I've only read your first paragraph but isn't that exactly what Samuel's checker program did by revising regression coefficients as it gained experience. We're talking late 1960s.
--- Frank C. Wimberly 140 Calle Ojo Feliz, Santa Fe, NM 87505 505 670-9918 Santa Fe, NM On Tue, Jun 16, 2020, 2:05 PM glen∉ℂ <[email protected]> wrote: > Right. What I set up was in preparation for an argument about exactly > that. Can a system (any system we know about) be programmed to factor in > its experience so that the next time around, the probabilities will be > different, even if only slightly so? Personally, I could go either way. FW) > I can see a situation where, immediately after some branch was taken, the > memory structure would dampen/lower (or raise) the chances of that branch > being taken again. NFW) Or, alternatively, maybe each situation is so > concrete, so forcibly contextual, that there is no such thing as "coming > around again". In the former, "free will" exists in the form of > successively modified ("deliberate") behavior [†]. In the latter, it > doesn't. I'm sure there are other ways to make the argument either way. > This argument boils down to pattern recognition, similarity between > "traces", approximation, and truncation. > > I'm sure it's not obvious how/if the (FW) case fits the typical > understanding of free will [‡]. But I think I can make the argument that > the scopes/degrees of the branch-points (including the speed of the events, > size of the clusters of events, etc.) suggest whether it falls under what > we'd normally call "free will". Scope that is too small/fast (biochemistry > up to limbic system) is below the threshold. Scope that is too large (being > reared in a society that forces some behavior like eating meat) is above > the threshold. But somewhere in between might be an adaptive trend that > kinda-sorta fits our usual understanding. > > > [†] I think this is distinct from, though related to, the concept of > _learning_ or entrainment. I think there's a sweet spot in between ignorant > and enslaved that we target with our concept of free will. > > [‡] One of the phenomena this setup could help test is the idea that "you > never know what you'll do until you're in that situation." I.e. the first > time you experience something (like a fist fight, or a hit of whiskey, or > whatever), there can be no free will. The 2nd time, maybe. The 100th time, > for sure. But the common understanding is that the "decision" is made 100 > times. This setup violates the vernacular in that the "decision" is smeared > out through the nearly-repeated experiences. But at some point, you fall > out of the "free will zone". After 50,000 glasses of whiskey, we might say > you no longer have free will. You're a slave to your addiction. > > Another phenomenon this setup might help think about is whether *some* > machines have free will but others don't. E.g. if the components that > remember and adjust the probabilities for the next time around are damaged, > the machine can't "deliberate" like it normally would ... or the free will > zone (event/process scopes in the sweet spot) might be shorter or longer. > > On 6/16/20 11:28 AM, Marcus Daniels wrote: > > But you also gamed this proposition: > > > > < That memory of lost opportunities is what we call free will. > > > > > Many people apparently believe they can defy their programming and think > it is reasonable to expect people to do the same. But punishing the sin > and the sinner are the same, and it only matters if the "trace" ever can be > exercised again. > > - .... . -..-. . -. -.. -..-. .. ... -..-. .... . .-. . > FRIAM Applied Complexity Group listserv > Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam > un/subscribe http://redfish.com/mailman/listinfo/friam_redfish.com > archives: http://friam.471366.n2.nabble.com/ > FRIAM-COMIC <http://friam.471366.n2.nabble.com/FRIAM-COMIC> > http://friam-comic.blogspot.com/ >
- .... . -..-. . -. -.. -..-. .. ... -..-. .... . .-. . FRIAM Applied Complexity Group listserv Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam un/subscribe http://redfish.com/mailman/listinfo/friam_redfish.com archives: http://friam.471366.n2.nabble.com/ FRIAM-COMIC http://friam-comic.blogspot.com/
