Part II of 4: Part I, Part III, Part IV .
A few weeks ago, the Mungers at Cognitive Daily explored the question, "Why is it that people seem to see faces so many places that there aren't actually faces?" They discussed the work of Pawan Sinha on developing computer models for face recognition.
It seems that we need very few clues to identify a face. (I would even extend that to a body part; a foot extending beyond a parked car in the lot, for example. Often, on the basis of that alone, we deduce age bracket, sex, direction, and decide whether or not to brake.) Dr. Sinha was experimenting with photos of 12 by 15 pixels only; Cognitive Daily went on to do an online survey with photos down to 6 by 7 pixels.
Go look them over and test yourself on their photos, here, here and here: you will be amazed!
Hint: many of these photos are nothing but a blur from close up. But walk across the room, and look. Even though they are much tinier at that distance, they usually resolve perfectly into clear images.
Does this have a bearing on why we see faces in random patterns?
And here are another couple of photos where we see faces. A puddle on the roadway and a rotting stump in Watershed Park.
(Laurie sees different faces than the ones I see. What do you see? Walk across the room; do the faces appear or change?)