pattern recognisers

I’ve repeated the following quote before, and I’ll do it again:

“to understand is to perceive patterns” - Isaiah Berlin

I’m intrigued by this because of working as a programmer, after seeing Jason Silva’s many outputs, and after reading Raymond Kurzweil’s “How To Create A Mind” last year.

What Kurzweil’s does in the book is first he attempts to explain how the brain works and how the mind then emerges from it. Later, he proposes some techniques on how this could be digitally replicated and what philosophical and cultural conundrums that would create. I find both interesting, but to me, the part about what constitutes the mind is more thought-provoking.

connections #

Kurzweil argues that the thinking mind emerges in the neocortex, with it’s approximately 300 million pattern recognisers. The pattern recognisers are then linked with each other, in a hierarchical grid of sorts.

Inputs, both external an internal, are processed within this grid by matching patterns and associations.

The hierarchy consists of information on different conceptual levels, as in curved shape - letter - word - sentence - context - world. This structure is often structured like a tree. As I understand, though, is that the connections need not be tightly structured, but can be made from any level to another. Matching a partial pattern would require imagination, and that is often drawn from combining past experiences and existing ideas.

The pattern recognisers are not just some immutable input/output functions, but provide the actual state of a person. This would mean that a person borns with next to no connections and no hierarchy, a “blank-slate”. The recognisers are then generated via inputs from external experiences and internal pondering. I suppose some are defined by genes, too.

repetition #

Repetition creates duplicates of these recognisers, some with different connections, some with similar. I’d imagine this is why memories formed within unusual connections tend to stick better. This would explain the better memory formation when the learning is combined with exercise or positive stress. Or just memories related to special events. And why telling stories help us remember details better, as they are related to something.

The speed of information processing in the brain is generally slow. At least when compared to a modern computer. Where the brain prevails, is that it does this bunch of relatively slow computations simultaneously. Repetitions of recognisers or connections are then more probable to match a stressor and become the chosen interpretation.

All of this remind me of graph databases. Simplified, data consists of nodes, and the base concept is that connections between any node is possible. Generally, the more connections, the higher the “value” of a node.

visions #

Kurzweil has his critics, and some of his visions do seem crazy. He also goes into general AI, out-of-body pattern recognisers, and of course to the Singularity. Some of his predictions in the past have also appeared silly, but many have still come true.

Whatever the true source of consciousness, thinking about the world as patterns is fascinating.

resources #


Now read this

programming, slow

Software development is generally considered a thinking person’s activity. Word thinking usually implies deliberate pondering and careful reasoning. However, there’s a clear-cut difference between thinking fast and slow (coined Systems 1... Continue →