Skip to content

“I was finding the test so irritating and boring that I just started to click through as fast as I could”

In this article, Oliver Sacks talks about his extreme difficulty in recognizing people (even close friends) and places (even extremely familiar locations such as his apartment and his office).

After reading this, I started to wonder if I have a very mild case of face-blindness. I’m very good at recognizing places, but I’m not good at faces. And I can’t really visualize faces at all. Like Sacks and some of his correspondents, I often have to do it by cheating, by recognizing certain landmarks that I can remember, thus coding the face linguistically rather than visually. (On the other hand, when thinking about mathematics or statistics, I’m very visual, as readers of this blog can attest.)

Anyway, in searching for the link to Sacks’s article, I came across the “Cambridge Face Memory Test.”

My reaction when taking this test was mostly irritation. I just found it annoying to stare at all these unadorned faces, and in my attempt to memorize them, I was trying to use tricks (keeping track of the skinny face, the scowling face, the jowly face, etc.). It wasn’t really clear to me whether I was supposed to try to use these sort of gimmicks or whether that was cheating, whether I was supposed to just look at the faces as if it were in a normal situation. The test seemed so artificial. But, really, I guess it’s no more artificial then tests in general (including the ones that I administer in my own classes!).

Here are my results form the test:

Out of 72 faces, you correctly identified 42. In other words, you got 58% correct.

On our previous version of this test, the average person with normal face recognition was able to recognize about 80% of the faces. If you correctly identified less than 65% of the faces, this may indicate face recognition difficulties.

Near the end, I gotta say that I was finding the test so irritating and boring that I just started to click through as fast as I could. On the other hand, my irritation was probably not “exogenous,” as the economists would say. If I were good at face recognition, maybe the test would’ve been mildly enjoyable.

P.S. Maybe the police should give this test to witnesses who are asked to identify a criminal suspect in a lineup. I certainly wouldn’t want somebody thrown into jail based on my eyewitness testimony! I’ve read a lot of criticisms of police lineups but only now have I realized that it might be possible to identify more or less accurate eyewitnesses. In fact, before reading Sacks’s article, I’d never thought of face recognition as a talent that some people have more of than others. I mean, sure, I knew that some people can recognize a lot of people, but I’d just thought of it as something that they did, not as a skill.

P.P.S. Just to be clear: I can recognize lot’s of people’s faces. I’m nothing like the way Sacks describes himself. But I felt that I have a little of this trait.

P.P.P.S. Sam Anthony writes in:

I was pleased to see you mention my boss Ken Nakayama’s prosopagnosia test on your blog. I’m sorry (but not too surprised) you found it so excruciating; the irritation inherent in being asked to repeatedly perform a nearly impossible task is an unfortunate side effect of many perception experiments. We’re currently working with a statistical researcher who is experienced with standardized testing to quantify the per-trial discriminative power and (hopefully) shorten the test, in part because the experience can be such an unpleasant one for the prosopagnosic.

If you’re curious enough about your own face recognition abilities that you want a more informative number, and you’re willing to subject yourself to the experience again, we have a much better-normed (but not shorter) version of the test (using real faces) online that I could point you to, but needless to say I understand where you’re coming from if you aren’t so interested in that.

My reply:

Thanks for the note. I imagine I would’ve taken the test more seriously had it been in a clinical setting rather than just something I did out of curiosity when writing a blog. So I wasn’t complaining about the test, just sharing my reaction with others. I think we tend when evaluating research to think of our measuring instruments from the perspective of the researcher rather than the perspective of the experimental subject, hence it seemed worth trying it out. In any case, I’m glad you’re working on improving the psychometric properties of your exam. I wish I could say we do as well with our own exams in the classes we teach!


  1. wcw says:

    I found that test far too easy and the faces really dull. I also went through as fast as I could click, to turn it into a bit of a challenge. I think I scored in the mid 90s. Recognition has always been exceptionally easy for me, though. I usually can't tell when twins are identical rather than fraternal, since the identical twin faces look quite different to me.

  2. DavidC says:

    Neat idea. Maybe for use by defense attorneys questioning eyewitnesses on the stand?

    I found myself thinking that this was testing my short-term memory rather than my facial recognition. But then again, maybe I have bad short-term memory for faces but not other things.

