Dichotomies

[This essay is about dichotomies in thinking and a few categorizations that might be of interest. I go into some potentially useful binary distinctions for thinking, attention, and action. Epistemic status: There is some strong evidence for certain dichotomies, like fast vs slow thinking (i.e. dual-process theory), but the rest is far more shaky.]

I sometimes half-jokingly say that any two opposite words, when appended to the word “thinking”, gives you something insightful. I’m going to try and make good on that claim. Here, I’ll try to explore some (maybe) useful dichotomies in thinking, attention, and action.

My main assumption going into this essay is that words have power. (See here to get the gist of the intuitions behind this claim.)

Let’s dive into some dichotomies:

1) Fast thinking vs Slow thinking:

This is probably the most iconic distinction, brought to us by the proponents of one of the dual-process theories of thinking. People like Keith Stanovich and Daniel Kahneman have done a lot of work in this area, and I think it’s a fairly solid way to classify thinking (with some minor caveats, of course).

Fast thinking is for mental processes that happen very quickly, without much conscious input. Examples include flushing the toilet, responding “Doing well. How about you?” to a greeting, noting that a face seems familiar, and recognizing colors, like blue.

Slow thinking is for mental processes which might require more time and deliberate effort to get right. Examples include writing a story, remembering what you had for yesterday’s lunch, multiplying numbers in your head, and trying to follow along a high-level conversation, perhaps a lecture.

The main distinction here, then, is that of effort and time involved for each process. There’s often a tradeoff occurring between complexity and speed. Fast thinking often prevails, perhaps because stopping to calculate a car’s trajectory is far more costly then buying the refrigerator which looked the most visually appealing at first glance.

That is to say, using slow when we should have used fast is perhaps  worse than using fast when we should have used slow.

When it’s useful to note this dichotomy: If you find yourself being a “habit-zombie” and drawn to a lot of situations where you’re operating off of cached actions. Here, it might be useful to “pop out” and put in some effortful thinking to see how you may want to alter your default behavior in response to certain cues.

---

 

2) Focused mode and Diffuse mode:

This is from Barbara Oakley’s work on learning and studying.

Focused mode is about directing your cone of attention towards a task. It’s what we typically think about when we refer to solving problems. Here, focused thinking is about trying different tactics, adapting, and interacting with the topic. It’s the subject of our attention, it’s the topic in mind.

Diffuse mode is about letting your brain drift off into different directions and spit out new things. It’s the sort of thing that happens when you find yourself in a mindless task, like mopping the floor, and your brain is still racing, running around certain topics. These ideas and thoughts are characteristic of diffuse thinking.

The difference here is whether you’re directly engaging with the subject, or if you’re letting your mind go off and do its own thing. If you imagine your brain as a flashlight, your cone of attention goes from a thin, powerful beam to a suffused glow of light.

If you focus more on the model when applied to attention, you can think about it as the difference between attending to real life, in the present, or thinking about something else. Focused attention is about being in the moment, responding to things in the world. Diffuse attention is where your thoughts wander and you’re only dimly aware of the present.

When it’s useful to note this dichotomy: If you tend not to build in the sort of “mindless” time that allows for diffuse mode. When you are only in focused mode, you’re likely missing out on some additional thoughts you otherwise won’t be exposed to. In my experience, diffuse mode gives me different takes on certain things, or ways of looking at the world I typically don’t take.

---

 

3) Dirty thinking vs Clean thinking (or Villainous thinking vs Heroic thinking):

This basically approximates the consequentialist and deontology distinction. It would have also likely fit in the essay about decision theories in real life, but I hadn’t thought of it then.

Dirty thinking is about looking only at the end results. It’s about figuring out how to hack things together and optimize, second by second. For example, dirty thinking supports whatever actions you need to take to get to your goal. Some examples might include asking a friend to burn your wallet if you don’t get work done (self-blackmail), turning off WiFi to reduce distractions (precommitment), getting yourself hooked on a video game to substitute an even worse addiction, or finding the lowest energy productive thing you can do when you’re tired.

Clean thinking is about following rules, even when the rules themselves might lead to less beneficial short-term benefits. It’s about being policy-based. When you’re using clean thinking, devotion and discipline are virtues. Note that it’s not exactly about being charitable with yourself. In fact, if you find yourself struggling to get out of bed, clean thinking is about being able to get out of bed anyway. You do this because getting out of bed when it’s hard follows the general rule of “do hard things”, which you perceive to be beneficial in the long run.

The distinction here is also similar to deciding in-the-moment vs planning ahead. Clean thinking sacrifices dirty thinking’s reliance on flexibility and side effects for the powerful ability to stick to general principles.

Funnily enough, the means that clean thinking is actually closer to the typical villain archetype, who is willing to sacrifice everything for the greater good. Dirty thinking, then, reprises the traditional hero, who is forced to consistently adapt on the fly, as the treacherous environment destroys any chance of their heroic plan from working.

