reflect all possible sources of awareness and not be influenced by unaware processes (Reingold and Merikle 1988). Since Jacoby’s (1991) 19 seminal paper on process dissociation, it has

been recognized that these criteria are never satisfied, as no measure is ever process-pure. We are grateful for WP’s discussion of our work on process dissociation procedures as an important step forward (WP, p. 13) Twenty-five years after their introduction, process dissociation procedures are still at the vanguard of research on consciousness. One reason why they are often superior to other approaches is that by design they combine manipulation and measurement into one procedure. Yet, process dissociation approaches come with their own assumptions and limitations, most notably the strong assumption that conscious and unconscious processes contribute equally in different conditions (Hütter and Klauer 2016). Therefore parameter estimates obtained by process dissociation procedures cannot be taken at face value either, but (like other measures) need to be complemented with additional experimental manipulations. For example, parameter estimates for conscious and unconscious processes

need to be validated by manipulations of variables assumed to impact these processes differently (e.g, cognitive capacity, motivation or attention; Hütter et al. 2012; Hütter and Sweldens, 2016; Mierop, Hütter, and Corneille, 2016). Consciousness as a Continuum The previous sections highlight that it is nearly impossible to design a perfect measure of awareness. How can consumer researchers deal with this important constraint? Shall we stop investigating the role of consciousness? We would like to promote a more optimistic view and find inspiration in the perspective on consciousness as a continuum, highlighted by Plassmann and Mormann (this issue). We fully agree with these authors that consciousness need not be an all-or-none phenomenon. Furthermore, we believe that many problems and ensuing criticism in this field emerged from the fact that researchers often applied a dichotomous perspective on 20 consciousness to their theories and empirical evidence, while the awareness

measures they used are ill-suited to substantiate dichotomous claims (especially where it concerns the strictly ‘unconscious’ nature of an effect). However, the imperfections in awareness measures are much less problematic if consciousness is treated as a continuous construct and researchers refrain from hard claims regarding the unconscious nature of an effect. Instead, in most cases the measures would be able to validly support claims that ‘effects are less consciously mediated’ or ‘characterized by lower levels of awareness’ in some conditions compared with others. R6: Hard claims that a process is unconscious are difficult to support, given the limitations of awareness measures. Softer claims that processes are ‘more’ or ‘less’ consciously mediated can be more validly entertained. CONCLUSION Despite the difficulties involved in the study of consciousness, we cannot agree more with WP that the role of consciousness has crucial theoretical and practical

implications for consumer behavior. The last few decades have brought about important conceptual and methodological advancements which we should embrace and continue to develop (e.g, process dissociation procedures, item-based analyses and hierarchical models, convenient eye-tracking equipment, the neuroscience toolbox). At the same time, it is crucial to be aware of the difficulties in studying consciousness. Researchers should try their best to apply or develop the best possible measures, yet at the same time be aware of and explicit about the limitations of the measure they use. They should not make claims that transcend the empirical evidence, acknowledging the limitations they encountered from R1-R4. Conversely, reviewers and editors should value advancements in this challenging field which acknowledge, but not necessarily overcome, these limitations in a single paper. 21 In closing, we would like to note that several of our comments are not restricted to the investigation of

awareness. Awareness is just one of several features of automaticity that may or may not co-occur. As rightly pointed out by WP, even if a process occurs without awareness, it cannot automatically be assumed that this process would also be uncontrollable or independent of processing resources. These features need to be investigated separately The same conceptual and methodological rigor that we promoted in this commentary needs to be applied to other features of automaticity to gain a more complete understanding of consumer behavior. 22 REFERENCES Allen, Chris T. and Chris A Janiszewski (1989), "Assessing the Role of Contingency Awareness in Attitudinal Conditioning with Implications for Advertising Research," Journal of Marketing Research, 26 (1), 30-43. Allen, Chris T. and Thomas J Madden (1985), "A Closer Look at Classical Conditioning," Journal of Consumer Research, 12 (3), 301-15. Bargh, John A. and Ezequiel Morsella (2008), "The Unconscious

