[FRIAM] naive question
Marcus Daniels
marcus at snoutfarm.com
Thu Oct 20 23:21:05 EDT 2022
Some enzymes can be very hard to simulate since the reactions they catalyze involve quantum mechanical effects. Very hard is different from impossible, though, and enzyme/substrate interactions are studied using Density Functional Theory simulations, and higher fidelity methods. These atomistic-scale methods run on Turing machine-like computers, or maybe someday quantum computers.
One way in which 3-d biochemical systems have functional differences from computer programs is nuanced context dependence. We write computer programs to be comprehensible to other people and so we focus attention on a small number of variables at a time. Nature doesn't care about what humans can understand, so this is not a consideration for how biological systems work.
To me, algorithms that are friendly to human cognition as components of models of biological functional behaviors are just zeroth order hypotheses to test. Understanding that the heart's functional role is to circulate blood is different from understanding how the electrical signaling of the heart works.
Marcus
________________________________
From: Friam <friam-bounces at redfish.com> on behalf of Prof David West <profwest at fastmail.fm>
Sent: Thursday, October 20, 2022 2:27 PM
To: friam at redfish.com <friam at redfish.com>
Subject: [FRIAM] naive question
Marcus and glen (and others on occasion) have posted frequently on the "algorithmic "equivalent" of [some feature] of consciousness, human emotion, etc.
I am always confronted with the question of of "how equivalent?" I am almost certain that they are not saying anything close to absolute equivalence - i.e., that the brain/mind is executing the same algorithm albeit in, perhaps, a different programming language. But, are their assertions meant to be "analogous to," "a metaphor for," or some other semi/pseudo equivalence?
Perhaps all that is being said is we have two black boxes into which we put the same inputs and arrive at the same outputs. Voila! We expose the contents of one black box, an algorithm executing on silicon. From that we conclude it does not matter what is happening inside the other black box—whatever it is, our, now, white box is an 'equivalent'.
Put another way: If I have two objects, A and B, each with an (ir)regular edge. in this case the irregular edge of A is an inverse match to that of B—when put together there are no gaps between the two edges. They "fit."
Assume that A and B have some means to detect if they "fit" together. I can think of algorithms that could determine fit, a simplistic iteration across all points to see if there was a gap between it and its neighbor, to some kind of collision detection.
Is it the case that whatever means used by A and B to detect fit, it is merely the equivalent of such an algorithm?
The roots of this question go back to my first two published papers, in AI Magazine (then the 'journal of record' for AI research); one critical of the computational metaphor, the second a set of alternative metaphors of mind. An excerpt relevant to the above example of fit.
Tactilizing Processor
Conrad draws his inspiration from the ability of an enzyme to combine with a substrate on the basis of the physical congruency of their respective shapes (topography). This is a generalized version of the lock-and-key mechanism as the hormone-receptor matching discussed by Bergland. When the topographic shape of an enzyme (hormone) matches that of a substrate (receptor), a simple recognize- by-touch mechanism (like two pieces of a puzzle fitting together) allows a simple decision, binary state change, or process to take place, hence the label “tactilizing processor.”
Hormones and enzymes, probably/possibly, lack the ability to compute (execute algorithms), so, at most, the black box equivalence might be used here.
[BTW, tactilizing processors were built, but were extremely slow (speed of chemical reactions) but had some advantages derived from parallelism. Similar 'shape matching' computation was explored in DNA computing as well.]
My interest in the issue is the (naive) question about how our understanding of mind/consciousness is fatally impeded by putting all our research eggs into the simplistic 'algorithm box'?
It seems to me that we have the CS/AI/ML equivalent of the quantum physics world where everyone is told to "shut up and compute" instead of actually trying to understand the domain and the theory.
davew
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20221021/b6099338/attachment.html>
More information about the Friam
mailing list