(This is a guest post by Jake Quilty-Dunn, replying to my reply to his original post on on rationalization and (ir)rationality. 1000 words; 5 minute read.)
Thanks very much to Kevin for inviting me to defend my comparatively gloomy picture of human nature on this blog, and for continuing the conversation with his thoughtful reply.
Kevin wonders whether negative affect might signal not cognitive dissonance per se, but rather an impetus to do some high-effort problem solving. This is an interesting—and empirical—question.
Fortunately, there is some relevant evidence. Negative affect correlates with worse performance on problem solving. It also hampers learning how to solve puzzles and transferring that learning to novel cases. And there’s some evidence that negative affect specifically makes people more likely to engage in low-effort as opposed to high-effort problem-solving strategies.
I pause to note that there is disagreement about this issue, and the evidence is nuanced in a way that makes it hard to capture in this short blog post. Schwarz and colleagues argued that negative affect enhances detail-oriented vs. heuristic-based cognition. A later review challenged this hypothesis, as does the more recent evidence just cited, though other results suggest negative affect can indeed re-orient problem-solving strategies toward seeking additional information. Given the evidence for impaired problem-solving performance overall, however, there is basis for skepticism about the idea that dissonance effects could be explained by a more general phenomenon of negative affect spurring on problem solving.
Interestingly, a fairly recent study provides evidence that confusion can aid learning, but the experimenters also found that self-reports of confusion failed to correlate with confusion induction. The negative affect of dissonance, however, is typically reportable (even if participants are unaware of its source). Thus the sort of confused feelings Kevin suggests drive effortful problem-solving differ from dissonance in key ways.
Some theories posit confusion-like feelings as a spur to further cognition, but don’t assign them a negative valence. Gopnik uses the provocative (if slightly cringey) metaphor of orgasm to describe the feeling of successfully explaining some phenomenon. Even if Gopnik’s hypothesized “theory drive” may provoke negative feelings when frustrated by a particularly confusing problem, it doesn’t seem to come prepackaged with negative valence as dissonance does. Similarly, Fingerhut and Prinz argue that the “cognitive perplexity” that motivates exploratory behavior is a key component of the (typically positive) feeling of wonder underlying aesthetic experience. These cognitive emotions don’t seem to have the affective profile of dissonance.
Another relevant point concerns the factors that mediate the generation of negative affect in dissonance experiments. As I argued in my original post, these seem to include self-esteem in a way that would be hard to explain if the affect was a general problem-solving-motivator. This brings us to Kevin’s next point: perhaps dissonance effects can arise in third-person as well as first-person cases.
Fortunately, again, there is relevant evidence. One of the interesting turns in 21st-century dissonance research concerns so-called “vicarious dissonance”. People can indeed experience dissonance when viewing others engage in counter-attitudinal behavior––but only when subjects share a strong group identity with the people they’re observing. Students who strongly identify with their college are more likely to rationalize and shift their own attitudes when observing somebody from their own college freely engage in counter-attitudinal behavior, but not when observing somebody from another college do the same thing. These effects aren’t driven by empathy or perspective-taking, but instead rely on the self-concept of the observer.
Effects of this sort are also increased when subjects see themselves as similar to the in-group member. In one study, subjects heard similar in-group members preach the values of sunscreen and later mention that they hypocritically fail to use sunscreen regularly; the vicarious dissonance caused by this hypocrisy led subjects to be more likely to take some sunscreen a moment later if the observed similar in-group member freely chose to make the hypocritical statement. When subjects are told that the lights and ventilation system may cause discomfort—similar to the placebo pill in Zanna and Cooper’s classic study—they’re less likely to grab sunscreen afterwards, strongly suggesting that this effect is mediated by the psychological discomfort characteristic of dissonance. These results indicate that vicarious dissonance involves threats to the subject’s self-image that are filtered through shared group identity rather than a rational goal of working out what people’s attitudes are.
Finally, Kevin asks why dissonance should arise in cases where evidence contradicts negative aspects of our self-image. I think there are speculative but plausible irrationalist explanations. For example, suppose evidence of a particular failing (e.g., I’m bad at sports) originally caused dissonance, but over time you adjust and maintain a set of beliefs that you’re comfortable with (e.g., I’m bad at sports, but it doesn’t matter because sports are dumb). Now if you face evidence that you don’t have that failing, all of your subsequent rationalizations are at risk, creating new openings for harmful conclusions about yourself that you haven’t yet built up defenses against (e.g., If I’m good at sports, then I was wrong to shy away from them; and I was wrong to think they’re dumb; and I was wrong to focus on X instead…). If belief updating has an immunodefensive function, it makes sense that there should be a drive to stick with old problems for which you’ve already developed cognitive coping strategies.
That said, the story I’ve been pushing for is compatible with the co-presence of rationally motivated drives for consistency, and/or more primitive analogues of dissonance. Thus it’s worth ending on a conciliatory note: the picture I’ve sketched of a psychological immune system is not only compatible with other rational forms of belief updating, it really only makes sense alongside rational updating. The empirical evidence shows that people are extremely good at inferring consequences of their behavior for their self-image, and it is only because these inferences are so sensitive to the strength and content of incoming evidence that they create a threat to self-esteem. This threat then has to be compensated for by introducing motivational forces that push cognition in a direction that preserves a stable, positive self-concept. Rationalization suggests that the function of cognition is not exclusively rational, but that doesn’t undermine the existence or importance of rational belief updating.