Mind," Perspectives on Psychological Science, 3 (1), 73-79. Baumeister, Roy F., Cory J Clark, Jonghan Kim, and Stephan Lau (forthcoming), "Consumers (and Consumer Researchers) Need Conscious Thinking in Addition to Unconscious Processes: A Call for Integrative Models," Journal of Consumer Research. Brewer, William P. (1974), "There is No Convincing Evidence for Operant or Classical Conditioning in Adult Humans," in Cognition and the Symbolic Processes, ed. Walter B Weimer and David S. Palermo, Hillsdale, NJ: Lawrence Erlbaum, 1-42 Brunswik, Egon (1952), The Conceptual Framework of Psychology, The University of Chicago Press. Carmon, Ziv, Klaus Wertenbroch, and Marcel Zeelenberg (2003), "Option Attachment: When Deliberating Makes Choosing Feel Like Losing," Journal of Consumer Research, 30 (1), 15-29. Chartrand, Tanya L. (2005), "The Role of Conscious Awareness in Consumer Behavior," Journal of Consumer Psychology, 15 (3), 203-10. Chartrand,

Tanya L., Joel Huber, Baba Shiv, and Robin J Tanner (2008), "Nonconscious Goals and Consumer Choice," Journal of Consumer Research, 35 (2), 189-201. Dalton, Amy N. and Li Huang (2014), "Motivated Forgetting in Response to Social Identity Threat," Journal of Consumer Research, 40 (6), 1017-38. Dedonder, Jonathan, Olivier Corneille, Denis Bertinchamps, and Vincent Yzerbyt (2014), "Overcoming Correlational Pitfalls: Experimental Evidence Suggests That Evaluative Conditioning Occurs for Explicit but Not Implicit Encoding of Cs–Us Pairings," Social Psychological and Personality Science, 5 (2), 250-57. Dijksterhuis, Ap (2004), "Think Different: The Merits of Unconscious Thought in Preference Development and Decision Making," Journal of Personality and Social Psychology, 87 (5), 586-98. Dijksterhuis, Ap, Ad van Knippenberg, Rob W. Holland, and Harm Veling (2014), "Newell and Shanks Approach to Psychology is a Dead End," Behavioral and Brain

Sciences, 37 (1), 25-26. Epstein, Seymour, Rosemary Pacini, Veronika DenesRaj, and Harriet Heier (1996), "Individual Differences in Intuitive-Experiential and Analytical-Rational Thinking Styles," Journal of Personality and Social Psychology, 71 (2), 390-405. Evans, Jonathan S. T (2014), "The Presumption of Consciousness," Behavioral and Brain Sciences, 37 (1), 26-27. Fiedler, Klaus and Mandy Hütter (2014), "The Limits of Automaticity," in Dual Processes Theories of the Social Mind, ed. Jeffrey W Sherman, Bertram Gawronski and Yaacov Trope, New York, NY: Guilford Press, 497-513. 23 Fitzsimons, Grainne M., Tanya L Chartrand, and Gavan J Fitzsimons (2008), "Automatic Effects of Brand Exposure on Motivated Behavior: How Apple Makes You ‘Think Different’," Journal of Consumer Research, 35 (1), 21-35. Forehand, Mark R. and Andrew Perkins (2005), "Implicit Assimilation and Explicit Contrast: A Set/Reset Model of Response to Celebrity

