Taking the Reading the Mind in the Eyes test

The “Reading the Mind in the Eyes” is meant to test Theory of Mind (ToM) or the ability to recognize and understand another person’s mental state. It’s supposed to be a more advanced test than “Fear, Anger, Joy“, which tests simple emotion recognition.

The original 1997 version of Reading the Mind in the Eyes consisted of a set of 25 photos showing the area around the eyes and a choice of two possible mental states for each photo.

However, the limited number of items on the test combined with the choice of only two responses resulted in a test that wasn’t very reliable. Parents of autistic children were scoring as far below the controls as the AS/HFA group was. Additionally, the original version of the test included some expressions for basic emotions (happiness, sadness) which were considered too easily recognizable and not a true test of ToM.

The revised version of Reading the Mind in the Eyes contains 36 items with 4 answer choices for each item, increasing the possible range of scores along with the difficulty level. It also contains a balance of male and female photos, a choice between more closely related mental states (i.e. not a choice between opposites like sympathetic/unsympathetic), and is composed entirely of photos representing complex mental states.

In the original study to validate the test, the AS/HFA group scored a mean of 21.9 while the control had a mean of 26.2. However, the AS/HFA group had only 15 participants versus 239 controls. A sample size of 15 is small, especially for study in which participants only have to complete two questionnaires (the AQ and Reading the Mind the Mind in the Eyes. I’m curious why the researchers didn’t make an effort to obtain a larger AS/HFA sample when they had the resources to administer the test to so many controls.

Pros and Cons of the Reading the Mind in the Eyes test


  • Tests recognition of complex mental states
  • Balanced presentation of male and female expressions
  • Offers subtly similar answer options to increase difficulty level
  • Self-scoring
  • Provides a list of items that were answered incorrectly (with the correct answers)


  • Validation study relied on a small sample size
  • Sets up artificial constraints not present in real life (limited choice of options, time to study “frozen” expressions)
  • Allows for unlimited time to answer each item

Taking the Test

You can take the test here. It’s all on one page. Just look at each set of eyes and then choose which of the four options best describes the state of mind that the pair of eyes conveys. Ideally, you should make your choice as quickly as possible.

It took me a little over 5 minutes to complete the test. I feel like I spent too much time on a few of the photos. For an idea of how unintuitive my process is when I take this kind of test, at one point I found myself unable to decide if a particular expression was content or defiant. These are very different mental states, but I ended up guessing (correctly!) because I couldn’t conclusively pick one over the other.

Once you’ve selected an answer for all 36 items, click the “get score” button and your score will be displayed at the top of the page.

Scoring the Test

Your score is a measure of how many out of the 36 items you answer correctly. You’ll also get a list of which answers you missed and a short summary of where your score fits in the distribution (below average, average or above average).

Here is my scoring information:

Your score: 31

A typical score is in the range 22-30. If you scored over 30, you are very accurate at decoding a person’s facial expressions around their eyes. A score under 22 indicates you find this quite difficult.

The correct answers for the ones you missed are: [I added in my answers in brackets so you can laugh at how wildly off some of them are]

    • 17: doubtful [I chose affectionate – this could be a serious gaffe in a social situation!]
    • 18: decisive [I chose bored]
    • 19: tentative [I chose grateful]
    • 28: interested [I chose affectionate – not that far off]
    • 35: nervous [I chose contemplative]

Like the “Fear, Anger, Joy” test, I scored slightly above average. I’m starting to question how much these tests say about a person’s ability to read facial expressions in “live” social contexts.

When I’m taking a test like this, there are two artificial constraints:

  1.  I’m forcing myself to intensely focus on and study each facial expression.
  2.  I’m given limited options to choose from.

Based on the availability of 4 choices, random guessing alone would result in, on average, 13 correct answers. If you look at the options for each expression, at least one and often two are obviously incorrect (to me, and that may just be me). One of my primary test taking strategies is process of elimination and my approach to this test was no different. If I can eliminate one or two options, my odds of guessing correctly go up significantly.

The artificial nature of the test seems to reduce its value in identifying problems with ToM. When I’m interacting with another person, I’m usually too preoccupied with trying to follow the conversation to spend much time “studying” the other person’s constantly changing expressions.

Often when I’m concentrating on a conversation, I’ll look away from the other person’s face because I find it easier to process information that way. You can’t gather a lot of facial expression data when you’re staring out the window. And, most importantly, there are no prompts. The other person’s expression could be saying literally anything and I have no helpful cue words to narrow that down for me.

