The missing ingredient is now available - Notes on attitude - Why Aren't We Saving the Planet?: A Psychologist's Perspective - Geoffrey Beattie

Why Aren't We Saving the Planet?: A Psychologist's Perspective - Geoffrey Beattie (2010)

Part I. Notes on attitude

Chapter 5. The missing ingredient is now available

The missing ingredient was introduced in 1998 in an article published in the Journal of Personality and Social Psychology and coauthored with Debbie McGhee and Jordan Schwartz. Greenwald introduced the new method with a thought experiment. Imagine being shown a series of male and female faces and having to respond rapidly by saying ‘hello’ to the male face and ‘goodbye’ to the female face. Then you are shown a series of names and this time you say ‘hello’ to the male name and ‘goodbye’ to the female name.

In that experiment subjects gave a response on a computer keyboard with the index finger of the right hand to words that named pleasant things and to names of flowers. With the left hand they were to respond to another two categories - words that named unpleasant things and insect names. This was a very easy task. Then we made one minor change: We switched hands for the flower and insect names. Now subjects had to give the same response to pleasant words and insect names and a different response to unpleasant words and flower names. Immediately the task became hugely difficult. The slowing on a response-by-response basis was on the order of 300 milliseconds, which was a magnitude of impact nobody could have expected. We certainly did not expect it.

I was the first subject in the experiment. When I experienced the slowing I found to my surprise that I could not overcome it - repeating the task did not make me faster. If I tried to go faster, I just started making errors when I was trying to give the same response to flower names and unpleasant words. This was a mind-opener.

The very first paper reporting the Implicit Association Test (IAT) provided psychologists with a much sought-after method to measure unconscious, implicit attitudes; but perhaps even more than that, it uncovered something that was extremely unsettling for Greenwald and colleagues, and no doubt for anyone who read the paper. In today’s society we like to think that race is no longer a significant issue: I am writing this particular paragraph in the same week that America has just elected its first Black president; surely the times of racial prejudice and stereotype are far behind us all in the West. The IAT revealed that this is not necessarily the case. The basic premise behind the IAT is that when categorising items into two sets of paired concepts, if the paired concepts are strongly associated, then participants should be able to categorise items faster into these category concepts. The IAT revealed that people were consistently faster at categorising Black and White names and pleasant and unpleasant words when the target categories were grouped ‘White’/‘pleasant’ and ‘Black’/‘unpleasant’ than when they were grouped ‘White’/‘unpleasant’ and ‘Black’/ ‘pleasant’, suggesting that the former concepts are strongly associated. When compared to explicit measures, the majority of White college students who took part in the study reported that they had no racial preference between White and Black, with some even saying they had a preference for Black. However, the IAT revealed that only one of these students showed a preference for Black consistent with their stated explicit attitudes. The remaining participants all showed a White preference, suggesting that White had positive associations, whereas Black had negative associations. As such, the IAT was able to successfully reveal underlying implicit attitudes that firstly cannot be masked by social desirability concerns and secondly a person may be totally unaware of holding.

In the first web-based experiment of its kind, Project Implicit measured implicit attitudes towards a range of social groups, including implicit measures of racial attitudes. The project collated a staggering 600,000 tests between October 1998 and December 2000, allowing for replication of the race IAT on an enormous scale using both White and Black participants, with surprising results. It found that White participants tended to explicitly endorse a preference for White but implicitly they demonstrated an even stronger preference for White names and faces. Black participants, on the other hand, demonstrated a strong explicitpreference for Black yet remarkably in the IAT, Black participants demonstrated a weak implicit preference for White names and faces. According to Nosek and colleagues who provided this overview of the IAT results in 2002, the preference shown for White by both White and Black participants is indicative of the American culture in which Black Americans are still often depicted in a negative light. The result is that these negative associations have penetrated into underlying racial attitudes and stereotypes, leading to the creation of automatic evaluations which show an implicit preference for White over Black people. For people who show strong explicit endorsement of racial indifference, the prospect that they may implicitly hold the very attitudes they strongly condemn can be a worrying thought (see also Gladwell 2005). As Fyodor Dostoevsky (1864) wrote in his Notes from the Underground:

