reward is just the temporary relief of punishment
- a draft
- You can't motivate someone to do something with money if they already have enough money or a better way to get more money. You can't motivate someone to do something using praise unless they secretly fear they're not good enough or respected enough.
- Technically "temporary relief of punishment" isn't quite correct. Being in a state of not having enough money isn't inherently a punishment. People tend to psychologically experience it that way, but that's not about the money itself.
- This connects with my remark at the top that most humans are in chronic fear. Reward represents temporary relief of that fear: don't worry, you're okay, you did the right thing—this time. But if you haven't actually dealt with the situation in which that fear arises, the reward doesn't solve your problem, it just relieves it for a period of time.
- Importantly—not all consequences are rewards/punishments. If you're trying to catch the bus and you make it (or miss it) that's not a reward (or punishment). This confusion is common, which is making me think just now that it must be part of the pattern, not a coincidence.
- While I'm at it: not all FEEDBACK is reward/punishment either. Letting someone know that something worked, or felt good for you isn't a reward. Letting them know it has a problem, or felt bad for you isn't a punishment.
- You can choose to interpret everything you experience as reward & punishment, or nothing. Both are coherent.
- (The latter doesn't mean being oblivious to fact "this person is trying to punish me", it means not internalizing and trying to counteract the sensefulness of your own system.)
- I'll write this up at some point but in the meantime you can learn more about it from one of my favorite books, Making Sense of Behavior: The Meaning of Control. You can also read Punished By Rewards for more on how rewards backfire in virtually all situations.
9 linked references
February 28th, 2022
- and my first draft has a section in it on reward is just the temporary relief of punishment
- but as I wrote it I found myself noting that that isn't quite true
- it's a useful pith but technically sort of incorrect
- the main reason it's incorrect is that it implies a punisher
- now in some sense in many cases the punisher is yourself...? hm.
- ANYWAY
- ultimately I think I'm concluding that in order for my point to not get distracted here or over-critiqued by skeptics, I want to just remove this whole section.
- aha! but it's easier I think to point at this wrt "rewarding yourself" (by feeling good or with cookies or something)
February 18th, 2022
draft of my thoughts on ethics / morality / moral realism › D) justification/blame-&-punishment/reward-based systems …
Podcast summaries
- Keywords - cooperation & defection aren't as obvious as they might seem (they depend on frame) IFS collective agency untangle self-authoring social technology is more of a bottleneck than material technology intentionality everyone is always doing whatever makes the most sense to them partswork Coherence Therapy everything makes sense Everyone is basically living in a dream mashup of their current external situation & old emotional meanings self-trust comes from respecting distrust Learning how to trust yourself when all men doubt you, but make allowance for their doubting too safety psychological safety Alfie Kohn punishment is goodharting waiting to happen protectiveness is damage control deliberate "negative reinforcement" is punishment even if you don't call it that intrinsic motivation adulthood An Everyone Culture in general, people don't not-care. they care about many things and don't know how to care for all of them together (PCT) reward is just the temporary relief of punishment scarcity much of human collaboration scales sublinearly but it's possible to scale superlinearly what is culture? how do we make groups with the union of their individual wisdoms rather than the intersection of their individual wisdoms? list of common schemas learned in school School is fake averagerianism post-average
- 20:12 - 37.07 - How to navigate between allowing people make their own mistakes and intervening punishment punishment can't steer deliberate "negative reinforcement" is punishment even if you don't call it that An Everyone Culture in general, people don't not-care. they care about many things and don't know how to care for all of them together (PCT) reward is just the temporary relief of punishment scarcity much of human collaboration scales sublinearly but it's possible to scale superlinearly adulthood
December 8th, 2021
- I suspect if you think about how behavioral is high-dimensional you will realize that guardrails are impractical
- they are somewhat practical in 1-3 dimensions
- but arbitrarily difficult with arbitrarily many dimensions, whereas a single attractor can work no matter how many dimensions there are
- rewards are one form of attractor but they require, in some sense, a sustained neg-reward-state (ie punishment) while in every state other than the attractor, otherwise the reward isn't motivating because the person can meet the need in an easier way (eg if you have enough money, nobody can get you to do something just by paying you—you need another reason to want to do it)
- rewards and punishments both inevitably goodhart
June 15th, 2021
- starting a thread on PCT/reward is just the temporary relief of punishment - tweeted
- Reinforcement learning is backwards because it implicitly says "when reward, behavior change"
when actually, change is the *default*, and the thing being called "reward" says "stop changing, you found it"
This figure-ground inversion is key. - The whole concept here of PCT (Perceptual Control Theory) is that brains are in some state of organization that is set up to produce a set of controlled perceptions, like getting money to buy food to not be hungry, or telling jokes to make people laugh to feel connection.
- When something is perceived to be at an unwanted level (eg hunger, safety, connection, thirst, sex, meaning) the system goes into a state of reorganization, inhibiting old strategies and trying new ones.
Once the error is resolved, the music stops and the organization stays. - This is independent from whether the new strategy is actually the thing that solved the problem (which is where superstitions come from).
- There's also a question of what happens when two strategies for two different needs conflict.
Often this produces oscillatory tug-of-war dynamics, but if you can gently hold all of the desiderata at once, there's usually some sort of win-win to be found.
twitter.com/nickcammarata/status/1403558910984966150 - One important implication of this whole thing is that you can only motivate/train someone with reward *to the extent that they are otherwise deprived of the thing* and
April 8th, 2021
- reward is just the temporary relief of punishment - not my original thought - comes straight out of PCT