Reading the Mind in the Eyes test

The Reading the Mind in the Eyes Test (RMET or Eyes Test for short) has been widely used to measure theory of mind—the ability to recognize and understand another person’s mental state—or social intelligence.

Take the test here:

This test is free, anonymous, and requires
no email sign-up for the results

Who the test is designed for

• Often used in research involving autistic adults without intellectual disability
• Adults with IQ in the normal range (IQ >=80).

Versions & translations

• The RMET has been translated into multiple languages.

Instructions

The Reading the Mind in the Eyes Test consists of 36 photographs, each giving you four choices.

For each set of eyes, choose which word best describes what the person in the picture is thinking or feeling. You may feel that more than one word is applicable, but please choose just one word, the word which you consider to be most suitable. Before making your choice, make sure that you have read all four.

1. 1. Once you have read the instructions, complete the practice item. Do not worry about what the answer is; the point of the practice item is for you to understand what to do.
   2. Now you are ready to begin the test.
   3. Please do not review the scoring until the test is complete, as it will affect your scores.

Scoring

The total score possible is 36.

Each item is scored as correct or incorrect. Total your correct items

• Autistic average & range: 24.9 ± 0.7 (23–30)^[1]Structural Correlates of Reading the Mind in the Eyes in Autism Spectrum Disorder (Sato et al., 2017)
• Neurotypical average & range: 27.3 ± 0.5 (18–29)^[2]Structural Correlates of Reading the Mind in the Eyes in Autism Spectrum Disorder (Sato et al., 2017)

Performance

In the table below, you can see the performance on the Revised Eyes Test and the AQ, according to a paper by Baron-Cohen et al. (2001).^[3]The ‘Reading the mind in the eyes’ test revised version: A study with normal adults, and adults with Asperger Syndrome or High-Functioning autism (Baron-Cohen et al., 2001)

Research by Sato et al. (2017) shows a performance of 24.9 ± 0.7 for autistics, and 27.3 ± 0.5 for neurotypicals—with typical ranges of 18–29 vs 23–30.^[4]Structural Correlates of Reading the Mind in the Eyes in Autism Spectrum Disorder (Sato et al., 2017) Note that the two ranges overlap, group differences on average have been observed in some studies.

Validity

The RMET is one measure sometimes used in autism research and assessment contexts to explore emotion recognition and social perception. The Eyes test is reliable and stable over 1 year.^[5]The “Reading the Mind in the Eyes” test: Systematic review of psychometric properties and a validation study in Italy (Vellante et al., 2013)

2001 update

Note that the original RMET—though a valid measure—had some shortcomings in that the items had a forced binary choice which left too much up to chance, and it was deemed that a longer test (36 vs. 25 items) would yield more statistically significant results that can better differentiate between autistics, Broader Autism Phenotype (BAP), and non-autistics.^[6]The “Reading the Mind in the Eyes” Test Revised Version: A Study with Normal Adults, and Adults with Asperger Syndrome or High-functioning Autism (Baron-Cohen et al., 2003) The revised test from 2001 rectified these issues. Consequent reports on the validity of the RMET are based on the revised test.

Not a ToM measure?

However, it should be noted that while a meta-analytic investigation from 2023 found that the RMET has an acceptable internal consistency (α = .73), it’s more strongly related with emotion perception (r = .33, ρ = .48) compared to alternative theory of mind measures (r = .29, ρ = .39). Furthermore, the RMET was weakly to moderately related with:^[7]Sty in the Mind’s Eye: A Meta-Analytic Investigation of the Nomological Network and Internal Consistency of the “Reading the Mind in the Eyes” Test (Kittel, Olderbak, & Wilhelm, 2021)

• Vocabulary (r = .25, ρ = .32)
• Cognitive empathy (r = .14, ρ = .20)
• Affective empathy (r = .13, ρ = .19)

The authors conclude that the RMET is more of an emotion perception measure than a measure of theory of mind per se.

