Seeing emotions in dots

Point-light displays can tell us an amazing amount about other people. Looking only at a few glowing spots corresponding to joints and set in motion, we can tell what people are doing, whether they are over- or underweight, and even identify a friend among strangers. We can also identify animals or determine the emotional state a dancer or actor is conveying. But some emotions are more difficult than others. Take a look at the following two animations (click on the image to view a quicktime movie):

i-21686267b685f7f8ef31a5c0fca1c723-point1.gif i-94f7e865e95606f9e51ef7a39e5eec00-point2.gif

Now, which emotion does each animation convey? Your choices are Anger, Joy, Sadness, Love, Fear, or Disgust. You can post your guess in the comments section, and we'll let you know when someone gets it right.

These figures came from an experiment conducted by Tanya Clarke and several colleagues (you can visit the Perception web site to see more examples, but be prepared for a long download!). They had several different actors perform the same scene but conveying different emotions (you may have seen comic Colin Mockery doing this same thing -- to hilarious effect -- on the improvisation TV show Whose Line Is It, Anyway?). The actors were allowed to repeat the scenes until they believed they conveyed the emotions adequately. Participants viewed the scenes as silent point-light displays, then rated them for how well they conveyed each of the emotions in the list. Here are the results:

i-91ade8f4561679004addf221c25e13a4-emotionpoint1.gif

As you can see, the viewers had little trouble with this task. When actors were portraying anger, for example, the participants rated the point-light displays as 70 percent effective in conveying anger, but all the other emotions averaged less than a 20 percent rating. The difference between the intended emotion and other emotions was significant in every case.

Next, the same viewers saw the same displays inverted -- upside-down. This time, the results were less consistent:

i-84fe0bec3e131ca9c1019b9f8f91a09f-emotionpoint2.gif

While viewers were still able to recognize anger, joy, sadness, and love, the two other emotions -- fear and disgust -- were rated as highly for other emotions as they were for the intended emotions.

Next Clarke et al. designed a new experiment to test what aspects of the displays were critical for expressing emotion. They generated two new types of animation -- one where one of the actors was cropped out of the picture, and another where instead of two actors interacting, the display was a reflection of the same actor's image. Here are the results:

i-89640a92c8e29fa2ea311120a875237f-emotionpoint3.gif

For anger, joy, and sadness, each display was equally effective at expressing the intended emotion. However, when attempting to express love, one actor alone was significantly less effective compared to two actors together. Viewers even rated the reflection of the identical actor to create two "people" as expressing love. Love, it appears, requires two people in order to be displayed effectively, but those two people needn't actually be interacting in a meaningful way. For fear, people rated the reflected image as less effective than the other displays. So fear appears to require one person reacting to another, or one person alone.

Perhaps the most important finding of this study is how often emotion can be conveyed with point-light displays, and how versatile the visual system is at perceiving emotion. Out of 27 types of point-light displays across two experiments, only four failed to express the desired emotion.

One more note: a tremendous resource for playing with point-light displays is the BioMotionLab. If you haven't tried this tool before, you should definitely give it a whirl. You can even create your own custom point-light display conveying any emotion or body type you desire.

Clarke, T.J., Bradshaw, M.F., Field, D.T., Hampson, S.E., & Rose, D. (2005). The perception of emotion from body movement in point-light displays of interpersonal dialogue. Perception, 34, 1171-1180.

More like this

Take a look at this video. Your job: decide which person speaks first: Click to view Video 1 (QuickTime required) Let's make this a poll: Which person in Video 1 speaks first?(web poll) Now, try another one. Once again, decide which person speaks first: Click to view Video 2 (QuickTime required)…
It's now taken as a given that the musical score of a movie can have huge influence on our perception of the movie. From the pulsating terror achieved in films like Psycho and Jaws, to the triumphant victories in Star Wars and Pirates of the Caribbean, it's hard to think about a great film without…
Classical Indian dancing is a tradition that extends back 2,000 years. Unlike much Western dance, it is intended to express specific emotions and tell detailed stories. The Natyasastra, a text from the first or second century A.D., offers instructions for how to depict nine primary emotions, and…
Disney's purchase of Pixar makes it clear that computer-generated (CGI) animation appears to be the wave of the future in movies. But one difficulty with CGI animation is conveying realistic emotions. While film animators (whether they use computers or not) can use artistic license to achieve the…

I think the first one is Sadness, and the second one is Anger. The first person is standing, kind of swaying a little and has one little cough at the end. The second set is a pair, with one looming over the other who throws (her?) hands in anger.

yeah, I agree with Scott.. the first one seems to be a gloomy depressed sorts, no hope.. the latter, a confrontation, the left one trying to defend itself, but kinda admits fault.. head hangs in shame..

