Why Aren't We Saving the Planet?: A Psychologist's Perspective - Geoffrey Beattie (2010)
Part III. Notes on dissociation
Chapter 10. In two minds
This book was always going to be a journey, a journey in uncharted territory, maybe even a little stop-start at times, but I felt that some slow progress was being made. I now knew certain things. I knew that attitudes were generally positive towards low-carbon-footprint products and that people were, inside, relatively green (in some domains at least), but I also knew that people didn’t really pick up on carbon footprint information in the time it takes to make a supermarket decision. The idea behind carbon labelling was (in principle) a good one, an empowering one. It allowed ordinary people to act in accordance with their underlying beliefs, but the actual mechanics of carbon labelling - how the information was represented, what kind of icons were used, what kind of numerical information was included, how the label looked - all needed a little bit more thought. Even those with the right underlying attitudes, the majority of the respondents, weren’t attending to the information in that 5-second or 10-second envelope that is critical to the purchasing decision in the supermarket. And perhaps people didn’t really care quite enough to find the carbon label icon in the necessary time frame. Their attention was not automatically drawn to it.
The research so far had also thrown up something else that was quite interesting. It allowed me to identify a certain group of people, between 10 and 20 per cent in the original sample, in which the two types of attitude (explicit and implicit) did not match, and it is to these individuals that I now turn. When I started thinking about this group a few months ago when I first saw the results, I called them ‘the green fakers’. However, I grew to dislike this term; it is unkind and unfair, unfair to them and unfair to me because when I asked Laura to run me through the experiment I discovered that I was one of them. In order to understand these people (and myself) I needed to understand a bit more about how these two attitudes are actually represented in the human brain, and in order to do this, I needed to consider in much more detail the limitations of the research that I had done so far on underlying attitudes. I suspected that this held the key to a lot of the important issues.
The research so far on attitudes was, of course, exploratory and by necessity purely descriptive. It didn’t really consider how the explicit and implicit attitudes were related within the individual. Were they significantly correlated or were they discrepant? Could there, in fact, be some level of ‘dissociation’ between them? (‘Dissociation’ is one of those psychological terms that seems so complex but at its heart is very simple - it just means ‘being separate’ or ‘lack of connection’, although I will discuss in a moment how psychologists use this term to refer to different states.) And if there are marked discrepancies within individuals, what are the psychological consequences for ordinary people (and me) of attempting to understand and explain their everyday actions concerning green choices in supermarket shopping? Furthermore, if implicit attitudes are largely unconscious, unlike explicit attitudes, can we rethink the role of unconscious impulses in everyday behaviour? Could we, in fact, look for evidence of the unconscious breaking through, perhaps as Freud (1901/1975) had done over a century earlier with respect to slips of the tongue? Any attempt to answer these questions will require a much fuller description of the relationship between implicit and explicit attitudes and a consideration of the new research into how thinking and speaking are represented and generated by the human brain. Interestingly, this will involve not just the analysis of speech itself (as Freud had done), but the analysis of human communicative behaviour more generally, because current research in psychology suggests that speech and the accompanying hand movements must be considered together to understand more fully speakers’ underlying thought processes.
But first the relationship between implicit and explicit attitudes. There is now a major controversy within psychology about the relationship between these two constructs. Many psychologists maintain that the representations underlying explicit and implicit attitudes are, in fact, dissociated. According to Greenwald and Nosek (2008:65):
This reference to dissociation implies the existence of distinct structural representations underlying distinguishable classes of attitude manifestations. In psychology, appeals to dissociation range from the mundane to the exotic. At the mundane end, the dissociation label may be attached to the simple absence or weakness of correlation between presumably related measures. At the exotic end, dissociation may be understood as a split in consciousness, such as mutually unaware person systems occupying the same brain. (emphasis in original)
So how exotic is this dissociation in attitudes likely to be? There are a number of critical lines of evidence here. Nosek and Hansen (2008) report that in a meta-analysis of 81 studies the IAT was only moderately correlated with self-reported attitudes (r = 0.24; Hofmann, Gawronski, Gschwendner, Le and Schmitt 2005) and, in a study of fifty attitude domains, Nosek (2005) found that the strength of the correlation between the IAT and self-reported attitudes varied from near zero for some attitude domains (e.g. attitudes to thin and fat) to approximately 0.70 in other domains (e.g. pro-choice/pro-life attitudes). So in Greenwald and Nosek’s (2008) terms there is clearly at least a mundane dissociation between the two constructs. But, in addition, there is the finding that other variables (for example, chronological age) can have a well-defined relationship with one of the measures (say, explicit attitude) but no relation with the other (say, implicit attitude). This has been found for things like explicit and implicit age preference. Here, there is a significant correlation between the age of the participant and age preference in the case of the explicit measure but no significant correlation in the case of the implicit measure.
