Face recognition, or lack thereof

[lorimt]

You know how I always say I'm bad at face recognition? Here's evidence!

I took this test of face, name and object recognition, run, I believe, by the University College London and Harvard.



-------------------------

Face Recognition Memory

In the face recognition test, you were asked to learn the faces of 10
women. When you were tested on these faces, you had to indicate
whether a face on the screen was one you learned or one that you
hadn't seen before.

Out of 20 familiar faces, you correctly responded that you had seen 12
faces before.

Out of 30 unfamiliar faces, you correctly responded that you had NOT
seen 24 faces before.

Overall, you got 72% correct! The average score on this test is 85%
correct. Your percentile rank on this test was 9.


Object Recognition Memory

In the object recognition test, you were asked to memorize 10 images
of eyeglasses. When you were tested on these eyeglasses, you had to
indicate whether a pair of eyeglasses on the screen was one you
learned or one that you hadn't seen before.

Out of 20 familiar eyeglasses, you correctly responded that you had
seen 19 eyeglasses before.

Out of 30 unfamiliar eyeglasses, you correctly responded that you had
NOT seen 26 eyeglasses before.

Overall, you got 90% correct! The average score on this test is 86%
correct. Your percentile rank on this test was 65.


Verbal Recognition Memory

In the verbal recognition test, you were asked to learn the names of
12 men. When you were tested on these names, you had to indicate which
of four names was one of the names that you had learned.

Out of 12 sets of names, you correctly chose the familiar name8 times.

Overall, you got 67% correct! The average score on this test is 77%
correct. Your percentile rank on this test was 25.

--------------------------------------------------------------------

I felt a sense of panic throughout the face recognition test, both in trying to learn the faces, and later when trying to recall them. I had this overwhelming feeling I'd never seen any of them before, and that aside from a few distinctive eyebrows and outlines I was blindly guessing. In contrast, the eyeglasses felt like a piece of cake, with distinctive shapes all over the place. The names were also sort of odd, in that I did better than I initially expected, but worse than I thought I was doing mid-test. That, and their suggestion to read the names aloud helped immensely.

The point is, maybe it isn't actually names I'm bad at (or at least, only names) but faces. Definitely an interesting test.

Comments

[purplebob.livejournal.com]

Surprisingly, I got 48th percentile for faces and 84th percentile for names. But this doesn't seem to help me recognize students that I'm TAing, or remember people's names at places like Mathcamp. They should have had a test for *connecting* names with faces.

At least this term I'm teaching a recitation so huge I won't be expected to know anyone's name.

[iluvsheep.livejournal.com]

I agree, it's the connecting faces and names that is the hard part.

[archonsengine.livejournal.com]

Okay, that was quite an interesting test. Thanks for passing it along!

I also felt really lost on the faces, but went with my gut and apparently was awesome (92% correct, 80th %ile). I found the name one easiest (83%, 65 %ile), though I blame my annoyance with the eyeglasses (96%, 86 %ile) on colorblindness.

I didn't post the scores just to show off; more to highlight how poorly my perceived skill and actual skill matched up.

[zwilichkl.livejournal.com]

Hm, I got 94% (84 percentile) on faces, 90% (65 percentile) correct on glasses, and 75% (48 percentile) correct on names. I always recognized a face if I had seen it before- it was a gut reaction, though I sometimes thought I had seen one before when I hadn't. Fun test!

[spiffystuff.livejournal.com]

XD I thought I was bad, though I guess you're worse ;D

Faces: 76% correct/16th percentile
Objects: 88% correct/55th precentile
Names: 92% correct/84th percentile

[inferno0069.livejournal.com]

I think I'm around where I'd expect to be based on how the tests work, maybe a bit better at faces than I'd have expected.

  Faces: I got 78%, 20th %ile, average was 85%
Objects: I got 96%, 86th %ile, average was 86%
  Names: I got 92%, 84th %ile, average was 77%

[jeffspender.livejournal.com]

Faces: 96% correct, 88th percentile.
Objects: 94% correct, 82nd percentile.
Names: 83% correct! 65th percentile.

It's interesting, doing that well on the eyeglasses took a bunch of conscious cognitive effort, and felt like I had to intentionally try to remember specific features about each one. The faces, on the other hand, took very little effort. I guess I couldn't really latch on to very many specific features, and was mostly perceiving the faces as actual people. In fact, when I looked at each of the people I was supposed to memorize, I could feel my brain immediately start making an empathic connection to them, trying to sense what they were feeling and what their personality was like... to the point where when I was trying to do the sorting, I started to stress a bit because I knew I would feel guilty if one of "my" people came up and I didn't recognize them. How stupid is that? :P

Or maybe not... maybe those were my ingroup/outgroup circuits kicking in...

[iluvsheep.livejournal.com]

Apparently, when I actually try, I'm good at recognizing images I've seen before. Names, not so much.

	Familiar	Unfamliar	Percent	Rank
Faces	20/20	28/30	96%	88
Objects	18/20	30/30	96%	86
Names	9/12		75%	48

Although, I do think it was cheating that they only quizzed the last names when I was trying to remember both.

[ztbb.livejournal.com]

90% on the faces and glasses, 100% on the names (93rd %ile, I think).

[tajmahall.livejournal.com]

92%, 80th percentile.
84%, 45th percentile.
92%, 84th percentile.

I think it'd be better if they showed you different photos of the people and objects each time you saw them. My strategy revolved more around detecting which actual images I'd seen before, not the faces in them exactly. Often the contrast or some weird pixel was the giveaway. Apparently this wasn't successful for objects.

Another test - Cambridge face memory

[lorimt]

Out of 72 faces, you correctly identified 48.
In other words, you got 67% 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.

--------------------------------------------------------------------

Yep. I'm still really bad at it. Maybe I should contact them...

[notsteven.livejournal.com]

Faces: 84% / 48th percentile
Glasses: 80% / 25th percentile
Names: 92% correct / 84th percentile

The names I'm not too surprised about. I was making mnemonics for most of them, which helped. My main problem in remembering names is that I'm usually so focused on navigating the introduction that I forget to take note of people's names.

And I'm not too surprised at being low percentile on the glasses. A lot of the time, I have trouble even remembering if people wear them. I just don't notice them all that much.

[telso.livejournal.com]

Reminds me of the alt-text of http://xkcd.com/302/ .

[hmdavid83.livejournal.com]

16th percentile on faces
85th on objects
65th on names

No wonder it takes me FOREVER to learn students' names!