>> Crossing the Levels of Analysis: Implicit Attitudes

	Interview with Elizabeth Phelps, New York University From Studying The Mind, VHS © 2003, W. W. Norton

How have cognitive neuroscientists and social psychologists collaborated to study implicit attitudes?

A lot of what we've done in studying the amygdala in humans is look at emotional responses to different types of situations, and particularly learned emotional responses. And what we found with patients with amygdala damage is that they show this disassociation in the physiological expression of an evaluative response. For instance, they'll learn that a blue square is paired with a shock, and while won't show a physiological expression of that learned fear, they can tell you that, "Oh, that was the one that was paired with the shock, you know, that's a negative thing." And the same is true for pictures. So they're very good at giving you explicit evaluations of stimuli in the world, but they're not so good at implicitly showing you that response within a physiological sense.

This disassociation between explicit evaluation (rating conscious knowledge of the situation) and a more implicit, subtle way of showing an evaluative response is something that's been observed by social psychologists as well when they look at responses to social groups. People might say, for instance, that they don't have prejudicial attitudes towards Black Americans, yet at the same time, on some implicit tasks, they will still show biases. So it seemed to us there was a link between the social psychological literature that talks about the difference between implicit and explicit attitudes and this work on the neurosystems of emotional learning—and especially the expression of those learned responses.

How did you test this idea?

I did a study with a social psychologist, Mahzarin Banaji, in which we showed pictures of Black and White male faces with neutral expressions to White American subjects. And at the time we did the study, the subjects didn't know that this was a study about race. Their task was simply to press one button if they had seen the face before, and another if they had not.

Later we did three things. First we gave subjects a scale that measured their race attitudes. This would tell us explicitly what they thought. We then gave them something called the Implicit Association Test (IAT), which has been used a lot by my collaborator and by other social psychologists. And the way this test works is that subjects see, one at a time, a picture of a Black face or a White face, or a word that has a "good" meaning or a "bad" meaning. For a White face or a good word, subjects were instructed to push a button on the right, and for a Black face or a bad word, they were to push a button on the left. Then after a while we switched the rules, pairing the left button with a Black face or a good word, and the right button with a White face or a bad word. And then we switched the rules yet again. And, in general, Mahzarin Benaji has found that it takes White Americans longer to respond when they have to pair a Black face with a good word as opposed to a bad word. In other words, there is some sort of inhibition or interference effect occuring. So this is a measure of sort of your implicit attitudes.

The third thing we did was to give subjects an eye blink startle test, our second implicit measure. If you're walking down the street in the middle of the day and you hear a loud noise, you startle. If you're walking down the street at night and it's dark out and you hear a loud noise, you startle even more. So, startle's a reflex that we all have, but it's a reflex that gets potentiated or modulated by our emotional state. When we're in a context that's somewhat negative, we will show an enhanced startle response. We can measure this in the laboratory by measuring your eye blink. You blink when you're startled, and when you blink harder, you've been startled more. So we just measure the muscles around your eyes. We showed people the same pictures of Black and White faces, played loud noises through head phones, and then measured the eye blink response. What we found as a whole is that White Americans showed more startle to Black faces than White, but that it was not a really strong effect. We then compared our brain imaging data (looking at the amygdala), and our three behavioral measures (the measure of explicit race attitudes and the two implicit measures, the IAT and eye blink startle). And while we found a lot of variability among the White American subjects, most of them—though not all—showed greater amygdala activation to the Black faces than the White faces. But most interesting was the relationship between our amygdala response and our behavioral measures. Again, there was variability among our subjects in all of these measures, but the subjects that showed a greater amygdala response with Black faces also showed a greater bias response on both the IAT test and a greater bias on the eye blink startle. There was no relationship between amygdala response and the explicit race attitudes.