Facial Recognition

If you read the post about Geons and Irving Biederman, you might have thought to yourself that visual perception must be more subtle and nuanced than the discussed Recognition by Components (RBC) theory suggests – and you’d be right. RBC does a very good job of describing how we identify object types, but it does not explain our ability to distinguish more qualitative differences.

Biederman’s research does in fact describe not one but three different processes by which we recognize objects. The first, as described by RBC theory, works well for shape-based objects whose parts are differentiated and whose edges can be readily identified. But there are two special cases where this mode of identification just doesn’t work. The first is faces, and the second is textures. We can distinguish between individual faces readily, even when they are very similar. With textures, we can readily identify the skin of an orange, a honed slate surface or a sheet of steel even without visible edges, vertices or any shape-based cues at all.

Faces present a special problem for the brain in two ways. First, faces have the same parts in the same places, and in the case of similar people they may be in almost the same relationship to one another. Second, it is essential that we are able to distinguish between people of similar appearance rapidly and unmistakably – who is friend, and who is foe. If RBC theory does not adequately explain this, then what does?

According to Biederman, we use a very different and very specialized approach for the recognition of faces. Geons help get us started by allowing us to ‘lock on’ to the basic shape and spatial orientation of the face using reference points such as the corners of eyes, but from there we apply a much more holistic approach in which we imprint the memory of the overall face as a remembered three dimensional contour, and compare it to our stored memories of the contours of other faces, until we find a match.

With textures, a visual imprinting of each surface texture may be stored and recalled in a similar way. Indeed, the texture mapping of the tone and texture of a person’s face can augment our imprinted memory of that person’s face, and in this way aid in facial recognition.

RBC processing takes place in an entirely different part of the brain than face and texture processing. For this reason a given person can have greatly differing levels of ability in these two skills; in unusual cases of brain injury, it is even possible for one ability to remain unaffected while the other is severely limited or even absent.

It is interesting to contemplate how these recent scientific insights into how we perform recognition tasks may add a new dimension to the way we think about objects and design. Each of our intuitive responses to places, spaces and things must undoubtedly be influenced by our relative strengths in these different recognition processes, and by the unique internal database of memory maps that each of us develop during our lifetime.

Image Credit: Cydonia Region of Mars, NASA
Learn more about the face on Mars here

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s