Every man has some reminiscences which he would not tell to everyone, but only to his friends. He has others which he would not reveal even to his friends, but only to himself, and that in secret. But finally there are still others which a man is even afraid to tell himself, and every decent man has a considerable number of such things stored away. That is, one can say that the more decent he is, the greater the number of such things in his mind. (1864/1972:55)

However, the research has also suggested that implicit attitudes can be changed. In 2001, Greenwald and Dasgupta found that by exposing participants to pictures of a range of admired Black Americans such as Martin Luther King and disliked White Americans such as Al Capone, the pro-White effect usually found in the race IAT was substantially reduced, immediately after and even twenty-four hours after the initial exposure. While this was only a temporary modification, there is the possibility that being consistently exposed to exemplars of admired Black people (particularly in the media) could lead to more permanent changes in underlying implicit attitudes.

It just so happened that Laura took the US Election 2008 IAT on Project Implicit in the week after Barack Obama was elected as the next US President. Her results speak for themselves: ‘Your data suggests a strong automatic preference for Black people over White people’ and ‘Your data suggests a strong automatic preference for Barack Obama over John McCain’. This could be an example of the malleability of implicit attitudes operating in the real world; all the positive exposure to Obama during that week, in all probability, had a significant effect on Laura’s implicit attitude. Continued positive exposures to Black role models could lead to more permanent positive associations for Black people in general.

At present there are something like fifteen versions of the IAT online at Project Implicit:

✵ Disability IAT

✵ Age IAT

✵ Gender-Science IAT

✵ Asian IAT

✵ Arab-Muslim IAT

✵ Native IAT

✵ Religion IAT

✵ Sexuality IAT

✵ Obama-McCain IAT

✵ Skin-tone IAT

✵ Weight IAT

✵ Presidents IAT

✵ Weapons IAT

✵ Gender-Career IAT

✵ Race IAT

For a measure of unconscious processing, engaging on the IAT is an oddly self-conscious process. I am strangely anxious every time I do it, maybe because I think that this may reveal the uncomfortable truths about me. It is a quick test, almost too quick, and the computerised IAT flashes the results up at you without embarrassment or pause after the completion of each test. You sit nervously by the screen prepared to view your own prejudices and biases, secretly hoping that none will be revealed. Or at worst, hoping to see just a slight prejudice in your reaction times and error rates, and the expression ‘Your data suggest a moderate preference for X over Y.’ Laura and I both sat all the tests that we thought might help produce a reasonable psychological profile for each of us, one after the other like a set of challenges. The results are given in Table 5.1.

Table 5.1 Our own IAT results



Obama-McCain IAT

Your data suggest a strong automatic preference for Barack Obama over John McCain

Your data suggest a strong automatic preference for Barack Obama over John McCain

Race IAT

Your data suggest a moderate automatic preference for European American over African American

Your data suggest a moderate automatic preference for European American over African American

Skin-tone IAT

Your data suggest a moderate automatic preference for light skin over dark skin

Your data suggest a strong automatic preference for light skin over dark skin

Weight IAT

Your data suggest a strong automatic preference for thin people compared to fat people

Your data suggest a moderate automatic preference for thin people compared to fat people

Sexuality IAT

Your data suggest a moderate automatic preference for straight compared to gay people

Your data suggest a moderate automatic preference for straight compared to gay people


Your data suggest no automatic preference for young compared to old.

Your data suggest a slight automatic preference for young compared to old.

We discovered that we were both strongly pro-Obama, which was fine, even a little reassuring (to my conscious mind). I admit that I could never take John McCain’s voice seriously, because of its pitch and general tone and the fact that it sounded like something computer-generated by Disney, and I am sure that Mahmoud Ahmadinejad, Hassan Nasrallah and Muqtada al-Sadr couldn’t take it too seriously either. There was not much threat in that voice and the danger would always have been that the voice would have had to be backed up with what the military, and the new head of Central Command General David Petraeus, were now calling ‘kinetics’ (a term interestingly borrowed from mainstream psychology but now being used to refer to military action rather than action in general). After that I either had no preference (age) or a series of moderate preferences, except when it came to weight, where I had a strong preference for thin people compared to fat people (Laura’s only other strong preference was for light skin over dark skin).

