This morning I had a fascinating conversation with my friend Sankar. We get together, more or less monthly, to discuss ancient Indian philosophy over a cup of coffee. At an earlier conversation I recommended Max Tegmark’s Our Mathematical Universe since some of the patterns that Tegmark describes seem similar to some of the philosophical works that Sankar studies.
Today we wandered through some of Tegmark’s thoughts on how we perceive motion. If I’m recalling correctly, we perceive a collection of discrete instances in time, like frames in a movie. Those snapshots are analyzed for their differences. The differences are mapped to a concept we call motion. This mapping to a concept we call motion is a form of knowing if you assert that knowing or understanding something is your ability to map a pattern of sensory impressions onto an internal generalization with minimal discrepancies. The conversation proceeded down the path of realizing that there’s an input filtering, summarizing, dispatching controller layer just behind the senses and a symmetrical controller layer on the action/output side as well. An architecture began to emerge and here’s a sketch…
It turns out there are names for some of these elements in the literature that Sankar studies. Unfortunately, I didn’t retain them from our conversation.
Several very cool things come out of this picture…
- As you can see it provides me with an ability to tease apart and understand the System 1 and System 2 notions introduced to me by Daniel Kahneman’s Thinking, Fast and Slow. Including possible sources and explanations for various forms and variations of availability biases and heuristics.
- It begins to hint a bit about a mechanism for “flow” as described by Mihaly Csikszentmihalyi in Flow; The Psychology of Optimal Experience. Basically what happens when there are few discrepancies to report back to the System 2 Difference Analyzer. Processing is unimpeded from input to output.
- Per conversations I’ve had with my friend Jim, it suggests an interesting architectural pattern for Internet-of-Things based systems. Note that any particular block on the diagram may be a distributed system and the actual functionality is more of an emergent characteristic.
- Sankar and I discussed the notion of death in this context. Basically if any of the processing elements along the System 1 path cease to function then you’re dead. Note that you can become blind or lose a hand and you aren’t dead. They aren’t points of catastrophic failure. You may also be able to lose your difference analyzer, basically System 2, and continue to be alive. This is likely what’s going on when you become a vegetable.
In any case, it was a fun conversation and generated the epiphany of the day. We’ll see how it holds up over time.