Can you guess what these pair of eyes are saying without any cue words to help you? Give it a shot in the comments if you like.
Can you guess what this pair of eyes is saying without any cue words to help you? Give it a shot in the comments if you like.

Then there is the fact that recognizing an expression is one thing; attributing causation is another thing entirely. Facial expressions are supposed to provide the clues that allow us to understand what another person is experiencing (the content of their mental state). Recognizing an expression of anticipation is the first step; deducing what the other person is anticipating should logically follow. Together these make up the concept of Theory of Mind.

To say that the Reading the Mind in the Eyes test is a measure of Theory of Mind is only partially true, especially for those of us on the spectrum. The second step of the process–understanding the content of the other person’s mental state–is where I often go wrong.

The Bottom Line

This is an interesting test of static facial expression reading. It’s value as a test of Theory of Mind is less certain.

40 thoughts on “Taking the Reading the Mind in the Eyes test”

  1. So this guy could be amused, particularly if I just look at his left eye (the one on the right facing me) but he could also be angry. Maybe he’s surprised, but then again he could be asking a question and this is how he looks while waiting for an answer… In the end, I looked at him and thought, I have no idea what he is feeling, and without words or asking him directly, it’s impossible to know! There. And I’m neurotypical – ish. 🙂

    1. Alas, none of what you guessed is quite right. If you hover your mouse over the image, you’ll see that I named it with the correct answer. My first thought when I saw this image was that he looks like he’s trying to read my mind. 🙂

      These tests feel awfully contrived. When the list of possible answers is taken away, it’s impossible to guess the correct answer. (Of course, now someone probably will.)

      1. *Laughing* Okay, so never would have guessed that, not in a million years. That is hilarious. Also I have to admit, I found it really creepy to look at him and try to figure it out. It reminded me of an Autistic friend of mine who told me that looking at another person was physically very distressing as she felt she could see into their soul and what she often saw was extremely painful! That’s how I felt looking at his eyes. It was painful! And now, kind of funny too!

        1. I find eye contact really difficult in most situations. It quickly becomes far too intense and I have a physical need to look away, much like a fight or flight response (though only the flight part!).

  2. My score was a 25. I don’t understand how these tests can be very accurate because it’s very rare to have to gauge emotion from eyes only – we usually have the whole face and body to tell us information. Also, and this might have nothing to do with how people score, but I think it’s strange that all the eyes they showed seem to belong to Caucasians. And I felt that the women’s eyes seemed to be inordinately bedroomy! The pair of eyes you posted struck me as leering – but I see now that’s not right.

    1. I couldn’t find any specific reason for the test focusing only on the area around the eyes. Maybe it has to do with the fact that aspies tend to avoid eye contact? That’s a good point about the majority of the faces being Caucasian. There was no mention of that in the research paper, but it is a weakness. I also found that some of the states of mind seemed to be gender stereotyped, with more of the female faces being things like flirtatious and more of the males being serious-type expressions.

  3. I couldn’t even finish the test. I found the choices completely unsatisfactory. And I was completely off on the picture that you included in your post. “Most people surprise themselves with how well they do.” Um, no. I’m surprised by how anxious the whole test made me (an NT) feel.

  4. lol I took this test during my “researching” stage. I was determined to try really hard and get a realistic idea of where i was at. It brought me out in a genuine sweat of fear. I was convinced that at the end it was gonna tell me it was all a big joke and of course no-one could possibly know…umm yeah it didnt and i sucked lol. I didnt even know that emotions like “flirtatious” showed in the eyes..which explains a few problems in itself! I then had to go and examine my own eyes in the mirror cos i realised i didnt even know what colour they were!!

    1. It’s so funny that you thought the test was a trick until the end. I wish! It’s rather impossible to figure out emotions from those static photos.

      And the eye color thing! Until I was in my thirties, I thought my eyes were brown. Then one day my husband mentioned how much he liked it that my eyes get very green when I’m in a good mood. Upon closer inspection I saw that my eyes are green with gold flecks around the iris. That was quite a revelation. I still haven’t changed my drivers license info . . .

