Performance on Indirect Measures of Race Evaluation Predicts Amygdala Activation

You think you're not racist. You've considered it, maybe filled out a survey, and your conscious mind agrees with your self-assessment. So why is a structure buried deep in your brain firing differently when it sees a Black face versus a White one? That’s not a rhetorical provocation — it’s a precise empirical question. A team led by Elizabeth Phelps placed White American subjects into a functional magnetic resonance imaging scanner and found that the answer lies in the amygdala, correlating with something your conscious mind can't access, or at least won't report. To understand why this matters, you need to grasp the gap the researchers were trying to measure. For decades, surveys of racial attitudes in the United States have shown declining explicit prejudice. People report more positive views toward Black Americans than they did in the 1950s or 1960s. But psychologists have noticed something uncomfortable: indirect measures, the types that don't ask people to state their views directly, kept revealing something different. These implicit measures — which just means they operate outside of conscious awareness or control — consistently showed negative associations with Black social groups, even in individuals who sincerely reported no bias. The question Phelps and colleagues were asking is whether that gap between what we say and what we implicitly feel shows up in the brain itself. To answer that, the team used three different measurement tools in combination, which makes this study methodologically interesting. The first was the Implicit Association Test, or IAT. This is a reaction-time task where you sort words and faces into categories. The logic is that when two concepts are strongly associated in your mind, you're faster at grouping them together. The IAT captures associations you may not even know you have — it measures the speed of mental pairing, not your stated opinion. The second measure was potentiated startle, specifically the eye-blink response. When you're in a negative emotional state, your startle reflex is amplified. By measuring the twitch of the orbicularis oculi muscle, the one around your eye, using electromyography while subjects viewed Black and White faces, researchers could get a physiological read on negative emotional arousal that bypassed self-report entirely. The third tool was the fMRI scanner, focused specifically on the amygdala, a subcortical structure — meaning it sits below the cortex, deeper in the brain — known to play a central role in emotional learning and evaluation. The experimental setup for Experiment 1 was straightforward in design, even if it was technically demanding. White American subjects viewed photographs of unfamiliar Black and White male faces. The photos came from college yearbooks, filtered down to grayscale images showing neutral expressions and non-distinctive clothing — a careful attempt to strip out confounding variables. Fourteen subjects were ultimately analyzed after excluding twenty due to excessive head motion in the scanner. They also completed the IAT and a self-report scale called the Modern Racism Scale, which asked directly about racial attitudes. About a week later, the same subjects returned for the startle measurement session. Now, the results. The amygdala activated more strongly when subjects viewed Black faces compared to White faces. But here's where it gets sharp: that differential activation — Black versus White — correlated with both the IAT score and the potentiated startle response. The more implicit bias a person displayed on those indirect measures, the stronger their amygdala response to Black faces. Critically, the Modern Racism Scale — the direct, conscious self-report — showed no such correlation. The brain's response tracked the unconscious measures, not the stated ones. The team also ran whole-brain analyses using Statistical Parametric Mapping, a standard fMRI analysis package, and found event-related activation patterns consistent with the area of interest focused on the amygdala. Composite maps of the correlations between IAT scores and brain activation confirmed that the relationship was specific and anatomically meaningful. Both the left and right amygdala showed associations with the behavioral measures of implicit bias. The convergence across three independent measurement modalities — reaction time, eye-blink physiology, and blood-oxygen-level-dependent brain signal — gives the finding its weight. Any one of those channels alone could be questioned. All three pointing in the same direction is harder to dismiss. But the researchers didn't stop there, and this is where the study gets genuinely illuminating. In Experiment 2, they ran the same basic paradigm with a crucial change: instead of unfamiliar faces, subjects viewed faces of familiar, positively regarded Black and White public figures. And the pattern evaporated. No differential amygdala activation. No correlation with implicit measures. Nothing. Think about what that means. If the amygdala response in Experiment 1 were simply a response to racial category — something hard-wired, automatic, and fixed — it should have appeared regardless of whether the faces were familiar or not. It didn't. The response was specific to unfamiliar faces, to strangers whose only available social information was their group membership. When individual experiences with specific people replaced that group-level default, the amygdala signal changed. Familiarity, and the positive regard that came with it, rewrote the response. This is the interpretive core of the paper. Phelps and colleagues argue that the amygdala, because of its role in emotional learning, encodes culturally transmitted evaluations. Not innate threat responses, but learned ones — associations absorbed from a cultural environment in which Black and White social groups are evaluated differently. The amygdala is picking up a signal that culture put there. Individual experience with particular people can override that cultural default, as Experiment 2 shows directly. The discussion Phelps and colleagues offer is careful and worth reflecting on. They note that the stimuli in Experiment 1 were presented supraliminally — above the threshold of conscious perception. Subjects could see the faces clearly. The amygdala response wasn't driven by subliminal processing. Yet it still dissociated from conscious attitude measures. That dissociation matters because it suggests the brain holds an evaluative signal that conscious self-report doesn't capture — either because that signal operates too quickly to be introspected, or because the social desirability of admitting bias suppresses the conscious response, or both. The study can't fully adjudicate between those possibilities, and the authors acknowledge that. There are real limitations here. The sample was exclusively White Americans — fourteen subjects in the fMRI experiment. That’s a small number, and the findings can't be generalized beyond that specific cultural context without further work. The study also examined only one direction of the racial comparison, and the authors are explicit that findings from White subjects viewing Black faces don't automatically transfer to other group combinations. The cultural specificity is a feature of the argument, not a bug — the point is that these responses are culturally shaped — but it also bounds what the data can tell you. What the data can tell you is this: the brain's honest answer and the mind's stated answer are not always the same thing. The amygdala responds to cultural evaluations that we may not be able to articulate, or choose to articulate. The IAT and the startle response capture something real about how the brain processes social group membership, and that something shows up in a subcortical structure that operates faster than conscious thought and encodes learning that accumulates over a lifetime of cultural exposure. When Phelps and colleagues showed that individual positive experience with specific people can quiet that signal, they also showed that it isn't fate. The response is learned. And if it's learned, it can, at least in principle, be modified. That’s not a guarantee — the study doesn't test interventions, and it doesn't follow people over time. But it’s a more tractable picture than one in which bias is simply wired in. The amygdala doesn't know what you believe about yourself. It knows what your culture taught it, and it learned quietly, the way all the most consequential learning does. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.