Threads

Moral intuitions as symmetry intuitions, and reconciling epistemic frames

4 tweets · August 2024 · 33 likes · 2 retweets · read on Twitter

"moral intuitions are generally symmetry intuitions, so fixing asymmetries in your direct phenomenology should allow much clearer direct perception of things likely to be perceived as moral/immoral" — @RomeoStevens76

@RomeoStevens76 "In practice," Romeo adds, "there are competing symmetry intuitions, hence [waking up doesn't automatically make you a better person]. But it should help quite a bit on the margin."

RomeoStevens @RomeoStevens76 ·

"Waking up doesn't automatically make you a better person" "Yeah, but, come on, my dude" Figured out a way to crisp up this intuition from watching @WystanTBS interview Jay Garfield. Essentially (ha), moral intuitions are generally symmetry intuitions, so fixing asymmetries

priority: synthesizing competing symmetry intuitions how much investment is going into that? that's actually a really simple way to put some of the most important work I've been doing with my life the past 4 years! eg these conflicting views on trust:

Malcolm Ocean 🏴‍☠️ @Malcolm_Ocean ·

"how am I relating to this?" I tried interviewing a friend about some wisdom/insight she has about being able to choose to trust life, since that nearly directly contradicts my orientation to choice/trust, and she shared this question, which helps you take your stance as object

@RomeoStevens76 the same way that Einstein's insight involved reconciling a symmetry intuition about physical reference frames so that all were valid and none was privileged the trust insight involves doing exactly that for learning systems' (eg ppl's) epistemic frames

Emmett Shear @eshear ·

@AnnaWSalamon @lVlarty @mherreshoff @ESYudkowsky You encounter an object in the wild. What makes you think it's an agent with goals, rather than say a rock? How do you come to the conclusion that any object should be understood as goal-oriented? Well, bc your best inference from observation, given your priors, is that you