  5. My email to Simon Baron Cohen
    “I am sure others have asked you this, but unfortunately I have not been able to find the answer online, so here it comes again. Where did you (or those you worked with) get the pictures for the Reading the mind in eyes test? Thank you for your time”
    the answer I got yesterday
    “Dear Kent. Sorry for the delay in replying. From fashion and other magazines. Best wishes, Simon Bc”
    so there is no way of saying if the mental states really are those they say they are, but I guess the test still works as a normality test if they make sure that the “right” answer is the ones that most vote on : /

    1. Thank you for sharing your email correspondence. It’s interesting that the images are taken from media rather than developed specifically for the test by actors intentionally modeling specific mental states.

  6. I scored 24, although that has been known to fluctuate, due to a lot of factors (mood, resolution of the device I’m using to take the test (whether it is a large screen monitor connected to my Mac or my iPad2)

  7. You scored better than me. I’ve never been diagnosed, but my boyfriend has and we are shockingly similar. I scored 17.
    I took another test on Cambridge’s website that asks situational questions and Aspie is 32 or higher and I scored 40. This makes me feel like crying, but it explains a lot. I think I’m going to have to see a doctor. I also have synesthesia of the four core processors; which has been diagnosed. I also have Tourette’s complex motor tics, but I manage it so well most people don’t even know. They laugh when I tell them because they think I’m joking.
    All of this makes so much sense, but I also feel incredibly upset now. Thank you for having this blog.

    1. It can take some time to adjust to the idea so take it slow. Perhaps you can talk to your boyfriend, since he’s diagnosed? He might be able to help you process what you’re feeling.

  8. hi i am trying to access the test but it seems like the link is broken somehow, can someone please send me the link to the test or help my find it please. thankyou

  9. I scored only 8 even though I spent lots of time studying facial expressions of those around me, it SO frustrating x_x
    I’m clueless with the picture you’ve posted. I’d say “mocking” 😀 but… 😀

  10. Well now I’m really confused. I scored a 40 on the AQ test and everything began to make sense… till I took this one and scored 34/36. What the heck does that make me?!

  11. I got 28/36 which puzzled me as it obvious to anyone who knows me well that I’m likely to be on the spectrum (diagnosis in about 2 months’ time).

    My theory is that I can observe people’s moods on their faces so long as I am not under time pressure, listening to them or trying to speak to them. Even then it feels more like trying to solve a puzzle rather than empathising. I can’t register people’s expressions when I’m trying to speak or listen.

    I also read that people from turbulent backgrounds tend to be better than average on this test because of the need to know when to stay out of the way of volatile people when they are likely to be aggressive. That could also partly explain the result I got. People with PTSD and BPD score particularly highly in this test as a result.

    1. That’s a really good point basilica. My background wasn’t nearly as turbulent as some but I was bullied from nursery school onwards and had to learn to become invisible at home when explosions were on the cards. A good education in reading the eyes.

      1. That’s pretty much what I went through. Bullied all the way through school – some of it either life-threatening or sexual abuse. Most of my relatives are very direct, but one had a propensity towards not saying what she meant and exploding out of nowhere weeks after I’d inadvertently offended her. I learned to read her so well we could at one point communicate by facial gestures without speaking at all. She’s mellowed over time and is starting to realise that if she isn’t direct with me I won’t understand. The only trouble is I got so so used to walking on eggshells around her as a child that I keep on falling in love with people who have similarly volatile personality traits to her which never works out well. So I’ve sworn off romantic relationships until I’ve had some serious therapy. 🙂

  12. The comments here are interesting. I found this test a while ago and had to research to make sure it wasn’t a joke. I don’t have a clue to any of the answers and I don’t understand how anyone could. So I find it very surprising that people can get a high score on this that’s not just luck.

  13. “The artificial nature of the test seems to reduce its value in identifying problems with ToM. When I’m interacting with another person, I’m usually too preoccupied with trying to follow the conversation to spend much time “studying” the other person’s constantly changing expressions.”

    That an indicator of a person having problems with ToM, though!

    NTs in coversation are not “too preoccupied with trying to follow the conversation”. For them, the “studying the other person’s constantly changing expressions” is even more important, and they don’t even have to spend much effort, it comes naturally.

    “Often when I’m concentrating on a conversation, I’ll look away from the other person’s face because I find it easier to process information that way. You can’t gather a lot of facial expression data when you’re staring out the window.”

    Well, that would just mean the test would do its job though: most aspies will score low, as well as show behavior like the above in conversations, which NT’s wont. So, an aspie’s “looking away from the others persons face to process information” is not unrelated to their failing the test (again, failing as a group, on average. Different aspie individuals might still score well).

Share Your Thoughts

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s