Decision Making
Thursday, August 19th, 2004 11:53 amBack when I was able to keep up with LiveJournal*, I wrote about fascination with Weakness of Will. It just struck me that a good model for willpower might be a neural net. In a simplified example, suppose you have a node "Sleep with this person" with two inputs: "Desire to get off" and "Desire to respect self in morning." Each input has a fixed (though it can change over time) weight. Perhaps the horny weight is 1 and the self-respect weight is -17. In that situation, the horny input would have to be 17 times as strong as the self-respect input (assuming a threshold at 0). The key to behaving virtuously therefore becomes training your decision-making neural net to the right weights. No homunculus needed.
(I suspect others have written extensively about this, but it just crystalized in my mind, so I thought I'd share.)
* When I haven't been working long hours, I've been hanging out at my girlfriend's house, and she doesn't have Internet access. I have therefore kept only rudimentary tabs on my friends page and haven't had opportunities to post. For the curious: Dragonfest was great. I then spent 24 hours from DIA dropoff to DIA pickup for a meeting in Riverside, CA which was also great. In a little over a week I'll be on my way to Burning Man, which should also be great. That's my month of August in one paragraph.
(I suspect others have written extensively about this, but it just crystalized in my mind, so I thought I'd share.)
* When I haven't been working long hours, I've been hanging out at my girlfriend's house, and she doesn't have Internet access. I have therefore kept only rudimentary tabs on my friends page and haven't had opportunities to post. For the curious: Dragonfest was great. I then spent 24 hours from DIA dropoff to DIA pickup for a meeting in Riverside, CA which was also great. In a little over a week I'll be on my way to Burning Man, which should also be great. That's my month of August in one paragraph.