Wisdom is Working with Frames (2/2)

How do we get wiser?

In Part 1 we looked at definitions of wisdom and arrived at a reasonably operationalizable one: 

Wisdom as adaptively working with frames.

And frames as the sets of assumptions, ontologies, and attentional lenses we use to navigate the world. 

In this part we sketch how we can become wiser by getting good at working with frames.

From Cartesian rat to trusting intuition

Growing up, I struggled with intuiting social norms and taking others' perspectives. This led me to over-rely on naive empiricism, science, and propositional logic to make sense of the world and communicate. One funny consequence of this approach was my attempt to follow Descartes in rebuilding my worldview from a very simple axiom system. This had mixed results.

I broke myself out when I encountered the fuzziness of machine learning, the limitations of formal systems via Gödel's incompleteness theorem, and the physical limits of math imposed by computation.

These concepts broke me out of my rigid frame and opened my eyes to fields that take subjectivity seriously: Jungian psychology, meditation, and various religious traditions. I began exploring these areas, using analogies to computation and math to navigate them and estimate the plausibility of their claims.

As we’ve seen, the ability to "break out" of previous assumptions is central to wisdom.

There’s something else here. I was giving primacy to reason, to symbolic, propositional thoughts and ignoring intuition. I was privileging one type of knowledge over others. 

The 4Ps of knowing

Vervaeke distinguishes between 4 types of knowledge we need to engage with: propositional, procedural, perspectival, and participatory. This video is a really nice explanation.

Propositional knowing is book-smarts. Facts, beliefs, rules. Thinky knowing. Knowing about things, about what is “true”. This is very useful but insufficient for living effectively.

Procedural knowing is knowing how to do things in practice, it’s about skills. e.g. knowing how to use a hammer (instead of knowing lots of facts about hammers). But is every problem a nail?

Perspectival knowing is about how to perceive the world. This kind of knowing lets us take a situation and know what skills are appropriate, what really matters in a situation. Not just thinking, but really inhabiting a perspective, knowing how things feel. It also makes clear the limits of our perspective and the value of others.

Participatory knowing is about being in right relationship to the arena we’re operating in. It’s also evidenced by being in a dynamic relationship with other people with their own perspectives, and being in flow without thinking about it, in the groove, being at home in a situation.

These types of knowledge support and constrain each other. A propositional theory will limit your options of things to try, making the search space manageable. Perspectival lenses will make some facts salient and not others. Participatory knowing will deploy the procedural skills you’ve built up at the right time, and so on.

Frames can exist at each level of knowledge. Facts and skills feel like things we “have” and are easier to treat as objects. Perspectival and participatory frames seem to be less conscious, lower level, closer to “the metal”, more like things that we “are” rather than things we have.

Embodiment, faith and sages 

This is where if I had more time I would connect my personal story at the beginning to stuff like Ian McGilchrist’s Left/Right hemisphere distinction, or Daniel Kahneman’s system 1 and system 2. Our unconscious consumes and processes vastly more information than our conscious symbolic mind.

The deepest frames are hard to see and steer because they are embedded in our bodies as instinctive patterns of thought and muscle tension. Knowing what frames to shift into and how can’t always be done with pros and cons lists, it requires subtle signals from our embodied intuitions and having the faith to follow them. 

Faith here means trusting without understanding the causal relationships involved. Trusting that your sense of curiosity and excitement and opportunity is intelligently picking up on complex information from your environment and guiding you in a way that the conscious mind can’t.

Even if we can’t solely rely on our propositional mind, we can still put it to use in designing our environments to be maximally supportive of our growth. Make good habits easy and bad habits high friction. Surround yourself with people you want to be more like.

This can be seen in wisdom traditions like monasteries where the environment and access to mentors and advanced peers is taken care of and tested over long periods. 

By hanging out with people you want to be more like, you get to observe the way they pay attention and respond to their environment. By emulating their perspectival and participatory frames - you can “internalize the sage” and accelerate your own transformation.

Working with frames

A way you can break “working with frames” into parts is to

• recognize unhelpful frames

• select appropriate ones

• break bad frames, and

• craft better ones.

Recognizing Frames

To recognize frames that are being automatically applied, we need what Vervaeke calls a “transparency-opacity” shift: What was transparent, the lens we were looking through, becomes opaque, an object to be examined. 

This same pattern of taking what was part of the subject as an object is also key for human development from Piaget to Kegan. Whereas relationships are seen as an inherent, fundamental part of the self in the socialized mind (Kegan 3), they become an object that can be changed in the self-authoring mind (Kegan 4). 

There are practices that can help us train this function. We can reveal our cultural assumptions through travel with deep cultural immersion, to bump against unfamiliar conventions. We can reveal our own psychological frames as contingent by practicing taking others’ perspectives through personal relationships or reading novels. And most thoroughly, we can rigorously investigate experience with insight (or Vipassana) meditation to see thoughts and emotions as objects. Michael Taft describes the whole process of meditation as “making the unconscious conscious.”

As we’ve seen, some frames can overstay their welcome. Upon recognizing a frame is there, we have the opportunity to drop it or break it if it’s unhelpful. 

Breaking Frames

Frames recognized as unhelpful can still be hard to drop. My friend Hunter Glenn found that clarifying what we value in perspectives makes them easier to let go. I think it could be because a bad frame may still be load bearing, and it feels safer to drop if we have clearly articulated its role in case we need to return to it later.

A master of recognizing frames is always conscious of what frames are active. However, this is not required to be able to break them. There are clear cues that an unhelpful frame is active, most notably a sense of being stuck, which can be the mindfulness bell that triggers you to question your frames.