When it’s useful to note this dichotomy: If you find yourself constantly compromising your morals to do better locally (i.e., lots of dirty thinking), you may want to try being more policy-based. I think that rationality too-often leads people down the dirty thinking path, where everything looks like a new tool to shape to their ends. Thus, over-correcting in favor of more clean values might be useful.

---

 

4) Inside view and Outside view

This also comes from Kahneman; it’s more a way of looking at things than a form of thinking.

The inside view is about incorporating information only someone in your position would know about the project. You’re looking at the details, the specifics, how certain things / people in your undertaking interact. As such, you’ll probably have good models about how the different pieces in your plan would respond to events and stressors.

The outside view is about using information related to the group you belong to. You’re looking at averages and aggregates to determine how things will go. For example, if you’re going on a difficult hike and you know that the trail is only completed by about 3% of people, you can expect failure to be highly probable.

As the reasoning goes, despite your inside view opinions about how prepared you are, 97 out of every 100 people likely thought the same thing, and yet still didn’t make it, then to assume otherwise would be arrogance.

In practice, the distinction is basically between whatever seems consistent and reasonable to someone inside a situation and what a neutral outsider might think. Note that this doesn’t mean that one view or the other is always correct, as there are examples where either one triumphs over the other.

Although I admittedly think overcorrecting in favor of the outside view is better than the other way around.

When it’s useful to note this dichotomy: If your predictions are overconfident, i.e. miss reality by a long shot. Especially note the inside-outside view distinction if your internal BS detector starts to ring in any situation—that’s a good trigger to consider what an outsider might say.

---

 

5) Object level vs Meta level:

Every rationalist’s favorite dichotomy.

The object level is about doing things in the real world. You might be doing some quick planning, but the focus is on getting things done. Things are happening, you’re adapting and taking actions. It’s about heading into reality, bumping into people and details, and collecting information. Examples include publishing a book, writing some software, giving a speech, and most other things you can do.

The meta level is about thinking about the things you can do. Here, you’re trying to answer the questions of “What can I be doing?” which might go into the question of “Why do I want to do things?” etc. Here, you’re looking at the bigger picture to see which things might be worth your time. Examples include planning, researching choices available to you, asking people for advice, and philosophy.

This doing vs planning dichotomy somewhat mirrors the inside vs outside view dichotomy. Actually staying on the object-level is somewhat analogous to the inside view as it gives you more information about the world. Going meta is, then, a little like the outside view, as you’re trying to use more high-level principles to reason about the world.

When it’s useful to note this dichotomy: If you find yourself doing too much of one or the other. Being stuck making plans doesn’t work if your plans don’t cash out into real world actions, and only doing stuff isn’t effective if you don’t first give it some forethought.

---

 

6) Generative mode vs Iterative mode:

(Similar to top-down and bottom-up design.)

This is a way to approach doing things, two different ways to consider how to start a project.

Generative mode is where you build up the object, piece by piece. Often, you already have the main skeleton and idea in your head, and the process consists of slowly building it up. An example is cooking, where you already know what the main ingredients are; the overall process, then, is about combining them to make the dish. For some projects, then, using generative mode encourages you to identify the big parts first, and then sequentially put it together.

Iterative mode is where you start you something rough and polish it, bit by bit, until it’s finished. It’s about starting, even if you lack a full picture, because, along the way, you’ll improve the project. An example is writing an essay, where the first step is to get the ideas down, then you edit, and then you finalize the product. It’s less about needing to get each individual step 100% correct because you know that the overall creation process encourages changing and improving.

As I mentioned above, these are two potential ways of getting started on a project. Both of these are far better actions than mere hesitation or inaction, especially if you’ve been meta for far too long.

When it’s useful to note this dichotomy: Generative mode or iterative mode are two ways of thinking about how to get something started. If a project seems difficult to start because you feel like you need to get all the parts right, it probably makes sense to dive more into an iterative mode. Here, poor design is merely part of the process, and you can perhaps penalize yourself less for not meeting whatever mental bar was holding you back.

---

 

Conclusion:

So clearly, the categorization I provided here has limits. A common theme is that going way too far into either end of the dichotomy might be problematic. (I do think that going full out for clean thinking and the outside view is much less bad compared to the other ones, though.)

There’s a more general question of when a gradient makes more sense than a binary distinction. A good example where I think this is clearly the case is the distinction between “free will” and “coercion”, where the reality of the situation points to a variety of things which can influence one’s decision. In situations like these, having a dichotomy does not appear to be as useful a distinction, compared to the expressive power provided by a gradient.

 

In all honesty, I do think the above meta-dichotomy is useful to examine to avoid falling into the failure mode of proclaiming that “everything is a gradient!” in a way similar to how people who first realize a lack of objective measuring units implies that “everything is relative!”

But in the meantime, I hope these distinctions provide some interesting lenses with which to look at the world.