How does dopamine work? I need to re-source this info, I think it was from a Google Deepmind article.

Brain log for 2025-03-26.

So let’s say for sake of ease that dopamine has a -5 to +5 scale. Minus 5 is the worst case scenario. +5 is the best possible outcome. Let’s also say for now that everyone’s dopamine baseline is a zero. If they were sitting on a couch staring into space, it’s zero.

You might think, as I did, that doing things results in an increase or decrease of dopamine that might fit into our system but it apparently (again, this needs sourcing) it’s more an anticipation and result system.

Let’s say you’re going to head out to the local shop. You predict a result of +2 because you get up and about and you’re gonna grab the stuff you need. Then when you’re at the shop you decide to treat yourself to a cheeky bag of Maltesers. By the time you’re home your brain notes the reward value was actually a +4 because “mmm, sugar”

Your brain bumps up your dopamine by the difference for a bit, +2, you feel happy. You’ll be wanting to do that again in future.

Next time you think about going to the shop you might think “I’ll grab myself some cheeky maltesers again, treat myself”. You anticipate the visit to the shop to be a +4.

When you get to the shop there are no Maltesers.. well that sucks. You grab whatever else you need but are short your sugary treat. For the sake of argument you didn’t think to grab something else.

If all else was the same your normal shop would be a +2 result, but you predicted +4. The difference is a -2. You feel a bit let down. You’ll try to avoid this result in future, maybe in future you’ll have a backup treat in mind just in case.

Anyway yeah so I want to make it so Brain will predict the outcome of its actions before doing them, then note the results after, and feed back into the decision making process the difference. Next time it makes a similar decision it will be able to anticipate a different result based on prior data. It will be able to learn from previous actions.

Of course the system outlined above may be completely inaccurate or far too simplified by my limited brainmatter, haha. Also note that this was just a single predict-result example. An actual person will be doing millions of these a day or whatever. Brain is gonna be a wee bit slower than a human mind for a while..

17:10

I bought some Maltesers