Concrete problem-solving contexts like strategy games and technical puzzles are great practice because they provide a clear feedback loop. In all of these examples, the practice is to:

1. Notice when you’re stuck on the problem at hand 

2. Recognize which frames are active

3. Let go of these frames and open up to new ways of looking 

4. (Wait for the insight to hit - “Eureka”, you’ve found a better frame!) 

Characteristically frame-breaking contexts include:

• Conventional therapy. Cognitive behavioral therapy works by identifying, challenging, and changing negative thought patterns.

• Vipassana meditation in high doses such as in a retreat will, beyond recognizing them, allow you to gradually dissolve limiting frames.

• Psychedelic trips will shatter frames and deliver you to similar states quickly, albeit less reliably.

After dropping a frame, there is a vacuum of power! How can we pick good frames? 

Selecting frames

We can try frames on as if they were clothes! 

You can practice going into a meeting with a “debate winning mindset”, focusing on the logical structure of arguments  or you can go in with the intention to relate to people, focusing on feelings in your body - the outcomes will be very different. 

You can adopt a strict rational material empiricist worldview, or an animist view where natural features are personified - and they will let you solve different sets of problems.

Different problems call for different frames. You can decide between frames based on your direct experience of outcomes, or you can feel for resonance between the frame and the situation in your body. (Tapping into relevance realization and predictive processing, from part 1 of this essay)

Propositional frameworks can help. Dave Snowden’s Cynefin, parses the world into different contexts where causality works differently (and which therefore require different frames):

For example, strictly rational methods (optimizations, simulations, etc.) can be applied in a complicated situation, whereas a complex context requires interactive safe-to-fail probes. 

It’s cool to be able to skillfully select frames but you know what’s cooler? Rolling your own 😎

Crafting frames

I take some care and pride in crafting my personal frames, ontology, and worldview. I’ve gotten a lot of mileage out of keeping a mostly consistent systematic model of the world, and expanding it by considering woo-sounding beliefs and asking “what if they were true?”. 

People like it when I make computer metaphors to translate woo concepts. It’s one of my favorite exercises - and if I find a plausible mechanistic explanation for how something may work - I’ll try it on as a lens for a while, in a separate context from the rest of the ontology, running in an “internal VM” so to speak. If it holds up and seems to add to my ability to make sense of the world and predict things, I’ll rely on it more and I’ll gradually let it integrate into my main worldview.

I enjoy learning little skills like pen spinning, contact ball juggling, contact improv lifts, or Mongolian throat singing. But for some reason I don’t really watch tutorials or try to follow the orthodoxy, I just try to figure them out on my own. I think it’s a powerful way to practice being in chaotic or unfamiliar environments and gradually adding structure and mastering a skill. You could call it ~the art of dabbling~.

Something similar can be said about improvisation disciplines, where frames are changing fast and one needs to adapt in real time. It can be jazz improvisation, where chord changes or just decisions by other musicians can radically change the context you need to respond to. It can be improv theatre, or contact improv dancing.

Since working memory is limited, we can’t really keep adding frames to our repertoire indefinitely. Luckily, we can merge and refactor frames! You just need to find ways to make the puzzle pieces fit together. In my case with woo things - it feels hard to square chakras with a materialist base worldview, but if you conceive of them not as Things that are Inherently Real, but rather as convenient abstractions to interpret signals from the body thematically (e.g. root chakra is about survival, heart is about compassion and motivation, etc), then there’s no problem and they can fit in the same ontology! 

Or another example. If you’ve experientially found value in a number of practices with diverse backgrounds that don’t all have obvious materialist explanations or connections to other unifying frameworks, you might be led by your intuition to write a blogpost about wisdom 😉

Conclusion

Wisdom is being good at working with frames! 

Frames can exist at different levels, and be progressively harder to shift: Propositional, Procedural, Perspectival, and Participatory.

You should structure your physical and social environment to support your flourishing!

You can break frame work into recognizing, breaking, selecting, and constructing frames!

And there are many suggested practices above!!!

Notes on Automation

Making the Bayesian connection with relevance realization (in part 1) and transitively with wisdom was one of the last things I did writing these posts. 

Expressing wisdom in terms of having good hyperpriors that modulate learning and switching between cognitive modes is satisfying to me as a description of wisdom for AI. 

I’d be interested in investigating “frame dynamics” with mechanistic interpretability methods. Current (Oct 2024) LLMs are often good at inferring appropriate frames from context - e.g. how to interact with you given your language cues - or given a problem without all the background spelled out. However they’re still not great at questioning and refining frames that are given to them - they will usually sandbag on this front and take the user’s framing as gospel even if they do have access to more complete or adaptive frames.

I also think that many of the examples laid out across the post may provide a broad sample from which to construct evals for adaptive and skillful work with frames, in which models are gradually provided with new information and measured as to whether they’ve appropriately adapted to the situation or not.

Notes on Preference Drift

Ultimately, the frames we use to make sense of the world are selected and measured based on other frames, other ways to filter out information and decide what to care about. This means our values must change over time, and that’s probably a good thing. 

The fact that humans don’t encode a teleological end means that we get to do some “lighthouse hopping”, sailing towards the lighthouse of a value function close enough to see the next one along the shore and adjust course, instead of sailing straight into it and crashing into Goodhart’s law. 

Of course our values could change in ways that to us would seem abhorrent or like a great loss, and it would be difficult for us to tell what was lost after the change. To address that we can try to 1. Change slowly; 2. Allow backtracking or conditionally pre-commit to backtracking; 3. Change more wisely (i.e. more skillfully, with more implicit knowledge of the preference landscape).