How we see the world
I always imagined my eyes send a full color image/movie to my brain, letting the brain sort out the what happens in the world. The brain has a lot of processing power but how can it decipher what it sees so fast? The failures of first DARPA Grand Challenge shows just how difficult it is to interpret visual data in real time.
It turns out that the brain does not accomplish the task by itself. It gets a hand by the eyes which pre-processes the data in a way I never imagined.
Frank Werblin and Botond Roska explain the results of their research in The Movies in our eyes article in Scientific American. It turns out that the eye, of a rabbit in their tests, does not send one stream of information but 12 parallel "movies" of the world. This movie illustrates very well the different (simulated) signals sent to the brain:
Different layers of cells in the retina interpret the world and send separate “movies“. The orange movie shows the contours which is great for pattern matching whereas the beige movie focuses on bright areas. The movies are then combined by the brain to give the world we see.
Another key point is that they eyes do not send a continuous movies, it only sends highlights thereby reducing the work the brain has to do. Flashing a light for one second at a specific area in the retina does not send a continuous signal for one second. The retina sends a signal shortly after the bright flash starts, then it remains silent until there is a short signal when the light goes off again.
A great example of solving a complex problem by distributing the work and breaking it down in the manageable pieces.
Yet another lesson learned from mother nature.
No comments:
Post a Comment