Micro Expressions Test

Post Reply New Topic

Three.

Yikes. I only got one right. And I was guessing. Faces already look like just pieces and fragments already. Looking at a still shot of a split-second expression (which would never be that short in real life, which kind of irritates me -- people's faces rarely move that fast) is like asking for trouble for me understanding it. I couldn't even see most of the face. I just saw pieces.

I also wonder (although I don't think it would make me do any better, I just always wonder about such things) whether these were real expressions or acted expressions. Acted expressions don't necessarily test ability to read faces, they often just test ability to read stage conventions. Stage conventions are not the same as real facial expressions, they're just something that most people have learned by seeing them on TV and stuff so long. (That's why some people even think that real expressions look "unrealistic". Things that are real don't have to be realistic.)

So, this could well test your ability to detect a split-second view of a stage convention, something that rarely happens in real life.

I do better in real-life situations. I can sense the pattern of movement in someone's body and get a lot from context. As long as I'm not also dealing with language or numbers etc. or any of the things that lead to them (which this test forces you to do, and most tests force you to do).

Which is why I've told researchers how to design an experiment that wouldn't depend on language use and that would employ real emotions evoked in a real situation. Whether they actually go ahead with it is up to them, but I've let them know how.

I just gave the test to my mom and she got 9-10 right. She is an introverted NT and wanted to know what that meant about her. I hope I gave her the right answer!

You're missing one option in your poll : Couldn't do it, it FREAKED ME OUT!
I was there focusing to read the emotion, the person seemed to be "interested" in every case which is not even in the suggestions ( lol) and suddently his face jumped!! i had to go through 3 images before i realised it was not just a bug, and that this scary jump was in fact an expression (the one i was supposed to read ) so my answer to all would be: super scary face tic. That is all.

Funny, I am so good at telling people's moods just by looking at their face. I can also tell if somebody is joking or not, just by looking at their face and ''reading'' their expression. But each time I take one of these tests on the computer, I fail, every time.

That's probably because these computer tests are very contrived and don't actually depict realistic situations.

Actually, I think I agree with anbuend that these tests are contrived. For instance, in this test, you do not see a face change from a static expression through a micro-expression back to a static expression. You see an image of a static expression followed by an image that is supposed to be a micro-expression and quickly followed by the first image again. The face doesn't move dynamically from one expression to another, it's just switch between static images.

ill do it later

I'm not sure I used the right words. What I mean is that practically all of these tests that purport to show that autistic people can't read facial expressions, don't test this in a realistic way. They will generally include things like (one or more of):

Forcing the person to use/understand words during the test.

Using actors rather than genuine situational facial expressions. This can mean that the test is a test of familiarity with cultural stage conventions more than anything else.

Choosing to represent which is the right answer by determining how a certain number of nonautistic people interpret the expression, with no way of telling whether those people are right or wrong. (One of the weirdest ones I ever saw was interpreting an expression as "desire" when it looked like an expression that would CAUSE desire in the viewer if they were straight and male, but not necessarily conveying desire in the person whose face it was.)

Taking the faces out of magazines and other situations that make it hard to verify what the expressions were.

Limiting faces and expressions to only one culture/ethnicity regardless of culture/ethnicity of the people being tested.

Using cartoons or line drawings that don't represent actual features of a face (for instance real smiles and cartoon smiles look different) and/or that use cartoon-style conventions that defy reality.

Making changes to the presentation of the face that differ from how things would happen in real life, for instance:

* Cropping the face to show a smaller piece of the face than would be visible in most real life situations.
* Making the face black and white or otherwise altering the coloring.
* Using stills rather than moving video.
* Flashing an expression for an abnormally short period of time, and without the natural progression of the physical movement of the face.
* Many others I don't remember this instant.

Testing only certain sorts of autistic people then generalizing the results to all of us.

Or worse in some ways, using common assumptions of severity when generalizing to all of us. For instance, testing people who can speak, understand spoken language, read, and write. Then assuming that not only do these results apply to people who can't do more than one of these things, but that these results apply more severely to people who can't do one or more of these things. So if speaking/literate/etc. people can't read the expressions then nonspeaking people or people with limited language comprehension must have even more severe trouble reading expressions.

Testing autistic people while they are in one particular mode and assuming it applies to them in other modes. For instance testing autistic people in a mode that forces them to use language then assuming that the same results apply to them when they can't understand language.

Assuming that when autistic people perform worse than nonautistic people on these tests, it means autistic people must perform worse than nonautistic people in reading facial expressions in real life. (Assuming that the test represents ability to read genuine facial expressions in natural situations.)

Assuming that the test shows not only abilities to read facial expressions, but body language in general, or worse, all nonverbal cues in general.

Testing only for typical face/body language reading abilities, not for the atypical body language reading that many autistic people describe in ourselves. Often done because the testers have never heard of the ways some autistic people can read other people. It's hard to test abilities that you don't even know exist.

And many more flaws, all at once adding up to a situation where what the test really measures and what it claims to measure are totally different things. Add all those things together and you have a situation where you can't even tell what you're measuring, if anything. So many things are potentially wrong, that you can't even tell what they are. And that's besides the ones I wasn't able to write down here.

Well... I grant that tests have many flaws, but how would you design a better test? Testing 'real life' situations has way too many variables, so you have to simplify - as far as I can tell, the best thing to do would be to come with something like this, which tries to get to the essence of what's going on. Do you have other ideas, though?

I showed this test to two NTs ->not at the same time, and both found the test impossible. The first got 2 right and the other got 4 right, and I played the mimics many times -> about 3 to 5 times per face.