Tricked Desires - 2nd order superstimuli

So, we have supernormal stimuli, which act on goals that evolution by natural and sexual selection has for us, specifically, on the desires that make us act on those goals.

But the individual might not care about evolution’s goals, or do so for different reasons. I have some set of diffuse values and e.g. nourishing my body is important to me bc it’s important to satisfying those values and goals I have, but passing my genes on, isn’t the goal/value I have. We are adaptation-executers, not fitness-maximizers¹ after all.

So, I won’t necessarily do the things that evolution “wants” me to do, for which it gave me the desires I have, but pursuing them makes me feel good. Evolution’s goals are not my goals, so goodharting them bits is fine by me!

But I do have my own goals and values. And those have their own desires, which can also get goodharted, can get satisfied in ways that feel good but don’t actually do anything to progress the goal.

2nd Order Superstimuli

The first causes I cared about are the environment and climate, as well as animal welfare. And working on them to make real progress is hard, but there are cute games out there where you clean up the planet (e.g. Clean Up Earth) or play in some floating solarpunk world where you grow your little poly-/permaculture garden, sell your veggies to the local co-op and build PV panels for power decentralization.
I’d love to play them, but NO! They satisfy a desire to create a better world, whilst the only impact they have is drawing power², at best they teach me a little bit on the side.

Then, there’s idle games that can create a feeling of productivity (we are making number go up after all!), without actually producing anything I value.

Series and podcasts might give a sense of community, when you spend a lot of time with the same people, except ofc they’re not actual friends, leading to parasocial relationship satiation³.

Engaging in slacktivism is another wanting to do good satisfier, or doomscrolling e.g. climate news might be emotionally hard in a sense, but in itself doesn’t do anything⁴.

When is this fine?

I’m not yet sure to what extend doing this is fine.
This seems outright bad in the “regular” case, where the same drive to do smth could either accumulate and build-up pressure until it’s directed to take real action on a goal or is satisfied pretty much immediately by doing things that feel good but are at best barely meaningful, at worst actively harming the goal.

But if we add more complexity, the evaluation also becomes more complex:

• If one’s motivation system is somewhat destroyed and even large amounts of loneliness or say climate despair don’t motivate action, then engaging in these things might even be beneficial if it has one regain energy, allowing one to act on the goal/value better later.
• Or if I have too many things I care about, I don’t want to feel bad by not being able to satisfy them all? To some extent, it’s probably good to take these feelings and see whether there really isn’t anything that could be done differently.
  • Maybe, upon closer inspection, there are pareto improvements for all values? How would we be sure we’ve exhausted the search space? Here, this tricking desires might be bad.
  • But assuming we did, we either already reached the maximum we can do or there are only net improvements where some goals have to be satisfied less for others to be satisfied more. Perhaps this is hard emotionally (since there might be divergence btwn the emotional salience and how much we value smth) and then tricking some of one’s own desires for some of those values allows one to put more effort into those goals that matter more to one.

---

So, while I might happily goodhart my evolutionary desires for fun and profit, it’s important to pay attention to superstimuli for desires stemming from my own goals and values. How much to avoid this in each case is tricky, in the general case this seems bad, but e.g. during burnout (or to prevent it) or value–salience conflict make this useful.

Footnotes

1. see LW, original quote:

   “Individual organisms are best thought of as adaptation-executers rather than as fitness-maximizers.” —John Tooby and Leda Cosmides, The Psychological Foundations of Culture.

   ↩

2. Andy Masley might add:
   if we assume a mid-range gaming PC is used, running a game might take 300-500W, or 300-500Wh for a short gaming session.
   or even if we assume a small 13” macbook air (m5) we might be at around 50Wh
   one prompt is 0.5Wh for training + inference
   that’s 100x less for the macbook and 600-1000x less for the mid-range desktop ↩

3. that’s not to say that those can’t do some work: modeling social norms, providing a safe place to mentally rest at (although content probably really isn’t optimized for this or might even be optimizing away from it?), but the ratio is still bad. ↩

4. here, there is probably some balance where enough news can motivate action, but too much doom and gloom can be demotivating, which is what I’m pointing at. ↩