The brain only sees a small part of the image in detail at any one time. That spot moves approximately 3 times a second (“saccades”). This is easy to see when you watch someone else’s eyes closely, and initially rather disturbing to notice in yourself.

The eye’s successive regions of interest are presumably dictated from ‘higher up’ in the hierarchy of brain function. It seems reasonable to suppose that the eyes are directed to wander over the most ‘interesting’ areas of the image - those areas that the brain is less sure that it understands (i.e. the brain knows that it has wider error bars on its predictions of a particular area before viewing, implying that it needs to be reviewed most urgently).

Your Internal Screen

It is very tempting to think of the brain as containing an internal screen on which the conscious mind watches the external scene. However, we know for sure that the image on this screen (if the screen exists at all) can only be being built up in successive patches.

Indeed, there is no need to actually have such an image if all that is required is to produce the correct scene at each position that the mind directs its attention to. It is sufficient to convince the higher levels in the brain’s hierarchy that they are correct in their claims of what is being seen.
From a practical point of view, the storage for the current scene needs only to be of the most significant features - pixel data can be discarded as soon as the relevant information is extracted.

Quick Thought Experiment

Question : Think about looking at a face in your mind - do you scan the features of the face in the same way as you would those of a person standing in front of you (with your internal mind’s eye looking from feature to feature)?

Answer : YES - pretty much the same experience as physically looking.

Martin Andrews

{Finance, Software, AI} entrepreneur, living in Singapore with my family.

blog comments powered by Disqus