Of course, these types of finding of possible attitudinal dissociation do not point to a single interpretation of the data, and Greenwald and Nosek (2008) have argued that they are, in fact, compatible with three different interpretations. First, the ‘single-representation hypothesis’, which maintains that the appearance of a dissociation is really just an illusion and all that is happening is that in explicit self-report measures, participants have the opportunity to modify their real response in their explicit reporting of their attitude. The second ‘dual-representation hypothesis’ is the real dissociation claim, and maintains that implicit and explicit measures of attitude have structurally distinct mental representations of attitudes and they are genuinely dissociated (see Chaiken and Trope 1999; Wilson, Lindsey and Schooler 2000). This hypothesis seems to be favoured by many in this area. Thus:
Abundant theory, and some evidence, point to the human mind being divided into two largely independent subsystems: first, a familiar foreground, where processing is conscious, controlled, intentional, reflective and slow, but where learning occurs rapidly; and second, a hidden background, where processing is unconscious, automatic, unintended, impulsive and fast, but where learning occurs gradually … This dual-process model is appealingly neat. It recalls Freud’s model of the mind, but with the inner sex maniac replaced by a dull but efficient zombie. (Gregg 2008:764)
In this hypothesis the implicit attitudes operate automatically and unconsciously (and have unconscious representations) while the explicit attitudes operate consciously and with deliberate thought (and have an underlying representation which is quite different).
The third ‘person vs culture hypothesis’ is that explicit measures capture the attitudes operating within a person while implicit attitudes represent the more general influence of what is known about a particular thing in a particular culture. Nosek and Hansen (2008) argue, however, that the evidence from variations in the level of correlation in the IAT and self-reported attitude across individuals largely rules out this third hypothesis, and that the IAT does not merely reflect the evaluative judgement of the culture as a whole. Their conclusion is that the data demonstrate that the IAT is an individual difference measure and is associated with individual-level thoughts, feelings, and actions.
But that still leaves two major hypotheses: one (the single-representation hypothesis) is that there is one underlying representation for both implicit and explicit attitudes; the second is that implicit attitudes have their roots in the unconscious (dual-representation hypothesis) and explicit attitudes have their roots elsewhere. The first hypothesis might suggest that when we report our attitude we may be aware of what we really feel but we modify our verbal response particularly in sensitive domains (like race, attitudes to obesity and attitudes to green issues which we might like to exaggerate to make ourselves look good). The second hypothesis suggests that the implicit attitude is grounded in the unconscious; we are unaware of this attitude and may well be puzzled by the slowness of our reaction times and our high error rate when we sit in front of the computer screen as we complete the IAT.
This is clearly a major issue for social psychology and for those wishing to promote behavioural change in many core areas. For example, when we find using self-report measures that people seem to be pro-low carbon in explicit attitude (but not in implicit attitude), how should we interpret this result: as merely a social desirability effect or as something more profound? This could be extremely important from the point of view of behavioural change. One reason for saying this is that there are clear practical implications when the two attitudes align. As Greenwald and Nosek (2008) pointed out, ‘When self-report and IAT measures were highly correlated with each other - a circumstance occurring especially in domains of political and consumer attitudes - both types of measures were more strongly correlated with behaviour than when implicit- explicit correlations were low’ (2008:78). In other words, once we know something about the nature of the correlation between these two measures, we will have a much better understanding of the likelihood of predicting behaviour from either or both of the measures. This is of major significance when it comes to climate change and how to tackle it.
So, I returned to the original data to take a fresh look at the scores and I carried out a batch of statistical tests, and in particular a series of correlations to work out the relationship between the various measures. A Pearson product- moment correlation coefficient revealed that there was no significant correlation between any of the measures of explicit attitude and the IAT measure (for the Likert scores and D scores r = 0.008, for the thermometer difference scores and D scores r =-0.057). A Pearson product-moment correlation coefficient, however, showed that there was a strong positive relationship between the various measures of explicit attitude (namely, the Likert and thermometer difference scores, r = 0.560) as shown in Figure 10.1.
