Cognitive dissonance reduction

Following on from my earlier post about the way psychologists look at the world, let me tell you a story which I think illustrates very well the tendency academic psychologists have for reductionism. It’s a story about a recent paper on the phenomenon of cognitive dissonance, and about a discussion of that paper by a group of psychologists that I was lucky enough to be part of.

Cognitive Dissonance is a term which describes an uncomfortable feeling we experience when our actions and beliefs are contradictory. For example, we might believe that we are environmentally conscious and responsible citizen, but might take the action of flying to Spain for the weekend. Our beliefs about ourselves seem to be in contradiction with our actions. Leon Festinger, who proposed dissonance theory, suggested that in situations like this we are motivated to reduce dissonance by adjusting our beliefs to be in line with our actions.

Obviously after-the-event it is a little too late to adjust our actions, so our beliefs are the only remaining point of movement. In the flying to Spain example you might be motivated by cognitive dissonance to change what you believe about flying: maybe you come to believe that flying isn’t actually that bad for the environment, or that focussing on personal choices isn’t the best way to understand environmental problems, or you could even go all the way and decide that you’re not an environmentally responsible person.

The classic experiment of dissonance theory involved recruiting male students to take part in a crushingly boring experiment. The boring part was an hour of two trivial actions — loading spools into a tray, turning pegs a quarter-turn in a peg-board. At the end of this, after the students through the experiment was over, was the interesting part of us. The students were offered either $1 or $20 to tell the next participant in the experiment (actually the female accomplice of the experimenter) that the experiment she was about to do was really enjoyable. After telling this lie, the participants were then interviewed about how enjoyable they really found the experiment. What would you expect from this procedure? Now one view would predict that the students paid $20 would enjoy the experiment more. This is certainly what behaviourist psychology would predict — a larger reward should produce a bigger effect (with the effect being a shift from remembering the task as boring, which is was, to remembering it being enjoyable, which getting ¬£20 presumably was). But cognitive dissonance theory suggests that the opposite would happen. Those paid $20 would have no need to change their beliefs about the task. They lied about how enjoyable the task was to the accomplice, something which presumably contradicted their beliefs about themselves as nice and trustworthy people, but they did it for a good reason, the $20. Now consider the group paid only $1. They lied about how enjoyable the task was, but looking around for a reason they cannot find one — what kind of person would lie to an innocent for only $1? So, the theory goes, they would experience dissonance between their actions and their beliefs and reduce this by adjusting their beliefs: they would come to believe that they actually did enjoy the boring task, and this is the reason that they told the accomplice that it was enjoyable. And, in fact, this is what happened.

At this point I want you to notice two things about cognitive dissonance. Firstly, it requires the existence of quite sophisticated mental machinery to operate. Not only do you need to have abstract beliefs about the world and yourself, you need to have some mechanism which detects when these beliefs are in contradiction with each other or with your actions, and which can (unconsciously) adjust selective beliefs to reduce this contradiction. The second thing to notice is that all this sophisticated mental machinery is postulated to exist from changes in behaviour, it is never directly measured. We don’t have any evidence that the change in attitudes really does result from an uncomfortable internal state (‘dissonance’) or that any such dissonance does result from an unconscious perception of the contradiction between beliefs and actions.

So, to the recent paper and to reductionism. The paper, by Louisa Egan and colleagues at Yale [ref below] is titled ‘The Origins of Cognitive Dissonance‘, and represents one kind of reductive strategy that psychologists might employ when considering a theory like cognitive dissonance. The experiments in the paper (summarised here and here) both involved demonstrating cognitive dissonance in two groups which do not have the sophisticated mental machinery normally considered necessary for cognitive dissonance — four year-old children, and monkeys. The reductionism of the paper, which the authors are quite explicit about, is to show that something like cognitive dissonance can occur in these two groups despite their lack of elaborate explicit beliefs. Unlike the students in Festinger’s classic experiment we can’t suppose that the children or the monkeys have thoughts about their thoughts in the same way that dissonance theory suggests.