So what does any of this mean? Can I find any evidence from my own life that the implicit attitudes revealed by this test have any substantive or actual behavioural implications? I think that the answer with respect to my one strong prejudice (other than Obama) is probably yes. However, the behaviour is not to do with actual discrimination against fat people but behaviour directed against myself, and not dieting but something else. I have always been a compulsive runner, and compulsive here means compulsive, every day without fail, in any country no matter how inconvenient or difficult: through the centre of Tokyo at 5.00 a.m. because the flight back to the UK was to leave early; along a motorway in Sweden in the middle of the night in a snowstorm without a clue as to which direction led back to the centre of Gothenburg; along the Pacific Coast highway in California, just off the plane and suffering from jet lag, with no pavement for protection and with wide gaudy red and yellow trucks almost brushing my legs.

I was a child with a chubby face, never fat, but I am sure that strangers might have thought that I was fat because of my face (‘you had a face like the moon when you were a baby,’ my mother used to say proudly; ‘like the moon,’ and she would smile broadly whenever she said it, as if the memory made her happy) and I like the way that running makes my face look lean. I refuse to go on television unless I have a run first (many television producers will vouch in frustration for this fact).

But my one compulsion runs deeper than mere misplaced vanity, more rooted I am sure in my unconscious mind. This compulsion started when I was at school. I would run every day and twice on a Tuesday and a Thursday. I started when I was thirteen. A lot happened that year - I broke my arm doing judo, which meant that many sports were for a long while out of bounds, and my father died unexpectedly of some heart-related condition. I never understood what he died of, it was never properly explained to me as a child, and I am not sure that my mother properly understood what had happened anyway, except that it happened during an operation. I suppose that this made the fear more intense, the fear of life being interrupted in a sudden and unexpected way. I might have a similar congenital weakness, so I decided to get fit, and run and run to make my heart stronger and stronger. I never stopped. Running, we all know, can be very addictive.

But I have another image from that one life-changing year as well. An image that has never faded or been diminished by time, an image that has been silent and never discussed until now. An image, nevertheless, that has haunted me. On the night of my father’s funeral, after we had laid him to rest in Roselawn Cemetery with the wind lifting the dirty green carpet used to cover the wet clay grave, we were in my aunt’s house for the sandwiches and tea because our own house (a two-up, two-down in North Belfast) wasn’t big enough. Everyone was there, drinking quietly, the quiet, subdued sobbing made worse by the image of the coffin juddering down into its final position: everyone except my cousin Myrna, that is, who inexplicably had gone to work that day. Nobody had explained why. Life at the time seemed to be full of things that were never properly explained, at least to a thirteen-year-old boy. My cousin walked in right in the middle of the wake. She seemed to cling to the doorframe, not entering, just standing there, staring at us all; and I can picture her now, an image etched on my mind for ever, an emaciated grey ghost, already dead in the eyes and the mouth. I had heard, overheard, that she had got some kind of eating disorder, anorexia nervosa - ‘slimmer’s disease’, my mother called it - but I hadn’t seen her for months, as the slimmer’s disease took hold. She avoided seeing relatives. But there she stood in the silence and the sadness, and everyone looked at her and nobody said a thing, as if she looked normal and healthy and was just late for the funeral. I think she walked slowly past us all into the kitchen to stand alone, the place where food is prepared and eaten, but not for her.

She died a couple of weeks later of pneumonia and was buried in the row opposite my father, which is handy from the point of view of people bringing flowers to either grave. Her mother, my Aunt May, a sweet, lovely woman with a giggly, girly voice, always said that what triggered the anorexia was a chance remark from a doctor at work during a routine medical examination, a remark that she was a little overweight. From that day on, my aunt always said, she never ate properly again. It sounds almost ridiculous that such a life-threatening disorder could be triggered in this way, but years later I supervised a postgraduate student who analysed the social construction of anorexia in the families of sufferers and the number of interviewees who pointed to a similar ‘chance remark’ as the cause of the whole thing was extraordinary.

