Negotiating With Yourself [Transcript]post by orthonormal, jacobjacob · 2020-06-26T23:55:15.638Z · score: 22 (4 votes) · LW · GW · None comments
orthonormal: So, I'm doing a generalisation of the post that was curated [LW · GW] and this post is sort of an elaboration of what Vaniver talked about [LW · GW]. If you notice that there are differences between your private intuitions, and what you can publicly acknowledge, this is a system fast versus system slow thing-
orthonormal: This is a question of negotiating with yourself. I'm going to present a model and talk about some consequences, but I’ll start with a question: why do people in our sphere tend to burn out or go nuts?
orthonormal: This is a pretty important question. I’ll use an analogy many of us have heard before — the elephant and the rider. The conscious mind is the rider and the elephant is the unconscious mind. The rider wants to get somewhere, but the elephant has its own preferences about what happens.
orthonormal: Some features of this analogy I think are true and useful for minds, for humans, is that the elephant has these immediate preferences and some longer term needs, just like we have subconscious desires and subconscious needs. The rider has their own preferences and some carrots and sticks, but the real advantage is having a map. The elephant can just completely ignore the rider if its preferences are strong enough. So how this connects to being human is that our subconscious has these desires and needs and fears, and our consciousness may have a little bit of willpower but what it really has is strategy and planning, the ability to pick out a path so that the elephant won't want to deviate from it too much. If you go right by a river, the elephant is going to want to drink. So, if you don't want to stop for that, don't go by the river right now.
orthonormal: Finally, the subconscious, the elephant, can just overwhelm you in two ways: one of which is it controls your motivation, so it can burn you out or get you depressed if you're trying to defy it too much. And the second is, it can induce bias.
orthonormal: This is the subject of The Elephant in the Brain by Robin Hanson and Kevin Simler, claiming that a lot of motivated reasoning comes when the elephant wants something, the rider doesn't, and the elephant changes the rider's cognition to make the rider feel like it wants that thing (for noble reasons, of course).
orthonormal: This would be very bad. Depression is its own thing, but it doesn't change your way of thinking about the world. It doesn't make you go crazy. Going crazy is really bad. Citation — don't need it in this community.
orthonormal: What can you do about this? There are a couple of things. First: you can keep the elephant happy. You can choose a path along the map so that the elephant will be reasonably well fed, have enough to drink, not get tired going up and down mountains, etc. And you can still get to a place you'd like to go. Maybe not the place that's absolute best, but good enough.
orthonormal: This is analogous to a lot of things in Effective Altruism where I'm telling people, “give yourself permission to be happy”. Don't take a job that's going to make you miserable just because you think it is the best thing to do. Find something that meets you in the middle. I don't recommend living on minimum wage and giving away everything else to charity because you're going to burn out from that, or you're going to come up with some crazy reason why doing something else is better. So just let yourself be happy. 80/20 things.
orthonormal: The second thing is about positive versus negative reinforcement. I mentioned carrots versus sticks earlier, and this is really good for also keeping the elephant happy and keeping the elephant liking the rider. There's a wonderful book called Don't Shoot the Dog, which is primarily about animal training, but also about interacting with people — and even about interacting with yourself. It talks about achieve things in animal training by rewarding the animal or by punishing the animal. Rewarding the animal, you can get them to do great things. Punishing, you can get them to do some things... but they'll also just want to avoid the trainer. You don't want your subconscious mind to want to avoid your thoughts. It'll make it even harder to find out what's going on with your desires.
orthonormal: Finally, real quick, treat the elephant with respect, even if you disagree with it. It's really important for you to be able to say, not "Your desires are wrong", but "I understand why you want that, I want this other thing, let's find common ground." And I think those are some of the really important lessons about the elephant and the rider.
orthonormal: Thank you.
Ben Pace: Cool. Thank you very much, orthonormal.
Ben Pace: I like the emphasis you made on having a respectful dialog with the elephant. You spoke about making the elephant happy. I understand the point you're making. But often my relationship with my elephant, when I try to have an internal dialog, is more about asking what it wants and making a commitment to getting it that thing. And those things are not necessarily happiness. They're sometimes respect, or status, or just commitments to find time for the elephant to do the things it wants, whilst also making agreements to work what the rider wants. Some of those motives are not directly about immediately pleasurable experiences. So I always make that distinction.
Ben Pace: Abram has a question.
Abram Demski: Yeah, sometimes I've heard this advice that you should identify with the elephant instead of the rider. It's also a diversity question, you're speaking to people who identify with the rider rather than the elephant but some people identify more with the elephant — or so I've heard. One part of it is normative. Like, maybe we should identify with the elephant instead of the rider? So the question is: what do you have against that, if anything?
orthonormal: It would be nice to be unified, but one thing I think is true is that the rider is good at language and the elephant is not. So the part of you that just asked me that question is the rider.
Abram Demski: I guess I have this drug experience where I was high and I completely separated my consciousness and my audio loop. So, my inner dialog did not feel conscious and instead, I felt like I was the consciousness that the inner dialog is talking about. Which doesn't change my day-to-day thinking that much but makes me able to take that framework where it's like words are coming out of my mouth from this thing that's looking at my actual conscious experience. But my conscious experience is not this thing. I don't have conscious access to... Compare how people know grammar without having explicit knowledge of grammar [Editor’s note: source]. So it's like there's this grammar thing here that somehow knows grammar and it's looking at my conscious experience and producing words that try to describe my conscious experience but that doesn't mean that my conscious experience is the thing that's... You're sort of not talking to my words, my words are like a special case module. You're interacting with my conscious experience indirectly through my words, but my words are kind of dumb.
orthonormal: This is just a hard thing to talk about. I very much believe your experiences and I very much believe that there is something to, through meditation or drugs or whatever, getting more in touch with the non-verbal part of you and having more compassion and connection to that. It's just very complicated to describe in words what that looks and feels like, for obvious reasons.
Abram Demski: Yeah.
Ben Pace: Thanks, Abram, I appreciate that way of thinking about yourself. I think I will probably meditate on that some more afterwards.
Ben Pace: Kamil, do you want to ask a question?
Kamil: Yeah I think that this concept looks like internal double-crux [LW · GW], and if so, my question is, maybe there would be some more sub-personalities, more than just elephant and rider — maybe, some other decision makers in our mind?
orthonormal: Absolutely. The elephant/rider is an extremely simplified version of things. Personally I like the internal family systems [LW · GW] approach to understanding myself. Again, all of these are metaphor, but metaphors can be very useful. The internal family systems metaphors treat different desires and feelings as different agents, more or less, that can talk to each other. So, whatever metaphor works well for people, I encourage them to use that while being aware that it's a metaphor, and also to experiment with other ways of thinking about themselves.
Ben Pace: Cool. Thanks, Kamil, does that sound good to you or do you want to follow up?
Kamil: Yeah, thanks. If so, what are the constraints of this model? Of the model of the elephant and the rider?
orthonormal: Right. The fundamental constraint for my metaphor, at first, is that the conscious part of the mind, which for me includes the verbal part of the mind, is just less strong than everything else that happens, whether that everything else is unified or an aggregate of other parts.
Comments sorted by top scores.