To demonstrate this the authors employed an experimental method that could be used with subjects who did not have language, but would still allow them to observe the core phenomena of dissonance theory — the adjusting of attitudes in line with previous actions. The method worked like this. For each participant — be they a child or a monkey — the experimenters identified three items (stickers for the children, coloured M&M’s for the monkeys) which the participant preferred equally. In other words, if we call the three items A, B and C then the child or monkey liked all of the items the same amount. Then the experimenter forced the participating child or monkey to choose between two of the items (lets say A and B), so that they only got one. Next the child or monkey was offered a choice between item C and the item they did not choose before. So, if the first choice was between A and B and the participant chose A, then the next choice would be between B and C. What does dissonance theory predict for this kind of situation? Well, originally the three items are equally preferred — that’s how the items are selected. After someone is forced to make a first choice, between A and B, cognitive dissonance supposedly comes into play. The participant now has a reason to adjust their attitudes, and the way they do this is to downgrade their evaluation of their unchosen item. This will is known as being happy with what you got or “I must not like B as much, because I chose A”. So on the second choice (B vs C) the participants are more likely to choose C (more likely that chance, and more likely than a control group that goes straight to the ‘second’ choice). This prediction is exactly what the experimenters found, in both children and monkeys, and the startling thing is that this occurred despite the fact that we know that neither group was explicitly talking to themselves in the way I outline the dissonance theory prediction above (“I must not prefer B as much…etc”). Obviously something like cognitive dissonance can be produced by far simpler mental machinery than that usually invoked to explain it, conclude the experimenters. In this way, The paper is a call to reduce the level at which we try and explain cognitive dissonance.

How far should you go when trying to reduce the level of theory-complexity that is needed to explain something? Psychologists know the answer to this immediately — as far as possible! So when our happy band of psychologists got to discussing the the Egan paper it wasn’t long before someone came up with a new suggestion, a further reduction.

What if, it was suggested, there was nothing like dissonance going on in the Egan et al experiments? After all, there was no direct measurement of anxiety or discomfort, so why suppose that dissonance occurred at all — perhaps, if we can come up with a plausible alternative, we can do away with dissonance all together. Imagine this, see if you find it plausible: all of us, including monkeys and children, possess a very simple cognitive mechanism which saves us energy by remembering our choices and, when similar situations arise, applying our old choices to new situations, thus cutting down on decision time. That sounds plausible, and it would explain the Egan et al results if you accept that the result of the first, A vs B, decision is not just “choosing A” but is also “not choosing B”. So, when you get to the second choice, B vs C, you are more likely to choose C because you are simply re-applying the previous decision of “not choosing B”, rather than performing some complicated re-evaluation of your previously held attitudes a-la cognitive dissonance theory.

At this point in the discussion the psychologists in the room were feeling pretty pleased with themselves — we’d started out to with cognitive dissonance, reduced the level of complexity of mental processes required to explain the phenomenon (the Egan et al result) and then we’d taken things one step further and reduced the complexity of the phenomenon itself. At this point, we had a discussion of how widely the ‘decisional inertia’ reinterpretation could be applied to supposed cognitive dissonance phenomena. Obviously we’d have only been really satisfied with the reinterpretation if it applied more widely than just to this one set of experiments under consideration.

But further treats were in store. What if we could reduce things again, what if we could make even simpler the processes involved? We’d already started to convince ourselves that the experimental results could be produced by simple cognitive processes rather than complex cognitive processes, perhaps we could come up with a theory about how the experimental results can be produced without any cognitive processes at all! Now that would be really reductive.

