A Developmental Framework for Rationality

[A way of looking at rationality as a set of transitioning worldviews that operate under different assumptions. The short summary of progression looks like techniques → habits → introspection → feelings → multiple ontologies. Skills which appear to be at odds with one another can be seen as operating under different models of what the question of rationality is even about.]

I started mindlevelup with the goal of figuring out self-improvement.

Over 2 years later and over 100 (!) essays later, I’m still at it.

Looking back, one thing that stands out is how I can identify distinct shifts in the ways I’ve approached this problem. The concepts I’ve written about, they haven’t all been operating on the same level of abstraction or assumptions. Some are concrete techniques, while others are overviews of mechanisms, still others are generalizations about the process of learning the techniques, and there’s even a few musing about the nature of self-improvement as a whole.

I’d like to outline what I think is a plausible way that someone’s thoughts might evolve as they approach this problem, based off my own experiences. I think this will be useful in identifying what the underlying models behind different rationality skills looks like. In particular, I’ll be using this to give additional context for where I think a lot of my own essays slot into the worldviews I’ll be introducing.

A] Techniques Rule:

You’re on a quest to find the One True Rationality.

When you first encounter rationality, it’s in the context of cognitive biases, pesky glitches which lead us down suboptimal paths when we can clearly see a better way forward.

For example:

When we say we’re going to finish something in an hour, it’ll really take us more than that. Somehow, projects basically never get finished on schedule, even when everyone knows this sort of thing happens all the time!
Lots of say that we want to exercise, and somehow only a few of us actually get it done. This disparity exists for just about anything else, from reading more books, writing more code, or meeting new people!
People on different sides of a topic will get into a room to share their opinions. And, yet, when both sides leave, they end up more sure of their original opinion then before they began to argue!

Given these observations, you look at self-improvement as one of overriding your body’s silly built-in defaults. The right thing to do, then, is finding algorithms that can outperform your current ones, and then doing those instead. Under this framework, techniques are very appealing.

They’re magic spells you can plug-and-chug to immediately swap up your defaults—just chant, cast, and optimize!

Things I’ve written which fall under this viewpoint:

Fighting Procrastination is all about different ways to try and deal with procrastination. Precommitment was a big thing that I recommended in that essay which I think is quite appealing under a technique-based worldview. It’s simple to describe and has all the hallmarks of enforcing your “rational” desires onto your “irrational” self.
IAT and CEV is about a way of querying other parts of you to see what actions are “actually” best for you. I think it falls under this mindset when you use it as a tool to accommodate for things hyperbolic discounting, e.g. asking yourself “Would I really reflectively endorse this action?”
Reference Class Forecasting, Murphyjitsu, and Back-planning, the three techniques featured in the Planning 101 primer very much espouse this view of “All you have to do is <this other thing> rather than <naive thing you already do> and then things will magically get better!”

There’s a definite sense of wrangling with yourself here, of saying things like “Stupid body! I clearly know what the correct thing to do is, and you’re getting in my way!”

This is also characterized by a sense of wanting to collect techniques. Given the initial problem specs of overriding defaults, it feels like you need a skill to counter each bias you have. Each countermeasure you learn about feels like a little boost up. You hunger for more spells.

And yet, despite having read up about all of these great spells, you can’t seem to recall the right incantations and wand movements to call upon the right one in the right moment…

B] Building Automaticity:

The problem, you reason, isn’t just about beating biases,

“Just do X instead of Y” isn’t a viable strategy if there isn’t a way to actually get yourself to do X at the right moments. But it seems quite difficult to be intentional all the time, to be able to reflectively say, “Ah, and now is the right time to cast my spell Premorta, which will show me an instance of the future where things have gone terribly wrong!”

That just won’t do.

Thus you decide to become a cyborg.

What matters, after all, you reason, is just that the desired algorithm gets executed at the right time. You turn to habits, towards installing the aforementioned algorithms into yourself. There’s a subtle shift in focus that comes in. Whereas you previously thought about techniques in terms of nullifying your biases, the techniques themselves come into full focus here: It’s about diving deep into the question of what it “really” means to practice a rationality skill.

