SkepticblogSkepticblog logo banner

top navigation:

Facial Recognition Culture

by Steven Novella, Jan 16 2012

It is well established that people generally have a well developed ability to recognize human faces. There is a substantial part of the visual cortex dedicated to doing just that, giving us the ability to recognize any of thousands of familiar faces in a fraction of a second. This is a useful skill to have for a social species like Homo sapiens. In addition to being able to recognize individuals, we can also gain information about health, fertility, gender, age, mood, and intention from looking at the face. We can also convey a great deal of non-verbal communication by facial expression alone.

A recent study published in PLOS One looked at the small question of what part of the face do people generally look at when trying to recognize an individual? In particular the authors were exploring the question of cultural differences.

Prior research has shown that Westerners tend to look at each eye then the mouth when sizing up another person. This suggests that they are looking at the details of these individual features. Asians, however, generally look at the center of the face, around the nose, perhaps so that they can take in the spacial relationship among the various facial features. These are two different strategies that can be used for facial recognition. This finding raises at least two questions – are these differences genetic or learned, and is there an adaptive reason for them? The research suggests that it is largely learned (Asians who grew up in the West use a combined Western and Asian style of facial recognition.)

The question of adaptation is difficult to study, but the speculation is that it relates to the difference in feature homogeneity. Western societies tend to have more differences in hair and eye color, for example, while Asian societies tend to have more uniform facial features. Asians, therefore, have to rely upon the spacial relationship among the features, rather than the individual features themselves.

The current study used eye tracking software to look at Malaysian Chinese. They found that, like Western born Asians, they employed a mixed strategy of facial recognition, but one unique to them. They tend to look at the eyes and nose more than the mouth. It is thought that this is also a cultural adaptation to the multicultural society of Malaysia.

The main author said of her study:

“We have shown that Malaysian Chinese adopt a unique looking pattern which differed from both Westerners and mainland Chinese. This combination of Eastern and Western looking patterns proved advantageous for Malaysian Chinese to accurately recognise Chinese and Caucasian faces.”

It’s interesting that we would learn what part of the face to focus on in order to optimize facial recognition. It’s not surprising, however, as we have, as I wrote above, a large part of the cortex dedicated to processing facial information. We tend to focus a great deal of our attention on the face, so it makes sense that we would employ a nuanced and adaptive strategy for facial recognition.

Of course this also explains why we tend to see faces everywhere, including in random patterns. We can even see emotion in the non-existent faces of inanimate objects.

8 Responses to “Facial Recognition Culture”

  1. John K. says:

    How fascinating!

    If facial recognition strategies are indeed learned, I wonder how so? After all, new eye tracking software was required to observe this at all. How could such a normally unobservable thing be taught? Perhaps the facial patterns a person grows up around will affect the type of recognition strategy that is developed?

    Just amazing.

  2. Max says:

    Saudi men must be good at recognizing women just by their eyes.

    People are pretty good at recognizing familiar faces, but not so good at recognizing unfamiliar faces, mistaking folks for celebrities and for crime suspects.

    The Asian strategy sounds closer to what computers do, because the spacial relationship of facial features is harder to change with makeup and plastic surgery.
    The Western strategy sounds more like the old facial composite “potato head” technique, selecting the eyes, then mouth, etc.

    There’s a lot of computer pareidolia, i.e. false detections, in Google Street View images. When you see random round things blurred, it’s because they were mistaken for faces. Blurred text on street signs was mistaken for license plates.

  3. Max says:

    “Of course this also explains why we tend to see faces everywhere, including in random patterns.”

    It’s not obvious that this should be the case. Perfect face recognition would have no false alarms.

    • Old Rockin' Dave says:

      If you think about it, false positives – seeing faces where there are not any, is more adaptive than false negatives. It could be worse to miss seeing a face that is there than to see one where there is not because your social relations would suffer and at worst an enemy or competitor could be upon you before you knew it. Beyond that, identifying an image as a face is one of the earliest visual skills to develop in an infant. They will make a face out of two roughly eyelike objects at first and discriminate between real faces soon after. The ability to recognize first a face, and then single one out as Mama, is part of parent-child bonding and would tend to promote survival, so the tendency to over-identify is likely to be inborn.

      • Max says:

        Makes sense. My point is that a high false alarm rate indicates a lack of nuance, at least in the detection stage.

  4. Max says:

    I wonder if nearsighted Westerners use more of an Asian strategy, since they see less detail.
    Similar to computer facial recognition with low-quality images.

  5. Tom says:

    So, have we found the cause of the “They all look alike” cliche?

  6. Max says:

    I wonder if Westerners looking at Asian faces adopt the Asian recognition strategy.
    http://themoderatevoice.com/wordpress-engine/files//2011/01/chinese_military0305.jpg