Eye test revisited

Post Reply New Topic

I went again through the Reading mind in the eyes test by Baron-Cohen, but using a differen method. I printed the test and did it without looking at multiple-choice answers (I covered them with a sheet of paper). This method is probably more realistic, because in real life we don't get multiple choices conveniently displayed next to other person's face.

I assigned 1 point if I answered the exact correct word, a synonym, or a word of similar meaning (similarity being defined quite liberally), for example:
amorous = flirtatious
thinking = contemplative
playful = happy
anxious = anticipating
decisive = confident

I assigned 1/2 of a point if I selected an incorrect feeling, but one that is at least in the same category as the correct one, for example:
distrustful - cautious
happy - fantasizing
serious - hostile
scared - concerned

The result: 17.5 correct out of 36 (14 correct not counting half-points). When I could look at the multiple-choice answers, I got 31 out of 36. The typical range is 22-30. Interesting discrepancy, wouldn't you say?

Last edited by magic on 12 Aug 2004, 3:21 pm, edited 1 time in total.

I think that that is a much better way of going about it, Magic. I also felt that the multiple choice options were very unrealistic. I don't think I'm going to print the test to retake, but I do recall that taking the test, I was often going through a process of elimination before choosing an answer, rather than just picking the right one.

We also don't get a lot of time to figure out what a person's face is saying, in real life. Even if we can stare at the image for a few seconds and puzzle it out, most people's faces change at a much faster rate. They are also probably speaking to you at the same time, as well. I often find that when I'm trying to work out someone's expression, I completely miss what they are actually saying. Or vice versa.

Yes, but the typical range is also measured using the site's method of multiple choice. I imagine that the typical range would also drop slightly if it were not multiple choice.

I just finished taking the test (the one on the computer that you posted the link to). I got 20 out of 36, which means that I "find this quite difficult."

You are right. To illustrate the difference I calculated that the expected result of selecting multiple choices at random is 9 (=36/4). On the other hand, the expected result of the doing the "free choice" method randomly is 0.49 (=36/74, assuming that one has a vocabulary of 74 words describing feelings, as used in answers in the test). There is no way to calculate the typical range for the "free choice" method without a sufficiently large sample of answers, but you are right that it will likely be lower than in the "multiple choice" method. I tried to level this discrepancy by grading my answers liberally. In fact I gave 14 correct answers out of 36, the rest of the score came from "halves" that were in fact wrong answers.

I agree with you that when doing the test with multiple choices, one can rely very much on elimination. Just figuring out if the emotion is positive or negative in most cases cuts number of possible choices by half. This is likely the reason for the difference in my scores.