Cognitive Daily

How long does it take to decide if someone’s attractive? It might be before you even know you looked.

Researchers can use a masking technique to show an image of a human face subliminally — without the observer being aware of seeing it. To do it, they first show a scrambled face (39 milliseconds). Next, the face itself (13 milliseconds), a blank screen (13 milliseconds), and a cartoon face (39 milliseconds). I’ve tried to duplicate the technique using an animated GIF file, but I think it’s beyond the capability of an ordinary web browser.

Click here to see it

Did you see a human face? I had no trouble recognizing my son Jim in the animation, but you might be less aware of the image.

But Ingrid Olson and Christy Marsheutz were able to effectively mask images in the laboratory to try to measure how quickly humans can judge attractiveness. How did they do it?

They found 40 attractive and 40 unattractive faces from highschool yearbooks. A separate panel of observers rated the faces on a scale of 1 to 10, and the average rating of the attractive faces was 7.44, while the unattractive faces averaged just 2.45. Next, they showed these faces to a new group of observers using the masking technique I described above. None of the observers reported being aware of seeing a face at all, but the experimenter asked them to make their “best guess” about the attractiveness of the face. Attractive faces rated a 5.79 on average, while unattractive faces were rated significantly lower: 4.71, on average. So after seeing faces for only 13 milliseconds, we are still able to make reliable judgments about attractiveness.

Does this ability have any effect on our behavior? In a second experiment, attractive or unattractive faces were flashed for 13 milliseconds, followed by either a positive word (“laughter”) or a negative word (“crying”). Observers were told to ignore the face and simply judge whether the word was “good” or “bad.” For good words, reaction times were significantly faster when observers had seen attractive faces beforehand.

So does this effect only apply to facial attractiveness, or does attractiveness of other items affect us the same way? In a new experiment, Olson and Marsheutz found 40 attractive and 40 unattractive houses from real estate listings. Again, independent observers agreed with their assessment, with the “attractive” houses rating 8.03 on average and “unattractive” houses rating only 2.2. They repeated the word evaluation task with a new set of observers who saw both faces and houses, after seeing each for just 13 milliseconds. Here are the results:


For identifying good words, reaction times were significantly faster after viewing attractive faces compared to unattractive faces, whereas there was no comparable effect for identifying good words after seeing houses. One possible explanation for this difference is that humans have a specialized mechanism for identifying attractive faces — that we’re slower at judging houses than faces.

Regardless of the cause, evidence abounds that attractive people receive preferential treatment — they’re more likely to be hired, more likely to be viewed as intelligent, more likely to be seen as good teachers. And the judgement that makes all this preferential treatment possible can be made in, literally, the blink of an eye.

Olson, I.R., & Marshuetz, C. (2005). Facial attractiveness is appraised in a glance. Emotion, 5(4), 498-502.


  1. #1 coturnix
    April 10, 2006

    Now, there are some obvious follow-up questions, but this is not an X-rated blog…

  2. #2 staffpsy
    April 10, 2006

    physical attractiveness is one of my areas of interest. this is a great little example of how quickly we make those judgements, which in turn activate a whole range of other networks of information and things we attribute to attractive people. good stuff.

  3. #3 Lab Cat
    April 10, 2006

    Those time lapsed clips were very interesting. I particularly liked the animated nerve cells.

  4. #4 mamagaea
    April 11, 2006

    I saw a younger man, shorter brown hair with glasses, and Bugs Bunny

  5. #5 outeast
    April 11, 2006

    Hmm… Interesting, but if anything I would have expected the impressions to be clearer – 4.7 and 5.7 seem pretty median to me, especially when extremes of attractiveness were used. I’d guess that the detection of just a very few key signifiers – eye and hair colour? weight? facial regularity? – would be all that would be needed to tip the scale so slightly to one side or the other of the median.

  6. #6 Kapitano
    April 11, 2006

    A few thoughts:
    * Do heterosexual men judge the attractiveness of other men as rapidly as they judge women?
    * Do prepubescent children or adults with low sex drive show the same pattern?
    * If a white person who admits to being racist is shown an masked attractive black face, is the racism ‘bypassed’?

  7. #7 Dave Munger
    April 11, 2006

    Kapitano –

    I can’t answer your second two questions (more research is needed!), but with regards to the first, the participants in this study were men and women, and they rated both men and women, with no gender differences: men rated men just as rapidly as women, and vice versa.

  8. #8 Dave Munger
    April 11, 2006

    Mamagaea –

    Yes, unfortunately I think I really didn’t duplicate the stimuli very effectively at all. The actual study used the refresh rate of the screen as the duration of viewing the face — literally as fast as the screen could update itself. I can specify a 10-millisecond interval in an animated GIF, but I’m certain the image is actually visible longer than that. I guess I need to get one of those high-speed cameras from to find out exactly how long my image was displayed, but it’s clearly considerably longer than 10 milliseconds. If I had to guess, I’d say it was more like 100 milliseconds.

  9. #9 encephalin
    April 11, 2006

    I remember an experiment where infants were shown images of attractive versus not-so-attractive faces. The infants turned their faces more frequently to the attractive faces, and looked at them longer…

  10. #10 Edward Hubbard
    April 16, 2006

    One possible explanation here: It has been shown that the amygdala (and perhaps via the amygdala, the orbitofrontal cortex) is sensitive to the overall, low spatial frequency structure of objects. Now, given that symmetrical faces are more likely to be rated as attractive, it is quite probable that the low spatial frequency content of the attractive faces were different from the unattractive faces in terms of this low spatial frequency content. This is likely to be what is getting through the unconscious processing pathway, to prime the subsequent lexical decisions. This is consistent with encephalin’s post, given that, for example, Mark Johnson suggests that early neonatal tracking of faces is mediated by subcortical mechanisms.