The 'eyes' test- ToM ?

Post Reply New Topic

Is it a good test of ToM ?

http://socialintelligence.labinthewild.org/mite/

I scored better than 54% of people but don't think my ToM is that good. Without a list of answers to choose from and having to guess an emotion on the spot,it would be a lot more difficult.

I took the test...

Mine is 15 out of 36. Half of the time they do not have the word I think of when I look at the picture so I have to just guess from the available options. I wish they had the answers to look at after the test.

My results:

I don't think this is a good test for Theory of Mind, this test test is more about reconising emotions in others, or Affective Empathy.

Theory of Mind is about being able to see things from other people's perspective, or Cognitive Empathy.

LostDragon, where did you find the bit saying you are equal to or better than 28% of participants? I didn't find that kind of information on my results page.

Weird, my percentage score showed up next to my results, and unfortunately I've already shut the tab down.

It didn't tell me what the average score was though, unlike it did for you. Try looking above or below that text. Make sure your web page isn't too zoomed in or out or something, as it may not be displaying all the text for some reason.

I got 27. I think this test would be much better if there were just a single, long list of the possible emotions, from which one would choose an answer for each question. Having to choose from only four options means that often one can arrive at an answer by a process of elimination of the obviously wrong answers, together with a bit of guesswork. It would be more realistic if one were not given clues by being presented with a restricted set of options for each question.

I agree with SpendidSnail, it's not a good ToM test. Attaching the correct "label" to each image does not require the ability to empathise with the emotion itself, and in principle could be learned purely by disinterested observation and noting people's word associations. As an Aspie with a lot of alexithymic traits, I use this kind of conscious "pattern recognition" quite a bit when I interact with people, and even to help me judge my own emotional state.

I think that there are other reasons why this kind of test can be misleading, and possibly more so when testing neuro-diverse people...

- Attention. It's all very well being able to comprehend the emotion when the stimulus has been explicitly pointed out to you. But a person who has difficulty with focusing their attention, is susceptible to distractions, or finds eye contact difficult or uncomfortable may not be aware of these cues most of the time. You can't recognise what you never saw in the first place. This might always be the case for someone with visual impairments.

- Sensory integration. The expression in someones eyes is only one emotional cue among many, overlapping with each other, and received by different senses. Difficulties reading one cue might influence a person's judgement of another, especially if they have sensory integration issues.

- Context. A person's expression doesn't always match what they are feeling right now. They could be telling a story, mimicking someone else, telling a joke etc. In this case, the "eye reading" might be accurate, but poor ToM when interpreting the context could still lead to the wrong conclusion.

My score was better than 36% of people. Somewhere middling was what I expected as aversion to eye contact, sensory processing and keeping focus seem to be the things that I struggle with more - I seem to judge what I notice quite reliably, but the cues wash over me without being noticed a lot of the time.

If you do that again, press ctrl + shift + t altogether and it will restore other tabs in the order that you shut them down. It really helps when you shut one down and then change your mind and want to look at something again- it even keeps the history so you can still press the back button in the restored tab.

I retook the quiz today, and got 26/36 (so two more correct than last time) and 45%



Also, thanks Goth Fairy- I shall have to keep that in mind next time.

Such a test would be quite difficult.

Especially considering that some emotions are quite similar and frequently confused (empathy VS sympathy, jealousy VS envy). I think if you were to have such a test, then you could have two circles (one where you put all the emotions you think it displays, and another for emotions that it doesn't display).

Then it would give you a mark for the ones that are correct that you put in the circle for the emotions it shows.

If the correct answer was "anger" then it should probably count "frustration" as correct as well. Also, considering this quiz is already just restricted to the eyes it already has a level of difficulty, since when people read body language they don't only look at the eyes but also other signals (hand movements, posture ect.)

Personally, I have a friend who doesn't make eye contact, so I only have what his other body language says as well as what he's saying, and his voice tone to go by. Despite this, I still find it fairly easy to understand him.

Also, am I the only one who found out there's a facial expression for sarcastic from this quiz? I always thought it was just in the tone (and only the tone) but apparently not. A quick Google search confirms that there is a facial expression for sarcasm. I guess that makes sense.

I got 25, but i was guessing a lot as most times I didn't get anything at least not one of the four options.

That is a very good suggestion. I think for the test to have any real value, it should not steer the subject into choosing one out of small subset of possible answers, in the way the present test does. But also, it cannot simply allow "write-in" answers, because it might not know how to interpret what the person typed. Your solution, where one has to put all the plausible emotions, selected from a universal list for the entire quiz, into one bin, would be a good way to do it. The program could then score the responses according to some metric measuring the closeness of the selected answers to the "actual" emotion that was depicted.

More work to program it, but I think the results would be more meaningful.