Voice-Overs," Journal of Consumer Research, 32 (3), 435-41. Galli, Maria and Gerald J. Gorn (2011), "Unconscious Transfer of Meaning to Brands," Journal of Consumer Psychology, 21 (3), 215-25. Gawronski, Bertram and Galen V. Bodenhausen (2014), "The Associative - Propositional Evaluation Model: Operating Principles and Operating Conditions of Evaluation," in Duel-Process Theories of the Social Mind, ed. Jeffrey W Sherman, Bertram Gawronski and Yaacov Trope, New York, NY: Guilford Press, 188-203. Gawronski, Bertram and Eva Walther (2012), "What Do Memory Data Tell Us About the Role of Contingency Awareness in Evaluative Conditioning?," Journal of Experimental Social Psychology, 48 (3), 617-23. Genevsky, Alex and Brian Knutson (2015), "Neural Affective Mechanisms Predict Market-Level Microlending," Psychological Science, 26 (9), 1411-22. Holender, Daniel (1986), "Semantic Activation without Conscious Identification in Dichotic Listening,

Parafoveal Vision, and Visual Masking: A Survey and Appraisal," Behavioral and Brain Sciences, 9 (1), 1-23. Hütter, Mandy and Karl Christoph Klauer (2016), "Applying Processing Trees in Social Psychology," European Review of Social Psychology, 27 (1), 116-59. Hütter, Mandy and Steven Sweldens (2016), "Threatening Consumer Autonomy? Uncontrollable Effects of Affective Stimuli on Attitudes and Consumption.," Working Paper Hütter, Mandy, Steven Sweldens, Christoph Stahl, Christian Unkelbach, and Karl Christoph Klauer (2012), "Dissociating Contingency Awareness and Conditioned Attitudes: Evidence of Contingency-Unaware Evaluative Conditioning," Journal of Experimental Psychology: General, 141 (3), 539-57. Jacoby, Larry L. (1991), "A Process Dissociation Framework - Separating Automatic from Intentional Uses of Memory," Journal of Memory and Language, 30 (5), 513-41. Kahneman, Daniel and Shane Frederick (2002), "Representativeness

Revisited: Attribute Substitution in Intuitive Judgment," in Heuristics and Biases: The Psychology of Intuitive Judgment, ed. Thomas Gilovich, Dale Griffin and Daniel Kahneman, New York, NY: Cambridge University Press, 49-81. Kruglanski, Arie W. and Gerd Gigerenzer (2011), "Intuitive and Deliberate Judgments Are Based on Common Principles," Psychological Review, 118 (1), 97-109. Labroo, Aparna A. and Vanessa M Patrick (2009), "Psychological Distancing: Why Happiness Helps You See the Big Picture," Journal of Consumer Research, 35 (5), 800-09. Laran, Juliano, Chris Janiszewski, and Anthony Salerno (2016), "Exploring the Differences Between Conscious and Unconscious Goal Pursuit," Journal of Marketing Research, 53 (3), 442-58. Lovibond, Peter F. and David R Shanks (2002), "The Role of Awareness in Pavlovian Conditioning: Empirical Evidence and Theoretical Implications," Journal of Experimental Psychology: Animal Behavior Processes, 28 (1),

3-26. Mierop, Adrien, Mandy Hütter, and Olivier Corneille (forthcoming). “Resource Availability and Explicit Memory Largely Determine Evaluative Conditioning Effects in a Paradigm 24 Claimed to be Conducive to Implicit Attitude Acquisition,” Social Psychological and Personality Science. Mitchell, Chris J., Jan De Houwer, and Peter F Lovibond (2009), "The Propositional Nature of Human Associative Learning," Behavioral and Brain Sciences, 32 (2), 183-98. Moors, Agnes and Jan De Houwer (2006), "Automaticity: A Theoretical and Conceptual Analysis," Psychological Bulletin, 132 (2), 297-326. Newell, Ben R. and David R Shanks (2014), "Unconscious Influences on Decision Making: A Critical Review," Behavioral and Brain Sciences, 37 (1), 1-19. Nisbett, Richard E. and Timothy D Wilson (1977), "Telling More Than We Can Know: Verbal Reports on Mental Processes," Psychological Review, 84 (3), 231-59. Page, Monte M. (1974), "Demand