Here’s what was suggested, not as definitely what was happening, but as a possibility for what could potentially be happening — and remember, if you are sharing a table with reductionists then they will prefer the simple theory by default because it is simpler. You will need to persuade them of the reasons to accept any complex theory before they abandon the simple one. Imagine that there is no change at all going on in the preferences of the monkeys and the children. Instead, imagine — o the simplicity! — that any participant in the experiment merely has a set of existing preferences. These preferences don’t even have to be mental, by preferences all I mean are consistent behaviours towards the items in question (stickers for the children, M&Ms for the monkeys). From here, via a bit of statistical theory, we can produce the experimental result with out any recourse to change in preferences, cognitive dissonance or indeed anything mental. Here’s how. Whenever you measure anything you get inaccuracies. This means that your end result reflects two things: the true value and some random ‘noise’ which either raises or lowers the result away from the true value. Now think about the Egan et al experiment. The experimenters picked three items, A, B and C, which the children or monkeys ‘preferred equally’, but what did this mean? It meant only that when the experimenters measured preference their result was the same for items A, B and C. And we know, as statistically-savvy psychologists, that those results don’t reflect the true preferences for A, B and C, but instead reflect the true preferences plus some noise. In reality, we can suppose, the children and monkeys actually do prefer each item differently from the others. Furthermore this might even be how they make their choice. So when they are presented with A vs B and choose A it may be because, on average, they preferred A all along. Now watch closely at what happens next. The experimental participants are given a second choice which depends on their first choice. If at first they chose A over B then the second choice is B vs C. But if they chose B over A then the second choice is A vs C. We know the results: they then choose C more than the unchosen option from the first choice, be it A or B, but now we have another theory as to why this might be. What could be happening is merely that, after the mistaken equivalence of A, B and C, the true preferences of the monkey or child are showing through, and the selective presentation of options on the second choice is making it look like they are changing their preferences in line with dissonance theory. Because the unchosen option from the first choice is more likely to have a lower true preference value (that, after all, may be why it was the unchosen option), it is consequently less likely to be preferred in the second choice, not because preferences have changed, but because it was less preferred all along. In the control condition, where no first choice is presented, their is no selective presentation of A and B and so the effect of the true values for preferences for A and B will tend to average out rather than produce a preferential selection of C.

Now obviously the next step with this theory would be to test if it is true, and check some details which might suggest how likely this is. Did Egan et al assess the reliability their initial preference evaluation? Did they test preferences and then re-test them at a later date to see if they were reliable? These and many other things could persuade us that such an explanation might or might not be very likely. The important thing, for now, is that we’ve come up with an explanation that seems as as simple as it could possibly be and still explain the experimental results.

For psychologists, reductionism is a value as well as a habit. We seek to use established simple features of the mind to explain as many things as possible before we get carried away with theories which rely on novel or complex possibilities. The reductionist position isn’t the right one in every situation, but it is an essential guiding principle in our investigations of how the mind works.

Link: to earlier post about how psychologists think.

References

Egan L. C., Santos L.R., Bloom P. (2007). The Origins of Cognitive Dissonance: Evidence from Children and Monkeys. Psychological Science, 18, 978-983.

Festinger, L. and Carlsmith, J. M. (1959). http://psychclassics.yorku.ca/Festinger/”>Cognitive consequences of forced compliance. Journal of Abnormal and Social Psychology, 58, 203-211.