Discussion

Dr. Natalie Engelbrecht:

I scored 35/36, which is above the average ranges reported in some studies.

• Happiness: Contracts the orbicularis oculi around your eyes, making your eyes smaller and contracting the muscles under the eyes, and at the sides of the eyes.
• Anger: Tightened eyelids, eyebrows lowered and drawn together. Eyes open wide and staring intensely.
• Sadness: Inner corners of eyebrows pulled down and together. Upper eyelids drooping and looking down.
• Surprise: Eyebrows raised, but not together. Upper eyelids raised, lower eyelids neutral.
• Fear: Eyebrows raised and pulled together. Raised upper lids. Tensed lower lids.
• Disgust: Lowered eyebrows, wrinkling at the bridge of the nose.

I do dislike that the test uses only caucasian faces. In addition, there seems to be gender bias in the emotions in that the male faces are more often angry faces, while the female faces are more often flirting.

Kendall:

When I did the eyes test 8 months ago, I scored 19/36. In a recent retaking, my score was 27/36. The 2nd took 12 minutes—I didn’t time the 1st. Using the process of elimination and sometimes simply guessing was my approach on both occasions. I believe my higher 2nd score is accounted for by remembering some of my initial errors.

Were word prompts not provided, 80% of the photos would be an out-and-out mystery to me. Thus, I was skeptical of the test’s workings and results. However, my score was consistent with autism. With many photos, I was close, but others were way off, for example, answering shy for 26, which is hostile. Or 29, where I answered irritated—the answer is reflective. The test is quick, straightforward, and easy to score. One of its more valuable aspects is pointing out the emotions I may need to take extra care in interpreting.

Eva:

I first did the test some years ago. I didn’t save my results, but I remember getting just a few items wrong (probably 3 or so); and what I remember from that time is that I had a tendency to evaluate facial expressions as more negative than they were due to my alexithymia. This tendency generally applies to people with some kind of trauma history; especially ambiguous facial expressions will be interpreted as more negative and threatening.

I retook the test again just now at the end of October 2024, and my first surprise was that while I was quite convinced I got all items correct, I actually got 6 items wrong! I did the test in about 3.5 minutes, so perhaps I could have gotten some of the items correct if I had given them more consideration. But still, I think that’s significant. I actually retook the test to assess the scoring module, so after my initial honest attempt, I tried to get a perfect score to see if the test would add up to 36. To my surprise, I still got three of the items wrong, and I still got a few of them wrong on a third attempt. It seems two of the items in particular are quite ambiguous, but that might just be me.

But what I’ve noticed in all but two of the items I got incorrect is that I evaluated the factual expressions to be more positive than they were!

• Item #7 I judged to be friendly, while they actually felt uneasy.
• Item #11 I judged to be amused, while they were actually regretful.
• Item #17 I judged to be affectionate, while they were actually doubtful.
• Item #23 I judged to be curious, while they were actually being defiant.
• Item #25 I deemed to be incredulous, while they were actually being interested.
• Item #29 I deemed to be impatient, while they were actually reflective.

So it seems I veered into the other direction, and now evaluate facial expressions to be more positive and innocuous than they are. Did I get more naive and trusting? I don’t know. I’m not sure what this means for me. But what I do know is that autistic people often have more challenges with interpreting facial expressions correctly, which may be in part because we spend more time looking at the mouth compared to neurotypicals. I wonder if I would have gotten more of the items correct if I could have assessed the whole face instead of just the eyes…

One final thing I want to add is that I’m now close to 3 years on HRT, which makes me wonder if there could be a hormonal influence on how I now read facial expressions compared to some years ago.

Reading the Mind in the Eyes Test

Answers

The table below shows the correct answers in bold and denoted with a black star (★). For easy comparison, we suggest opening this post in another window, so you can compare the table with the images of the test.

RMET answers

If you are looking for an autism/AuDHD assessment,
our (autistic) assessment team can offer help!

You can find more information here:

Autism/AuDHD assessments