Once, not so long ago, I was testifying about perception and memory in Anchorage, Alaska. An eyewitness claimed to have recognized some individuals (the defendants in a murder trial) when they were 500 feet away. I pointed out that such recognition would be impossible: It would be akin to sitting in high in an end zone seat of a large football stadium and recognizing someone sitting high in the opposite end zone.
Based on this experience, I initiated a project designed to (1) determine what kind of information is lost when a witness views another person from varying distances and (2) implement the result of this project as an image-processing program such that one could enter a photograph of a person and a distance„which would then produce a degraded version of the original photograph corresponding to the information available to the visual system when the individual depicted in the photograph was viewed from the specified distance.
I had believed that the first part of this project would be easy. The plan rested on the following logic. First, the visual world is made up of a continuum of spatial frequencies: Very roughly speaking, "large blobby things" are composed of low spatial frequencies, while "small detailed things" are made up of high spatial frequencies. The visual system, like any optical device, processes different spatial frequencies with differing efficiencies: Again, roughly speaking, higher spatial frequencies, are not processed as well as lower spatial frequencies, and above some limit, high spatial frequencies cannot be processed at all.
As distance from a witness to some object (such as a person) increases, geometry dictates that features composed of low spatial frequencies when the person is close, become composed of higher spatial frequencies as the person moves further away. This means that as distance increases, the details of the viewed person disappear, and essentially, the witness's image of the person becomes blurred.
Although this plan was straightforward, it turns out to be more difficult than I had originally imagined to pin down the exact quantitative means by which the visual system perceives different spatial frequencies with different efficiencies. As a result, I've been doing some experiments to investigate this issue. In my talk, I will describe the problem sketched above in more detail, and I will provide some preliminary data.