8 Comments

  1. edmo
    Posted December 18, 2007 at 6:59 pm | Permalink

    Tom/Vaughan – thanks for all the great stuff you guys write for Mindhacks.
    Re. the kids & monkeys and choices between A, B, and C – though I’m just a joe layperson, this article got me thinking about a couple of things that I thought might be useful if shared.
    You mention a couple of alternatives to cognitive dissonance being the cause of the behaviour – including that their preferences remained static but were not accurately measured at the outset. A relatively simple alternative is that this is not a demonstration of cognitive dissonance reduction, or of static but initially inaccurately measured preferences, but of preferences that simply changed due to the effects of new experiences and new information.
    Perhaps the kids/monkeys truly did like A, B, and C equally at the outset, but when presented with A and B only, and then forced to choose A, they interpreted this new info/experience in a way that said, “B is inferior to A”, or, “B sucks in general”, or “Why isn’t C part of this game…why can’t I have C?”
    To me the experiment’s results could simply be a matter of new experiences and information leading to new conclusions and preferences. There is not necessarily a need to look at static, unchanging conclusions and preferences (which I think are less likely), or at revisionist history/dissonance reduction.
    One additional challenge I see with calling the kids’/monkeys’ behaviour dissonance reduction is that there is no measurable evidence of their minds distorting the truth about their actions or their beliefs (i.e. no evidence that they were trying to convince themselves and/or the experimenters that they had always preferred C over B, even though they initially indicated they liked C and B equally.) While I don’t think there is anything in this experiment that rules out cognitive dissonance and dissonance reduction as possible causes or contributing factors to the kids/monkeys behaviour, to me the evidence here only allows us to surmise that either their preferences changed, or else their preferences were not accurately indicated/measured at the outset.
    Curious to know anyone’s thoughts. Thanks for reading.

  2. Posted December 18, 2007 at 7:36 pm | Permalink

    As a general point about reduction, I agree within a single theory domain, but I see psychologists as much more able than other disciplines to entertain parallel explanations using different levels of explanation.
    For example, being quite happy to work with theories at both the biological and psychological levels, but within both, aim for Ockham’s Razor.

  3. Bob
    Posted December 19, 2007 at 6:17 pm | Permalink

    Ditto edmo’s props to you on MindHacks… great stuff. Not sure about the monkeys, but we do know that preference assessment methods in humans can generate reversals. A “fill in the blank” assessment (which is pretty much what the kids did in this study to determine A = B = C) lead to different preference orderings than binary choice approaches. This effect could lead to the effect observed. See an old paper I co-authored on this issue at http://mdm.sagepub.com/cgi/content/abstract/21/3/208.

  4. devenant
    Posted January 19, 2008 at 10:13 pm | Permalink

    The experiment has some resemblance to the Monty Hall problem, though I haven’t thought through the details. Wouldn’t it be nice if you could completely reduce the explanation to Bayesian math? ;)

    http://en.wikipedia.org/wiki/Monty_Hall_problem

    • parreira
      Posted October 17, 2012 at 2:47 pm | Permalink

      Regarding the Monty Hall Problem, I recall the Mythbusters episode where they tested it.
      It was interesting to watch the psychological side of their experiment. Nobody (or almost nobody) chose to change the original choice. This can be, indeed, related to the A, B and C experiment. If the original choice is A, we can interpret it as not-B. So, when the choice is between B and C, and if the subject prefered not-B, the “obvious” choice would be C.

  5. Enya
    Posted April 13, 2014 at 11:07 am | Permalink

    How do we explain that after we make the choice and the cognitive dissonance played its role, we can still regret our original choice? The grass is always greener on the other side?

  6. David Winter
    Posted August 27, 2014 at 10:01 am | Permalink

    There would seem to be some support for the theory that irrationality can be explained by applying noise to rational choices – https://news.usc.edu/29328/Irrationality-May-Be-the-Result-of-Mental-Noise/

  7. David Winter
    Posted August 27, 2014 at 10:11 am | Permalink

    Applying this to more complex cases of cognitive dissonance. Maybe what we are seeing is an exposure of the difference between beliefs as they are expressed (espoused theory) and the beliefs that are actually held (theory in practice).

    Taking your example of flying to Spain. The actual belief held might be “I believe in supporting the environment to the extent that it doesn’t inconvenience me.” However, when asked, the person might express their belief in simpler, less conditional terms or in a way that they perceive to be more socially-acceptable. But their actions reveal the true beliefs that were there all the time.


Post a Comment

Required fields are marked *
*
*

Follow

Get every new post delivered to your Inbox.

Join 24,095 other followers