Characteristics and the Classical Conditioning of Attitudes Experiment," Journal of Personality and Social Psychology, 30 (4), 468-74. Persaud, Navindra, Peter McLeod, and Alan Cowey (2007), "Post-Decision Wagering Objectively Measures Awareness," Nature Neuroscience, 10 (2), 257-61. Plassmann, Hilke and Milica Mormann (forthcoming), "An Interdisciplinary Lens on Consciousness: The Consciousness Continuum and How to (Not) Study It in the Brain and the Gut," Journal of Consumer Research. Reingold, Eyal M. and Philip M Merikle (1988), "Using Direct and Indirect Measures to Study Perception Without Awareness," Perception & Psychophysics, 44 (6), 563-75. Rubin, David C. and Amy E Wenzel (1996), "One Hundred Years of Forgetting: A Quantitative Description of Retention," Psychological Review, 103 (4), 734-60. Shanks, David R. (2010), "Learning: From Association to Action," Annual Review of Psychology, 61, 273-301. Shanks, David R.

(forthcoming), "Regressive Research: The Pitfalls of Post Hoc Data Selection in the Study of Unconscious Mental Processes," Psychonomic Bulletin & Review. Shanks, David R. and Mark F St John (1994), "Characteristics of Dissociable Human Learning Systems," Behavioral and Brain Sciences, 17 (3), 367-95. Simonson, Itamar (1989), "Choice Based on Reason: the Case of Attraction and Compromise Effects," Journal of Consumer Research, 16 (2), 158-74. Stahl, Christoph, Julia Haaf, and Olivier Corneille (2016), "Subliminal Evaluative Conditioning? Above-Chance CS Identification May Be Necessary and Insufficient for Attitude Learning," Journal of Experimental Psychology: General, 145 (9), 1107-31. Stenstrom, Eric, Gad Saad, Marcelo V. Nepomuceno, and Zack Mendenhall (2011), "Testosterone and Domain-Specific Risk: Digit Ratios (2D:4D and Rel2) as Predictors of Recreational, Financial, and Social Risk-Taking Behaviors," Personality and

Individual Differences, 51 (4), 412-16. Stuart, Elnora W., Terence A Shimp, and Randall W Engle (1987), "Classical Conditioning of Consumer Attitudes: Four Experiments in an Advertising Context," Journal of Consumer Research, 14 (3), 334-49. Sweldens, Steven, Olivier Corneille, and Vincent Yzerbyt (2014), "The Role of Awareness in Attitude Formation through Evaluative Conditioning," Personality and Social Psychology Review, 18 (2), 187-209. Sweldens, Steven, Stijn M. J van Osselaer, and Chris Janiszewski (2010), "Evaluative Conditioning Procedures and the Resilience of Conditioned Brand Attitudes," Journal of Consumer Research, 37 (3), 473-89. 25 Timmermans, Bert and Axel Cleeremans (2015), "How Can We Measure Awareness? An Overview of Current Methods," in Behavioral Methods in Consciousness Research, ed. Morten Overgaard, Oxford, UK: Oxford University Press, 21-46. Tuk, Mirjam A., Peeter WJ Verlegh, Ale Smidts, and Daniel HJ Wigboldus

(2009), "Sales and Sincerity: The Role of Relational Framing in Word-of-Mouth Marketing," Journal of Consumer Psychology, 19, 38-47. Van den Bergh, Bram, Siegfried Dewitte, and Luk Warlop (2008), "Bikinis Instigate Generalized Impatience in Intertemporal Choice," Journal of Consumer Research, 35 (1), 85-97. Waroquier, Laurent, David Marchiori, Olivier Klein, and Axel Cleeremans (2009), "Methodological Pitfalls of the Unconscious Thought Paradigm," Judgment and Decision Making, 4 (7), 601-10. Wheeler, S. Christian and Jonah Berger (2007), "When the Same Prime Leads to Different Effects," Journal of Consumer Research, 34 (3), 357-68. Williams, Lawrence E. and T Andrew Poehlman (forthcoming), "Conceptualizing Consciousness in Consumer Research," Journal of Consumer Research