Anorexia is a complex disorder with cultural, personality and biological factors all implicated in its ontogenesis, but human beings like to identify a single cause that they can pick out and say ‘if only that hadn’t happened …’. This single cause is usually something fairly random (so that any random family could potentially be affected) and external to the family (so that no blame could be attached to the family). The PE teacher who commented that Tracy was too fat to be any good at games, the boyfriend who said that Jane’s bum was too big for her skinny jeans, the doctor who quipped that his patient could do with losing a little weight. It was always things like that. It reminded me of what Friedrich Nietzsche wrote: ‘To trace something unknown back to something known is alleviating, soothing, gratifying, and gives us moreover a feeling of power. Danger, disquiet, anxiety attend the unknown, and the first instinct is to eliminate these distressing states. First principle: any explanation is better than none’ (1871/1962:62).

My family, and many other families, had found an explanation, one that stressed the power of the word, of the chance remark (and I suppose, by implication, the dangerous power of the carelessness of those in authority). Nobody ever disputed my aunt’s account, and it became the true version of what had become of my beautiful cousin, and in that awful year of my life I sometimes think that my weight prejudices were probably laid down for ever.

Of course, this story may tell you why weight is an important issue for me, but why have I ended up with an anti-fat prejudice, why not an anti-thin prejudice? After all, it was not the fact that Myrna was a few pounds overweight (maybe more, maybe less) that killed her. Well, maybe it was the implicit message in the story, the implicit message being that if you are overweight then you can be killed by a chance remark, the unconscious message being that being fat makes you too sensitive to others’ insensitivity, the unconscious theme being that being fat means that others can control your life, and even your death. My compulsive running may just reflect my unconscious desire to escape from my father’s destiny, but it may also reflect this deep-seated desire to put myself out of harm’s way from chance remarks (and thereby make myself less vulnerable in life). It may be core to my psychological make-up and mean that I have an implicit and unconscious bias against fat people, who have not made the effort to shield themselves in this way. Of course the fact that my implicit attitude actually does connect to some core behaviours in my everyday life, namely my determination to run, is very encouraging from the point of view of my current academic concerns. It is also, of course, more than a little depressing for me.

But for the psychologist wanting to save the planet, the big question is whether the IAT gives us more insight than the measures of explicit attitudes. The beauty of the IAT is that because it measures automatically activated associations, it is resistant (some have argued ‘immune’) to faking (see Greenwald, Poehlman, Uhlmann and Banaji 2009). Would it, therefore, help us to understand and predict actual behaviour more effectively? The explicit tests that we had carried out had revealed that explicit attitudes are very pro-environment, so why wasn’t people’s behaviour falling in line with this? Would the IAT reveal something quite different here? Would this help explain what was going on? The focus on behavioural prediction was, of course, inevitable from the inception of the IAT and ‘plagued’ the IAT very early on (in the eyes and words of some of the core researchers). Banaji (2001) commented that ‘tolerance’ was needed if this question was to be answered: ‘Pushing fast and furiously to “show me what predicts” may be counterproductive. One first needs to understand the construct before asking what it may or may not predict’ (2001:132).

Greenwald et al. (2009) conducted a meta-analysis on the predictive validity of the IAT and concluded that, in general, when the IAT and explicit attitude measures are combined, they are better predictors of behaviour than either measure alone. However, when attitudes are ‘socially sensitive’, and where social desirability concerns are inevitably present (as in attitudes to race, age, gender or the environment), explicit measures are very poor predictors of behaviour and in these situations the IAT would appear to be a much better predictor of behaviour than explicit measures.

Recently, Greenwald along with other researchers has started to apply the IAT to consumer research since it has become increasingly apparent that consumer behaviour does not necessarily involve conscious and rational decision-making, but can also be influenced by all sorts of unconscious factors. You can often see people in the supermarket picking up products without even properly looking at them, let alone making complex decisions based on price or nutritional value or fat content. There is something at work here that is not based on conscious, reflexive, rational thought and slow decision-making. And, of course, the IAT could be particularly useful in the domain of green consumerism given the evident attitude-behaviour gap in the purchasing of green goods. If implicit attitudes were measured in this particular domain, then a less optimistic view of environmental thinking may well be revealed, which reflects all the other concerns that consumers have about things like price and convenience, as well as potentially their concerns regarding their lack of knowledge about carbon footprint, and their general anxiety as to whether one person’s individual shopping behaviour can actually make a difference.