Key assumptions here are that humans are stimulus-response machines, and that any rationality technique can be turned into a habit. One big insight at this stage is the distinction between declarative and procedural knowledge—it’s very possible for there to be an explanation of a rationality technique without giving any actual good information on how to implement said technique in real life.

Things I’ve written which fall under this viewpoint:

TAPs are the core technique here which bring everything together. TAPs are the ultimate meta-skill; here they exist as a proceduralized process that allows you to learn other proceduralized skills.
Hunting for Practicality has, as its core message the idea that you should be operationalizing, translating any advice you have into its most actionable form by asking yourself “How do I see myself actually acting differently as a result of <whatever information I just got> in the future?” It’s all about specificity and actionability.
There Is No Akrasia rejects the idea of akrasia as an all-encompassing bogeyman which causes all the productivity problems. Instead, it pushes you to look directly at what’s going wrong and come up with a specific fix for that thing, whatever it is.

Armed with your Hammer of Reductionism and the powerful welding power of TAPs, you get to work breaking things down and installing them. Things go well. You’re not just all talk anymore—you can point to specific instances where you have indeed performed better.

And yet, it’s not smooth sailing. Some of the software you install quickly becomes obsolete and stops working. Other times, your legacy code—your old defaults—seem to interact in quite chaotic ways with all of these new things you’re installing. Maybe, you think, it’s time to take another peek at that massive codebase…

C] Going Mental

You’re back here again.

You had one look at the codebase, and it was a total mess. No documentation and a bunch of hard-coded variables.

So, against all hope, you wonder if maybe you can gain some more insights by looking at exactly what’s going wrong with the installation process itself. No in-depth internals analysis needed, thank you very much.

You turn your attention inwards, on the practice, rather than what’s being done behind the scenes. By paying attention to the sensations of engaging in rationality, though, you realize that focusing only on execution is futile: The phenomenon of learning skills is still a mental one, no matter how much you’d like to abstract away the messiness of the mind and focus on implementation.

You can’t remove your mind from the equation.

Things I’ve written which explore this foray into the necessity of the internal experience of learning rationality:

Fading Novelty is about how any skill we try to learn is going to be subject to our brain’s tendency to eventually become accustomed to the newness of said skill. Thus, we should expect initial high interest in a skill (perhaps representative of someone seeing a new technique) which eventually fades as time goes on (perhaps to find yet another skill to learn).
Conceptual Similarity Does Not Imply Actionable Similarity looks at the challenges that can come up when our brains substitute an easier question for a harder one because it thinks the answer to both is the same. Concretely, someone might immediately dismiss advice that sounds “obvious” as it’s easier to answer the question of “Does this advice sound novel?” rather than “Would I benefit from taking this advice?”
Replace Stereotypes With Experiences is about how can often let affect and aesthetics be major decision-making factors, when in fact experience is a far better guide. There’s a mix-up happening here where your feelings about X get conflated with your feelings while doing X.
Explication explores how vagueness can be comforting and be a place of retreat for our poorly-made plans. It’s only by diving straight into the scary-feeling uncertainty and making specific claims/plans that give us an opportunity to receive feedback and verification.

Humans have an internal universe. Rationality, as a thing humans do, is going to interface with said universe at some level. When making decisions or practicing rationality, then, the ways our minds interact with our mental conceptions of rationality do in fact play a role. That much is clear now.

And yet, you can’t find good solutions by just detachedly observing things from a distance.

Sighing, you roll up your sleeves.

It’s time for a chat with your inner demons.

D] The Human Alignment Problem:

“Hello System 1, my old friend.”

You approach yourself tentatively.

“I tried going against you, molding you to my will. Then I tried ignoring you entirely, hoping I could do well without you. But it looks like I’ll need your help after all. Turns out the real Rationality was inside me all along.”

You say nothing in response.

(It’s yourself, after all.)

“How are we feeling right now?” you ask.

Which, of course, you already know the answer to.

Anger. Aversion. Calm. Sorrow. Worry. Want. Pleasure. Joy.