I second Scott's comment, Sadness and Anger. In fact, the anger could be determined from the still alone, and Sadness was my first guess for the still of the first image.

interesting how little information is needed by human perception.. in the light of this knowledge, its amazing why you vision system is as accurate and advanced when for a lot of things a much worse vision system would've worked as well..

also, this article might hint about how these emotions might itself be stored in the brain, and how so little information can help conjure up a detailed explanation of the situation..

Scott's right on the second one -- it's Anger. However, no one has gotten the first one yet. So Sadness has been eliminated. Keep those guesses coming!

It could be a nervous, middle school kind of love. Very "So, um, I was wondering if you, um, wanted to maybe get a coke sometime." I suppose it could also be disgust, but more in a "tut-tut" sort of way than an "ew, that's repulsive" sort of way. It's certainly not anger or joy; sadness has been eliminated; it could be on the extremely nervous side of fear, but I doubt it.

Maggie, you're right. It's love, disgust, or fear -- but you've got to make more of a commitment to a guess before I confirm the answer.

I should point out that actors were prohibited from making dead-giveaway gestures like shaking a fist in anger, and physical contact was also a no-no.

I know it's not an option, but the first one seemed like 'indecision' to me... Like a person listening to someone else hesitantly, not necessarily agreeing with what they are saying. I guess disgust is closest to that....

Going out on a limb here:

Is it possible that the first one is love? You indicate that love is difficult to tell when there's just one person, and it's possible that the dot-man is daydreaming??

But I thought it was sadness at first, too.

We have a winner! Jaclyn got it -- the first animation is Love. I was wondering if anyone would pick up on the fact that love was difficult to pick up on when only one person was present. Note that love was readily distinguishable -- even upside-down -- when two people are included in the animation.

Cool clips. Interesting too that Anger seems to be the most reliably recognized, even when it was 'distorted'. Seems to agree with the facial expression perception work that shows Anger is easiest to detect.

I wonder how much of the conveyed emotion comes from the *speed* of the movements, versus the movements themselves? I expect it's some combination of the 2 factors, but it would be neat look at how it breaks down -- i.e., if the actual motions themselves are enough, regardless of speed.

I'll go so far as to guess that the second clip was derived from a Charlie Chaplin silent movie. Something about the short person's (the one to the left) angry/moping gestures reminds me of him

I've heard from Tanya Clarke about this article, and just wanted to add her comments to the mix:

You might like to mention that not all emotions are easily conveyed via body movement; disgust was actually not easily recognised and was confused with sadness, anger and fear in our original experiment which is why this emotion was not used in the dyads vs monads vs reflections experiment. However, this emotion is easily perceived from a point-light display of the face (Bassili, 1978) and disgust is also easily recognised from static pictures of facial expression. This is not surprising given that disgust is primarily an adaptive response to protect us from ingesting noxious substances (although there are arguments about this narrow definition as Ekman 2003 argues that it also has a strong moral element). Therefore, it seems that the channel is a good channel for the expression of most, but not all, emotions.

I have been interested in your readers' comments about the movies and the work and would like to just point out that one person is spot on about the fact that both spatial and movement information contribute to the perception of affect, for example, it seems that low-level cues such as velocity are more important for anger whereas for sadness spatial info may be more important. There is a lot of research about the contribution of spatial/ kinematic info. The Charlie Chaplin comment made me laugh!

Funny, I thought the first one (before I knew what the choices were) was boredom. A guy with his hands in his pockets, standing around, nothing to do. Would have never guessed it to be love.

I think the first portrays anger and the second, happiness and excitement.