From the meta-analysis conducted by Greenwald et al. (2009), the overall conclusion would seem to be that the IAT does significantly predict behaviour (although the actual level of prediction can be modest at times). But as Gregg (2008) has commented, ‘the behaviours documented are often quite specific, so it is striking that general implicit associations predict them at all. Moreover, the IAT outstrips self-report in forecasting instances of discrimination and prejudice. Hence, it offers some genuine diagnostic advantages’ (2008:765). There appear to be something like fifteen published studies that have used the IAT in the consumer domain, as shown in Table 5.2. This table has been adapted from the Greenwald et al. (2009) paper, but I have tried to make the conclusions as specific and as accurate as possible by quoting directly from the original studies.

The trend in these results appears to be fairly positive with respect to the predictive power of the IAT and particularly under certain types of conditions. When people are under any kind of time pressure (like shopping in a supermarket, particularly just before closing time), or when they are having to control their emotional state (like shopping in a supermarket after a hard, stressful day at work), or when they are focusing on the kinds of enjoyment that they might get from their food choices (like shopping in a supermarket after a hard day at work and looking forward to dinner), or when they are under the influence of alcohol, the IAT is a very good predictor of behaviour. Measures of implicit attitude seem to predict behaviour best when the person concerned is under any kind of mental or emotional or time pressure: when there is a lot going on, in other words. Explicit measures, on the other hand, are usually better

Table 5.2 The IAT as a predictor of consumer behaviour


Behavioural measure

Did the IAT successfully predict behaviour?

Brunel, Tietje and Greenwald (2004)

Study 1: Self-report of ownership and usage frequency of Mac and PC.

Both IAT and explicit attitude measure predicted ownership and usage.

Friese, Hofmann and Wänke (2008)

Study 1: Behavioural choice task between apples and chocolate where working memory capacity is reduced.

When processing resources are ‘ample’, explicit attitude measure is a better predictor of behaviour. When processing resources are ‘taxed’, ‘behaviour appeared to be more strongly driven by impulsive processes as indicated by the increase in the implicit measure’s predictive validity’.

Friese, Hofmann and Wänke (2008)

Study 2: Consumption of potato crisps after watching a film where emotions were either controlled (depleting ‘self-regulatory strength’) or not controlled.

‘… when participants were depleted of their self-regulatory strength [by having to suppress their emotional response to a film], not only did the implicit measure gain considerable predictive power compared with the control condition but also the explicit measure was now unrelated to potato crisps consumption.’

Friese, Hofmann and Wänke (2008)

Study 3: Beer consumption after watching a film where emotions were either controlled (depleting ‘self-regulatory strength’) or not controlled.

‘When resources were scarce [because participants had to suppress their emotional response to a film] the implicit measure predicted behaviour well and showed incremental validity over and above both explicit self-report measures at the same time.’

Friese, Wänke and Plessner (2006)

Brand choice between generic and branded products in experimental conditions either under time pressure (5 seconds to make their choice) or not under time pressure (unlimited time).

‘Participants whose explicit and implicit preferences regarding generic food products and well-known food brands were incongruent were more likely to choose the implicitly preferred brand over the explicitly preferred one when choices were made under time pressure. The opposite was the case when they had ample time to make their choice.’

Gibson (2008)

Brand choice between Coke and Pepsi in experimental conditions when cognitive load was manipulated (by asking participants to remember an 8 digit number, or not).

‘… choice in this high load condition was related to implicit attitudes, while choice in the low load condition was not.’

Hofmann and Friese (2008)

Candy consumption when participants had been drinking alcohol or not.

‘Specifically, the predictive validity of implicit attitudes (as part of the impulsive system) was markedly increased for participants who had consumed alcohol as compared with sober participants.’

Hofmann, Rauch and Gawronski (2007)

Candy consumption after watching a film and being asked either to suppress emotions (depletion condition) or to ‘let emotions flow’ (control condition).

