The Mind Is Stranger Than Fiction

A warning: I have no idea how to explain the research results that I'm about to describe, and neither John Bargh nor Mark Chen or Lara Burrows can offer a compelling explanation either. It may very well be that the results are due to the entirely mundane effects of increased and decreased arousal that have more to do with the nature of the tasks than the content of the primes, but it's impossible to tell. The "experiments" don't really test any explanations, they just demonstrate effects, as is the social psychologist's way. Still, the "experiments"are pretty cool, and the results are really weird, and sexy (they've made Bargh a minor celebrity), so I'm going to talk about them anyway.

We've known for some time that attitudes, including those we don't really want to have (e.g., racist, sexist, or homophobic attitudes), can be activated automatically (i.e., without intention), and often unconsciously, and that they can influence our subsequent thoughts and behavior. John Bargh and his colleagues have argued that this is only part of the story. They believe that specific "social-behavioral responses" can be automatically (and unconsciously) activated by situational factors in the way that attitudes can. In their 1996 paper, they refer to James' "principle of ideomotor action" (by the way, if all philosophy is merely a footnote to Plato, then all psychology is merely a footnote to James), and give the following quote from The Principles of Psychology:

We may lay it down for certain that every representation of a movement awakens in some degree the actual movement which is its object.

In other words, the mental representation of a movement increases the likelihood of the performing of that movement. Since movements, or behaviors, involve mental representations just as attitudes do, then it stands to reason that representing a behavior increases the likelihood of performing it. The thought primes the action. Furthermore, since thoughts can be primed by external information, behaviors can too. The most common examples of such an automatic stimulus-behavior link, these days, refer to aggression (though people sometimes draw really strange conclusions from that research), but you get the idea. In fact, because attitudes and other conceptual representations are closely associated with behavior representations, activating one can activate the other, and thus prime behaviors.

Thus, Bargh, Chen, and Burrows set out to demonstrate the existence of "automatic social behavior." They conducted three experiments, each targeting different behaviors. In the first experiment, they first gave participants a scrambled sentence test, which involves presenting five scrambled words and asking the participant to form a grammatically correct sentence out of four of them as quickly as possible. They developed three different lists of scrambled words, one of which primed the concept RUDE, another that primed POLITE, and a third that was neutral with respect to rudeness/politeness. The primes in these lists included adjectives, adverbs, or verbs that were associated with the concepts (e.g., brazen, aggressively, or disturb for RUDE, and considerate, patiently, and respect for POLITE). While the participant was completing the scrambled sentence test, the experimenter left and began talking to a confederate (an experimenter posing as another participant). When the participant finished, he or she came out of the room to look for the experimenter to receive instructions for the next task (as the experimenter had instructed). However, the participant always found the experimenter talking to the confederate. Bargh et al. then measured the time it took for the participant to interrupt the conversation between the experimenter and the confederate.

Guess what they found. Of the participants who did the RUDE version of the sentence test, more than 60% interrupted in under ten minutes (they cut it off at ten minutes -- can you imagine how frustrated some of those participants were after standing there for ten minutes?), whereas fewer than 20% of the POLITE-primed participants interrupted in that time. The neutral list participants were in between at around 40%. The RUDE participants also interrupted a full 3 minutes sooner than neutral participants, and almost 4 minutes sooner than the POLITE participants.

But it gets weirder. In experiment two, they looked at the behavioral effects of a particular stereotype, the "elderly stereotype." Once again, they gave participants the scrambled sentence test, but this time, one version containing words associated with the elderly (e.g., worried, old, lonely, and no I'm not making this up, Florida), and the other was age-neutral. The participant was then "partially debriefed" (i.e., lied to), and then told that he or she could go. The experimenter then timed the participants walk to the elevator down the hall, and wouldn't ya know it, the people who had been primed with the elderly words, thus activating the elderly stereotype (which apparently includes slowness) walked to the elevator more slowly. It took them a full second longer (8.28 seconds vs. 7.30 seconds for those who took the neutral test), on average, to get there.

Recognizing that there might be at least one alternative explanation for their findings (other than the automatic activation of a stereotype which then influences behavior), they conducted a second version of this experiment to make sure that the elderly word list wasn't depressing the participants, and that sadness was responsible for their slowness. They gave each participant a version of the Affect-Arousal Scale, and found that the elderly-primed participants were no more depressed than those in the neutral condition. Thus, sadness can't explain their results. Who knows what can?

But it gets even weirder. Brace yourselves for this one. In the third and final experiment, they presented participants with a "boring and tedious" task that took forever (each participant completed 130 trials). Before each trial, a picture of an African American male face, or a "young Caucasian male face" popped up on the screen subliminally. This was designed to activate racial stereotypes without the participants being aware of it. After the 130th trial, a message came up on the screen saying that there had been a problem in saving their data, and they would have to complete the experiment again. They videotaped the participants' reactions to this horrible news (if you've ever participated in one of those tedious psychology experiments, you will understand just how terrible that news is), while the experimenter rated the hostility of the reaction in first person. The video tapes were used so that two other independent raters could rate the hostility later on. The inter-rater reliability wasn't really very high, with the two independent raters agreeing about 64% of the time, but since they were using a 5-point scale, that's not surprising. They just averaged the ratings that they couldn't agree on after discussion (if one had given a 2, and another a 4, then they used 3). Once again, the stereotype seemed to automatically affect the participants behavior. Those who had seen the African American faces were rated as more than half a rating-point more hostile than those who had seen the white faces.

I wish I had something really smart to say to wrap this post up, summarize the findings, and at least attempt to explain them, but I don't. If these results show what Bargh and his colleagues think they do, then it is both very interesting, and very scary. The unconscious and automatic influence of pernicious racial, gender, or even age stereotypes on our behavior is not something pleasant to think about, and could ultimately have major social and political implications. However, no one really knows what's driving these experimental results, and to date, no one's really attempted to tease out the causes experimentally. All we have is these really freaky results from a 1996 paper, along with the other, often equally bizarre effects that Bargh and others have since demonstrated since then. We're no closer to knowing what's really going on than we were 9 years ago. I do know one thing, though. Reading entire social psychology papers makes me write more speculatively and less analytically. Is that yet another case of stereotypes automatically influencing behavior? (OK, that was a bit of low blow, but it was funny, right?)

POST SCRIPT: I'm not really trying to be mean to social psychologists. I just wrote a post the other day that favorbably cited the research of several social psychologists. As with all science, each piece of social psychological research should be evaluated individually. It should not be assumed that because some social psychologists don't actually test explanations, and just look for really sexy results intead, that all social psychologists do that.

POST POST SCRIPT: And no, I'm not just making these jabs because I'm anonymous. I have been saying these things for years, often quite loudly and within earshot of social psychologists. In fact, in the graduate course I took on social psychology, I spent an entire semester saying these things to a whole room of social psychologists. It's all in good fun, people, all in good fun.