    One could put in questions unrelated to faces, to control for that, and time the questions in various ways to study:

    1) Ability to remember faces (as a function of time elapsed)
    2) Ability to remember other things (as a function of time elapsed)
    3) Ability to remember anything (as a function of how long you've been taking this annoying test)

  3. subdee says:

    This is really hard without haircuts.

    It became easier for me when I imagined personalities to go along with the faces, what kinds of expressions the people who owned the faces might make, their skin and hair colors…

    Scored above average, which surprised me, because I always thought I was terrible with faces. Maybe I am just terrible with names.

    I wonder whether the internet average for this test is lower than the world average. Probably, right? What about the social scientist average, the businessman average, and the statistician average?

  4. Phil says:

    I think I'm somewhat worse than average at recognizing faces, and I'm definitely worse than average at associating names with them. In fact, I occasionally go to quiz night at a pub with some friends, where one of the tasks is to recognize the people in about 10 or 12 photos…they might show Hillary Clinton's yearbook picture, that sort of thing. I can usually get one or two — four would be an excellent performance for me — but my friends can often just race through almost all of them with no problem, leaving only a few to guess at.

    But I scored about average on this test: 60 faces right out of 72, which is 83% correct. Of course, I may have gotten a bit lucky. For what it's worth, I didn't use Andrew's technique of memorizing specific features, but it did come naturally to think of "the guy with the chubby cheeks", "the guy who looks like an elf," and so on. I thought the test was really hard, and if you had asked how many I _thought_ I had gotten right, I would have guessed that I was in the 60-70% range.

  5. Ken Williams says:

    I recommend you go listen to the following program, one of the stories is very relevant here (and totally gut-wrenching):

    Again, this is closer to Sacks' mistook-for-a-hat stuff than mild facial nonrecognition though.

  6. K? O'Rourke says:

    Recall someone suggesting there was a particular area of the brain that is involved in face recognition and we have evolved to be good at it.

    Also has something to say about Chernoff faces as a ploting method.

    On a personal note – I am not very good either. When I was young I had to be very careful to schedule the meeting for the first date after meeting a girl at a dance or party – to ensure I did not get caught not quite recognizing her.


  7. JL says:

    I don't think I'm bad at recognicing faces, but I got 65 percent correct on that test. I recall that face recognition is one of the very few mental tasks that is not correlated with IQ.

  8. Manoel Galdino says:

    I thought I was not very good either, but gess what? I got 82% correct.

    But, in the end, I was finding the test irritating too. It's boring.

  9. Bob Carpenter says:

    It's damn near impossible to make statistical arguments in court because no one has any stats background among the lawyers, police, judges and jury.

    My sister's a criminal defense attorney and oversees lineups as part of her job. She also cross-examines eyewitnesses. She takes the time to go to the scene of an alleged crime and take measurements. Sometimes the judge lets her run the eyewitnesses through a simulation, but often they don't.

    Shockingly, though not surprisingly, now that the judges and prosecutors know that she's thorough before trials AND persuasive during trials AND willing to go to trial, they offer her clients much better plea bargains than when she moved back to Detroit six or seven years ago.

  10. K? O'Rourke says:

    Yes Bob – a real chicken and egg problem for stats – the judges can't let the lawyers make stat arguments becuase they [judges] don't have the background so lawyers have no incentive to learn [anything more than intro} stats …

    My favourite one was a judge who when he heard the expert witness mention standard error belowed "I will not allow errors in my court – standard or otherwise!"


  11. The Science Pundit says:

    60/72 83%
    I guess I'm average (which still isn't really reliable enough when it comes to eyewitness testimony of a stranger's face imho).

  12. Andy says:

    Interesting – just tried it and got 61/72 (85%).

    I had a feeling of OH COME ON, especially when it got to the last block with the noisy images, but initially it felt all effortless.

    I wonder how well you do at tasks like Block Design and Embedded Figures.

  13. Jeremy Miles says:

    I don't consider myself great at recognising people, and I got 65%. Towards the end, I felt like it was just a blur of faces – I had no idea if they were in the target or not. But I guess I had some idea, or I wouldn't have scored _that_ high.

  14. MacGBrown says:

    I was on a jury with a lineup a few years back. The prosecution avoided the issues of proving the validity of their lineup by allowing the defendant to choose the rest of the panel. We were shown photos of the lineup, and there was no way I could have picked out the defendant. I was glad that the eyewitness testimony was a fairly small portion of the total testimony.