‘… automatic candy attitudes showed a positive correlation to candy consumption in the depletion condition but not in the control condition. That is, candy consumption significantly increased as a function of automatic positivity toward the candy in the depletion condition but not in the control condition.’


Behavioural measure

Did the IAT successfully predict behaviour?

Karpinski and Hilton (2001)

Study 2: Behavioural choice between apples and candy bars.

‘… explicit attitudes and the IAT are independent … explicit attitudes predicted behaviour but the IAT did not.’ (Note: there was no time pressure/drain on cognitive resources etc. operating here. In this study participants ‘were informed that they could choose only one of the objects [apple or candy bar] to eat or to take home with them.’)

Karpinski and Steinman (2006)

Study 1: Brand choice between Coke and Pepsi.

Both IAT and explicit measures predicted choice of branded drink.

Karpinski, Steinman and Hilton (2005)

Voting intention in the 2000 US Presidential election and also brand choice between Coke and Pepsi.

‘… explicit attitude measures were better predictors of deliberative behaviours than IAT scores’ (emphasis added).

Maison, Greenwald and Bruin (2001)

Study 1: Self-reported drinking of juices or soda, and self-reported dieting.

‘… significant correlation between the IAT and Ss’ self-reported behaviour’ (emphasis in original).

Maison, Greenwald and Bruin (2004)

Study 1: Self-reported consumption of yoghurt brands/ eating at different fast food restaurants/consumption of Coke or Pepsi.

‘A meta-analytic combination of the three studies showed that the use of IAT measures increased the prediction of behaviour relative to explicit attitude measures alone.’

Olson and Fazio (2004)

Study 3: Self-reported behaviour of apple and candy bar consumption.

IAT predicted behaviour, particularly a more personalized IAT. ‘… the personalized IAT correlated more strongly with explicit measures of liking, past eating behaviour, and behavioural intentions than did the traditional IAT.’

Scarabis, Florack and Gosejohann (2006)

Choice between chocolate and fruit.

The IAT was a good predictor of actual choice, ‘… people rely more on automatic preferences that are independent from higher-order appraisals when they focus on their affective responses [what enjoyment they might get from the food] than when they think about the advantages and disadvantages of choice options.’

Swanson, Rudman and Greenwald (2001)

Study 2: Self-reported smoking behaviour or vegetarianism/non-vegetarianism.

The IAT and explicit attitude measures did predict vegetarianism/non-vegetarianism but not smoking.

Vantomme, Geuens, De Houwer and De Pelsmacker (2005)

Self-reported purchase intentions for real and fictitious brands of green and environmentally unfriendly cleaning products.

‘The IAT, but not the explicit difference score, differentiated between respondents intending to buy the real ecological all-purpose cleaner and those intending to buy the real traditional all-purpose cleaner.’

when there is no mental load or time pressure, when there is all the time in the world, thereby allowing the person to make slower, deliberate and reflective behavioural decisions. Something like supermarket shopping, however, is not, generally speaking, a slow, deliberate, reflective process for most people. It is fast and non-reflective and sometimes quite hectic. So, in a context like this, the IAT should be a much better predictor of consumer behaviour than a measure of explicit attitudes.

Only one study seems to have applied the IAT to actual green consumerism. Vantomme and colleagues in 2005 conducted an experiment that looked at implicit and explicit attitudes towards green cleaning products. They expected to find that implicit attitudes would reveal less positive results than the explicit attitude measures (because of the social desirability factors relating to green behaviour). In their first experiment they used fictitious cleaning products, introduced to participants in a ‘learning phase’ where participants were informed that one product was environmentally friendly while the other was harmful to the environment. They used fictitious products because of the dangers of brand image impacting on the results. What they found, however, was that in contrast to what they had hypothesised, implicit attitudes towards the fictitious green cleaning products were far more positive than explicit attitudes.

In a second experiment, real brands were used instead. This time, there was no difference in implicit and explicit attitudes towards green cleaning products. So the results from this study are a little inconclusive with respect to the underlying implicit attitudes towards green products that people might actually hold. But both sets of results went against the original hypothesis. This study, therefore, left our own empirical investigation into implicit attitudes wide open, which made the whole thing, of course, that much more exciting.