Further comparisons using a Spearman’s rank-order correlation coefficient suggested that while there was no relationship between age and D scores ( = 0.56) as shown in Figure 10.2, there was a positive relationship between age and Likert scores ( = 0.214) as shown in Figure 10.3
Figure 10.1 Likert and thermometer difference correlations.
Figure 10.2 Age and D score correlations.
Figure 10.3 Age and Likert correlations.
and between age and thermometer difference scores ( = 0.197) as shown in Figure 10.4.
In other words, the measures of implicit and explicit attitudes could well be dissociated because firstly there is no significant correlation between them, and secondly one additional variable (chronological age) correlates with one
Figure 10.4 Age and thermometer difference correlations.
(the explicit measure) but not the other (the implicit measure). Of course, this latter finding is interesting because it suggests that as people get older they become more concerned about explicitly reporting their green credentials.
The fact that there was no significant correlation between the implicit and explicit measures also allows us to identify different sets of individuals with differing patterns of implicit and explicit attitudes. Figure 10.5 compares the Likert and D score results. While the majority of participants showed some degree of convergence between implicit and explicit attitudes with D scores > 0.8 and Likert scores of 4 and 5 (outlined by a dashed line in Figure 10.5), there were 13 participants in this first sample of 100 who, despite saying explicitly that they were very pro-low carbon with Likert scores of 5, had implicit attitudes that were not as positive, with D scores < 0.8 (outlined by a bold dashed line in Figure 10.5).
Figure 10.6 compares the thermometer difference and D score results. Again, the majority of participants displayed convergent attitudes, expressing a preference for low-carbon-footprint products at both the implicit and explicit levels (outlined by a dashed line on the right-hand side of
Figure 10.5 Likert and D score comparisons.
Figure 10.6 Thermometer difference and D score comparisons.
Figure 10.6). However, there were two sets of participants that showed attitudinal divergence at both extremes. Overall, twelve participants (a similar number to those who displayed divergent attitudes using the Likert scale) explicitly stated that they were pro-low carbon with thermometer difference scores of 4, but their implicit attitudes were much less positive, with D scores < 0.8 (outlined by a bold dashed line on the right-hand side of Figure 10.6). Interestingly, a second set consisting of five participants (outlined in bold to the left of Figure 10.6) expressed explicit attitudes that demonstrated a preference for high carbon (using the criterion of < 0 for thermometer difference scores); however, implicitly their attitudes appeared to be much more positive towards low carbon (with D scores > 0.8).
To recap, although the explicit measures of attitude (the feeling thermometer and Likert scale) were significantly correlated, the explicit and implicit measures were not correlated. This discrepancy seems to reflect some kind of ‘dissociation’, and has been reported previously in a number of other domains (see Banaji and Hardin 1996; Blair and Banaji 1996; Bosson, Swann and Pennebaker 2000; Devine 1989; Fazio, Jackson, Dunton and Williams 1995; Greenwald, McGhee and Schwartz 1998), although in other domains measures of implicit and explicit attitudes appear to be positively related (see Greenwald and Nosek 2001; Hofmann et al. 2005; Nosek and Banaji 2002; Nosek, Banaji and Greenwald 2002; Nosek and Smyth 2005).
Nosek (2007) has argued that ‘measurement innovations [such as the IAT] have spawned dual-process theories that, among other things, distinguish between the mind as we experience it (explicit), and the mind as it operates automatically, unintentionally, or unconsciously (implicit)’ (2007:184). So we have here the distinct possibility of two largely independent subsystems in the human mind, one that is familiar and one that is not. (Whether we have any ‘conscious’ awareness at all of our implicit thinking, and whether the implicit process is always truly unconscious or whether we have some inkling of the underlying evaluation, remain to be properly investigated. The fact that something cannot be consciously controlled and manipulated does not of course mean that it resides purely and totally in the unconscious.) But how does this divergence between implicit and explicit attitude manifest itself within the individual, and does it have any effect on any aspects of observable behaviour? After all, a hundred years ago or so Freud showed how unconscious (and repressed) thoughts could find articulation through the medium of everyday speech in the form of slips of the tongue. And how might this dissociation impact on people’s willingness or ability to actually do something about climate change? These are potentially important questions from both a theoretical and a practical point of view. It surprised me that nobody until now had attempted to answer them.