Here's my interpretation, correct me if I'm wrong:
1. The g-factor and g-loadings are calculated by factor analysis based on a collection of various cognitive tasks
2. These collections typically *do not* include mind-reading tasks (even though mind-reading is arguably a cognitive task, what else could it be?)
3. If we were to include the RMET (and similar emotion-related tasks) into the datasets used to calibrate the g-factor, that would rotate the vector a little bit, and slightly change the loadings of the different tasks
4. That would slightly increase the average IQ score of women and decrease the average of men, possibly explaining a part of the gender gap
5. That may or may not improve the accuracy of IQ scores for predicting job performance/anti-social behaviour/etc (I suspect that recognizing emotions is relevant for job performance, at least in some jobs)
Yes, any aspect of intelligence not currently included may change group gaps a little bit. So face reading, spatial rotation, mental speed/chronometric tests etc. could all be added.
"In this test, the participant is presented with a series of 25 photographs of the eye-region of the face of different actors and actresses, and is asked to choose which of two words best describes what the person in the photograph is thinking or feeling."
It's not needed to be anything. Test authors can make more items and then just discard the bad ones (i.e. if 90% people think that item #2 is "assertive" it probably is, and then it people who are good at guessing items #2, #3, #4 can't agree whether item #5 is "sad" or "resting" (with 50% answers for each) then item #5 gets discarded).
My first question about the test is how the correct answers are determined. How do we know what the people in the photographs are thinking and feeling? Are we measuring what people look like when they're feeling pensive, or are we measuring what actors look like when you direct them to look pensive?
This distinction is crucial for trial lawyers, among whom it is well known that the jury's idea of how a witness should carry themselves, and what their demeanor says about their credibility, is generally unrelated to how witnesses do in fact carry themselves, or to their credibility. But actors directed to display an emotion will do so in a way that "the jury" considers natural, and recognizing this is not the same thing as recognizing what a person is actually feeling.
My other thoughts on the test:
- I seem to have scored about average, 25/36 by report of the website.
- In some cases the person's eyes were not visible. In one case (it was the second question for me), the picture appeared to be graphical noise, and I was not able to recognize that it contained a human. This seemed unfair. In another case, one eye was heavily occluded by shadow and the other eye, or rather the space where a normal human would have one, was entirely blacked out by shadow.
- In one case, the eyes appeared to be East Asian. This seemed unfair too, though less so than the random graphical noise. I see that you've noted that all the photographs are of Europeans. I stand by my opinion that, if this was a European, it was a European with remarkably Oriental eyes.
I suspect that all tests of this form will show strong racial effects. You might be interested to know that the Chinese have rejected the basic smiley 🙂 - in current use, it indicates sarcasm or mockery - on the grounds that a smiling person does not have open eyes. For friendly smiling, they use 😄 (or 😊 if female).
I will bet that people who think too much about the images do worse, and that timing it so that people were forced to rely on first impressions would improve the reliability of the results..
I have doubts when I finished the test as I have a very high score despite being a slightly autistic person, Do you know of any study that controls Hispanic voters tendencies by their race?
What does this mean for the claim that men's g-factor is higher than women's, that I think you mentioned some time ago?
According to a cursory search, it looks like women have been doing consistently better at mind-reading than men, with effect sizes around 0.2:
https://doi.org/10.1016/j.cpnec.2022.100162
https://psycnet.apa.org/record/2013-09240-009
https://www.pnas.org/doi/abs/10.1073/pnas.2022385119
Here's my interpretation, correct me if I'm wrong:
1. The g-factor and g-loadings are calculated by factor analysis based on a collection of various cognitive tasks
2. These collections typically *do not* include mind-reading tasks (even though mind-reading is arguably a cognitive task, what else could it be?)
3. If we were to include the RMET (and similar emotion-related tasks) into the datasets used to calibrate the g-factor, that would rotate the vector a little bit, and slightly change the loadings of the different tasks
4. That would slightly increase the average IQ score of women and decrease the average of men, possibly explaining a part of the gender gap
5. That may or may not improve the accuracy of IQ scores for predicting job performance/anti-social behaviour/etc (I suspect that recognizing emotions is relevant for job performance, at least in some jobs)
Am I understanding this correctly?
Yes, any aspect of intelligence not currently included may change group gaps a little bit. So face reading, spatial rotation, mental speed/chronometric tests etc. could all be added.
I always wondered how the "correct" emotion of the eyes was determined by the test's creators. A poll? Interviews with the subjects in the photos?
"In this test, the participant is presented with a series of 25 photographs of the eye-region of the face of different actors and actresses, and is asked to choose which of two words best describes what the person in the photograph is thinking or feeling."
https://docs.autismresearchcentre.com/papers/2001_BCetal_adulteyes.pdf
Hmm. "Actors and actresses"...so apparently these are scripted/faked emotions, not necessarily genuine emotions.
It's not needed to be anything. Test authors can make more items and then just discard the bad ones (i.e. if 90% people think that item #2 is "assertive" it probably is, and then it people who are good at guessing items #2, #3, #4 can't agree whether item #5 is "sad" or "resting" (with 50% answers for each) then item #5 gets discarded).
30/36, autism card revoked.
enjoyable post.
My first question about the test is how the correct answers are determined. How do we know what the people in the photographs are thinking and feeling? Are we measuring what people look like when they're feeling pensive, or are we measuring what actors look like when you direct them to look pensive?
This distinction is crucial for trial lawyers, among whom it is well known that the jury's idea of how a witness should carry themselves, and what their demeanor says about their credibility, is generally unrelated to how witnesses do in fact carry themselves, or to their credibility. But actors directed to display an emotion will do so in a way that "the jury" considers natural, and recognizing this is not the same thing as recognizing what a person is actually feeling.
My other thoughts on the test:
- I seem to have scored about average, 25/36 by report of the website.
- In some cases the person's eyes were not visible. In one case (it was the second question for me), the picture appeared to be graphical noise, and I was not able to recognize that it contained a human. This seemed unfair. In another case, one eye was heavily occluded by shadow and the other eye, or rather the space where a normal human would have one, was entirely blacked out by shadow.
- In one case, the eyes appeared to be East Asian. This seemed unfair too, though less so than the random graphical noise. I see that you've noted that all the photographs are of Europeans. I stand by my opinion that, if this was a European, it was a European with remarkably Oriental eyes.
I suspect that all tests of this form will show strong racial effects. You might be interested to know that the Chinese have rejected the basic smiley 🙂 - in current use, it indicates sarcasm or mockery - on the grounds that a smiling person does not have open eyes. For friendly smiling, they use 😄 (or 😊 if female).
I thought sense of rhythm was uncorrelated with general intelligence.
I will bet that people who think too much about the images do worse, and that timing it so that people were forced to rely on first impressions would improve the reliability of the results..
I got 34/36 and the two I got wrong were my second choice. I found my first instinct was almost always correct.
I have doubts when I finished the test as I have a very high score despite being a slightly autistic person, Do you know of any study that controls Hispanic voters tendencies by their race?
I was mainly baffled by the women's eyes, but still did okay with 29/36.