>I bet a computer can detect those edges with relative ease
It's not so simple. What edges? The computers sees no edges there -- only where they actually exist.
If it was made to "extend" the actually existing edges, where would it stop?
It's not so simple. What edges? The computers sees no edges there -- only where they actually exist.
If it was made to "extend" the actually existing edges, where would it stop?
I am certainly no expert on this matter, but it seems intuitively obvious to me that the reason they haven't had much luck is because they are trying to solve a much harder problem than the brain actually solves.
The brain almost certainly is using input from other "functions", particularly edge detection and large-scale pattern detection to determine that there is a repeating grid there, not just raw color information and inferred scene lighting. I bet a computer can detect those edges with relative ease, but that's not how the researchers are trying to solve this. They probably know that they could do it that way, but consider it cheating because that's not what they wanted to study.