I have noticed that people often mix up two cognitive tools, models and frames.

Models

A model is a set of concepts and beliefs that allows you to draw inferences. For example:

  • John likes hiking and John likes me (model) so if I invite him to join a hike he will say yes if he has time (inference).
  • Ibuprofen has a serum half life of 2 hours and a does below 15mg isn’t noticeable (model) so if I take 100mg for a headache in the evening and wake up without a headache 8 hours later, the headache is gone rather than suppressed (inference).
  • Specific changes to the status quo usually don’t happen (model) so a good starting point for predicting the probability of some policy without much traction being implemented is somewhere between 0-10% (inference).
  • The universe wants good things for me (model) so if I try to become pop star hard enough I will necessarily succeed.

A neural net or a linear regression is also a models, but in a more formal sense. Here, I care about models as tools of thinking.

Models can be faithful representations of reality or not. They can sometimes give wrong predictions, or correct predictions for the wrong reasons, but ultimately they try to make objective predictions about the world.

Frames

Frames subjective. Their purpose is not to draw inferences but to direct your thinking. To interrogate a frame, ask: What sort of thoughts will it generate?. For example:

  • My conversation with James is about status (frame). How can I tactfully mention my new watch / racing bike / set of holiday homes?
  • Or it’s about approval (frame). What can I do to get James to like me?
  • There is a lesson in every failure (frame). What can I learn from John not liking me?
  • I am not the sort of person who can X (frame). How will my attempt to do X fail?
  • Every person I meet is a potential friend (frame). ???

Frames can be helpful or unhelpful. They can be broad, informing your entire outlook on life, or narrow, influencing just one task or interaction.

Frames can be rooted in beliefs about the world, but they do not, in themselves, make claims about it. There is no such thing as falsifying a frame.

Mixing and confusing frames and models

Sometimes the model you have also imposes a frame1. It might be an objective fact in reality that you are bad at, say, dancing. If your frame is “Dancing is not for me,” you are unlikely to generate the thoughts that would allow you to improve. If your frame is “I can learn anything,” your chances are much better.

If “I’m bad at dancing” is your model, it might be true, and it’s good to believe true things, but it can easily become your frame too. There is nothing virtuous is holding on to unhelpful frames.

I have noticed specifically rat-adj people find something icky about statements like “There is a lesson in every failure.” On the face of it, that’s understandable. “Doesn’t Science say that the universe is just meaningless atoms bopping around?” (source, note the original is also quoted). There is no good reason to expect a lesson to be waiting there for you. In fact assuming it is seems to commit you to some belief in fate, or something similarly woo.

But that can miss the point, if that statement describes a frame (generates thoughts) and not a model (generates predictions about external reality).

I have found that rats2 sometimes miss the role that frames play in thinking. Even though we would like our thoughts to be a stream of pure objective truth, there is some subjective process in your mind that generates your internal monologue3. Often, for truth-seeking people, this frame is quite harsh. Understanding the frame you inhabit will give you more agency about your thought processes. Ignoring them arbitrarily limits your cognitive arsenal.

I should mention that on the flip side, non-rats (predominantly) will sometimes take their frames to be models. Say you believe that The universe wants good things for you. As a frame, this can be valuable, but it’s a terrible model. Perhaps people don’t want to let go of beliefs like that, because they are the foundation of useful frames.

It would be nice if we could learn to access frames that do not naturally arise from our belief systems. If you are religious, you might ask “Why did this happen to me?” or “What is the right thing to do here?”. I don’t know if you can ask these questions with the same earnestness as an atheist. I would like to experiment more here.

Another thing I would like to is to examine the frames that I find myself typically adopting, and to try putting myself into a different frame, by thinking about the thoughts I would like to be generating and getting a feel for it. I have found that language models are decent at this, if I prompt them with the examples from the first section and a clear statement of a frame I would like to understand better.

Related

  1. Here it would be very convenient to claim that studying economics makes you more selfish. Is it true? I looked at a meta-analysis which claims yes. But looking more closely (at Table 3 to be precise), there are only three studies which it identifies as having “Low” risk of bias across types of bias (read: those are the studies that are actually good according to the authors of the meta-analysis). One of these studies only looks at behaviours in games, the two others find no causal and only a selection effect. So alas, this bit of insight porn will not have made-up science to back it. ↩

  2. A term of affection, I promise. ↩

  3. Or, if you don’t have one, your internal felt-sense token stream (I guess). ↩