You have lots of feelings.

And they’re all important.

****

Given the intrinsically mental portion of learning rationality, the direct approaches given by techniques and habits are missing a crucial component. Namely, there are situations where, despite your best efforts to set up a system to do X, something inside of you is resisting. The sensation of forcing yourself to do something is not a pleasant one.

So instead of trying to bind the demon to your will, you fuse with the demon.

Practically speaking, this means coming to terms with all of the intuitive, wordless, gut pieces of yourself.

One component of this frame is letting go of some sense of direct control over yourself. You accept that motivation isn’t always something you can hack together with a formula. It’s not that you endorse a worldview where motivation is magical and irreducible, but just that, instrumentally speaking, there are ways of becoming driven which don’t involve thinking, “And now I must motivate myself to get X done!”

What you gain in return is a great deal more self-trust. You no longer need to be the person that watches over your own shoulder, making sure that you get things done. It’s now much less about “forcing” yourself to do things because the things you’re doing are things you want to do anyway. Internal conflict is generally removed from the equation because you’re giving all of your different sides a voice.

It’s not just that you’ve got a new technique that draws on a new way of getting techniques down. This is a viewpoint that isn’t about the techniques.

Things I’ve written which are about this shift into feeling out your feelings:

Feelings Matter. As advertised. A slightly longer explanation about how I think about feelings and some of the social incentives for not wanting to publicly endorse feelings as important (e.g. being associated with silly self-help gurus).
Lying On The Ground is about cultivating feelings and sensations by, as the name suggests, lying on the ground. It presents the idea of paying attention to feelings as, not exactly a productivity technique, but as a useful thing in itself.
Recovering From Failure looks at the commitments you make with yourself. It takes an honest look at what’s going wrong when you break them. It operates off the assumption that all the parts of yourself have needs, and commitments you break represent unmet needs.

I think the best exposition to this viewpoint is the Replacing Guilt series by Nate Soares.

Being whole with all of yourself is a powerful place to be in. You’re able to resolve your inner aversions, which are often upstream blockades for lots of endeavors. Moreover, words like “procrastination” and “motivation” are a lot less meaningful now that you’ve blurred the distinction between your “wants” and your “shoulds”.

You just do them. Because.

(What more reason do you need?)

And yet, even with this state of mind, you find yourself tempted...

E] Paradigmaster:

Having aligned yourself, you have a better measure of yourself. You know there are some things you can’t do.

Or can you?

Throughout your travels, you hear whispers. You hear whispers of other powers. Other powers which promise more than alignment:

Resolve, an ability to surpass your limits, to endure any trial, to always power on.
Attractor Theory, a graceful way to go with the flow and yet preserve your autonomy and will.
Folding, a self-awareness even deeper than alignment, which allows you to change yourself at a fundamental level.

Which should you choose?

You’ve always been a good student.

You study them all.

Time passes.

You now wear more than one hat.

You’ve got lots of models. Some of them conflict, but they all have their use-cases.

You’re a master at knowing which model to use at the right situations. Sure, there might be some underpinnings/assumptions that don’t change much from one frame to the next, but the point is that you’re fluid enough to recognize which way of approaching things is the most effective.

The most representative work I have of this frame is the Instrumental Rationality sequence as a whole, which is perhaps cheating because it isn’t so much as one thing which switches between models, but that the different essays contained therein use different models.

At the end, you found Many True Rationalities.

And yet, you have to wonder...the whole point of all these Rationalities is to get things done, after all. And, if you look at it hard enough, isn’t still just about countering cognitive biases? It’s not like the central problem has really changed.

Looks like you’re off on a quest to find the One True Rationality.

Coda:

Those familiar with Kegan’s stages of development will likely see parallels here, especially as I myself was reading In Over Our Heads during the writing of this essay.

I took care to try and paint this framework, not necessarily one where each stage is better, but as a set of growing considerations. I think the most useful insight I can offer here is where you start to see different rationality techniques emerge as a consequence of trying to solve different components (practical, mental, ontological, etc.) of the self-improvement problem.