PsychOS

The Reading the Mind in the Eyes Test: How Baron-Cohen Measured Cognitive Empathy

Cover every part of a human face except a two-inch strip around the eyes, and something strange happens: most people can still accurately guess whether that person is feeling arrogant, terrified, f...

Explore PsychOS

PsychOS combines psychology education with structured self-reflection tools. Explore assessments, articles, experiments, and related resources while treating results as educational signals rather than diagnosis.

THE HYPOTHESIS

By the mid-1990s, psychologist Simon Baron-Cohen and his colleagues at Cambridge faced a specific measurement problem. Existing tools for studying "theory of mind" — the capacity to infer what another person is thinking — relied heavily on tasks designed for young children, such as the classic false-belief test, where a child predicts that another person will hold a mistaken belief they don't personally share. These tasks were useful for studying basic theory-of-mind development, but they were far too simple to reveal anything meaningful about adults, including high-functioning autistic and Asperger's adults who could pass basic false-belief tasks without difficulty, yet still reported profound, lifelong struggles reading other people in real social situations.

Baron-Cohen hypothesized that a more sensitive instrument was needed — one capable of detecting subtle deficits in adult social cognition that simpler, childhood-designed tasks were structurally unable to catch. He also suspected that a specific, narrow channel of social information — the eye region alone, stripped of every other facial and contextual cue — might be unusually revealing, since humans rely heavily on the eyes specifically, more than any other single facial feature, to judge another person's mental and emotional state.

THE METHODOLOGY

Baron-Cohen, along with Sally Wheelwright, Jacqueline Hill, Yogini Raste, and Ian Plumb, developed what became known as the Reading the Mind in the Eyes Test , first published in 1997 and revised into its most widely used form in 2001. The test presented participants with a series of photographs cropped tightly around the eye region only — no mouth, no eyebrow movement beyond what naturally appeared, no broader facial context, and no body language.

For each photograph, participants were given four words describing possible mental or emotional states — for example, "jealous," "panicked," "arrogant," or "hateful" — and asked to select the single word that best matched what the person in the photograph was thinking or feeling. Critically, the researchers deliberately selected complex, subtle mental states rather than the small set of basic emotions, such as happiness or anger, that most existing emotion-recognition research had relied on. This design choice mattered enormously: basic emotion recognition is a relatively coarse skill that remains largely intact across most clinical populations, while recognition of complex, blended, or subtle mental states requires a more sophisticated inferential process — closer to actual cognitive empathy as it operates in everyday social life, and closer to the process examined in Cognitive Empathy vs. Affective Empathy: Why Understanding Someone Isn't the Same as Feeling With Them .

The test was administered to multiple groups: adults with high-functioning autism or Asperger syndrome, a matched control group of neurotypical adults, and, in later research, general population samples used to establish typical scoring ranges.

STEP-BY-STEP EXECUTION

Participants worked through 36 photographs (the original 1997 version used 25; the revised 2001 version, now the standard, uses 36), each requiring a forced choice between four emotionally or mentally complex descriptor words. There was no time limit in most administrations, meaning the test measured accuracy of inference rather than processing speed — a deliberate design decision, since the researchers were specifically interested in whether the capacity for accurate mental-state inference existed at all, not merely how quickly it could be deployed.

The results were striking. Neurotypical control participants scored significantly higher, on average, than the autism and Asperger syndrome group, despite both groups performing similarly on basic emotion-recognition tasks and standard intelligence measures. This confirmed that the deficit being measured was specific and narrow: not a general intelligence gap, not a basic emotion-recognition failure, but a specific difficulty extracting complex mental-state information from a narrow, subtle channel of social cues — the eyes alone, without supporting context.

Equally important, scores within both groups varied continuously rather than falling into two cleanly separated clusters. Some individuals with autism scored within the typical range, and some neurotypical individuals scored notably low. This detail turned out to matter enormously for how the test would later be used and interpreted.

PSYCHOLOGICAL REVELATION

The Reading the Mind in the Eyes Test demonstrated something that simpler theory-of-mind tasks couldn't: that cognitive empathy is not a single pass/fail switch, but a continuously variable capacity that exists on a spectrum across the entire population, autistic and neurotypical alike. This directly supported the broader shift in psychology away from treating empathy as one unified trait, and toward recognizing it as a specific, measurable, and separable component of social cognition — distinct from emotional resonance, distinct from general intelligence, and distinct from moral character.

The test's design also carried an unintended but significant implication once later research applied it to other populations. When researchers administered the same instrument to individuals scoring high on measures of psychopathic traits, results showed something that directly challenged an intuitive assumption: many individuals with elevated psychopathic traits performed normally or even above average on the Reading the Mind in the Eyes Test, despite well-documented reductions in affective empathy and emotional concern for others. In other words, the capacity to accurately read what someone else is thinking or feeling from minimal cues — cognitive empathy — was often fully intact, or even unusually sharp, in exactly the population least likely to act on that information with any genuine care. This finding lends direct empirical weight to the pattern explored in The Dark Triad: Narcissism, Machiavellianism, and Psychopathy Explained — accurate social perception, entirely uncoupled from compassionate motivation, is not a contradiction. It's a documented, measurable psychological profile.

The test also had a second, quieter consequence: it gave psychology a validated way to distinguish between two very different explanations for "poor empathy" that had previously been lumped together under one vague label. A person could score low because they couldn't infer the mental state from available cues — a cognitive empathy deficit, common in certain autism presentations. Or a person could score entirely normally on inference, while still failing to act with care once the inference was made — a profile far more consistent with reduced affective empathy or antisocial traits. Before this instrument existed, both patterns were often described identically as "lacking empathy." After it, researchers had an actual tool to tell them apart.

See Your Own Cognitive Empathy Pattern More Clearly

Reading about how researchers measure the gap between understanding someone and feeling with them is different from knowing where your own accuracy and emotional response actually fall.

→ Take the Cognitive Empathy vs Affective Empathy Assessment

Free. No account required.

Want to see how your empathy pattern connects to your broader interpersonal style, including manipulation awareness and communication under pressure?

→ Explore the full PsychOS Assessment Lab

Frequently Asked Questions

What is the Reading the Mind in the Eyes Test?

It's a psychological instrument developed by Simon Baron-Cohen and colleagues that measures cognitive empathy by asking participants to identify complex mental or emotional states from photographs cropped to show only the eye region, with no other facial or contextual cues available.

What did the original study find?

Adults with high-functioning autism or Asperger syndrome scored significantly lower on average than neurotypical adults, despite performing similarly on basic emotion recognition and general intelligence measures, indicating a specific and narrow deficit in reading complex mental states from subtle cues.

Does a low score mean someone lacks empathy entirely?

No. The test specifically measures cognitive empathy — the ability to accurately infer what someone else is thinking or feeling. It does not measure affective empathy, the capacity to emotionally resonate with that state once it's identified, which is a separate system entirely.

Can people with psychopathic traits score well on this test?

Yes, and this is one of the test's most significant findings. Individuals scoring high on psychopathic traits often perform normally or above average on this test, despite documented reductions in affective empathy, demonstrating that accurate social read and genuine emotional concern are separable psychological systems.

Is this test used clinically to diagnose autism?

It is primarily a research instrument used to study social cognition and has informational and research value, but it functions as one data point among many rather than a standalone diagnostic tool, and clinical diagnosis involves broader criteria and professional evaluation.

Related Psychology