Willpower duality (a very short essay)

post by Gyrodiot · 2017-01-20T09:56:50.441Z · score: 5 (6 votes) · LW · GW · Legacy · 8 comments

This is a link post for http://www.jrmyp.net/blog/index.php?post/2017/01/07/Willpower-duality


Comments sorted by top scores.

comment by Qiaochu_Yuan · 2017-01-20T20:33:56.296Z · score: 7 (7 votes) · LW · GW

I think willpower and akrasia are both kind of fake concepts. Here's a different model: you have a bunch of parts, approximately in the IFS sense. Those parts want different things, and sometimes they come into conflict. Different parts are in control of what you actually do at different times. Some of your parts are verbal and some are not. (I don't think "verbal = S2, nonverbal = S1" is necessarily a helpful frame here; for example, focusing seems to be related to nonverbal parts, but it sort of breaks the S1/S2 dichotomy by being nonverbal but slow.)

"Akrasia" is when a nonverbal part is in control (and wants to, say, watch TV) and a verbal part wants to do something else, and says so. "Willpower" is when a verbal part attempts to forcibly override a nonverbal part. The skill to try to build is understanding what your nonverbal parts actually want, facilitating communication between parts, and either making compromises between your parts or getting your parts to agree on what to do (I'm genuinely uncertain which is better).

Edit: Also, my impression is that this is essentially the model currently being taught at modern CFAR workshops, although people who have been to a CFAR workshop more recently than me should chime in about whether they agree or not.

comment by Gram_Stone · 2017-01-21T03:31:07.211Z · score: 3 (3 votes) · LW · GW

(for example, focusing seems to be related to nonverbal parts, but it sort of breaks the S1/S2 dichotomy by being nonverbal but slow.)

Noncentral nitpick that is meant to be helpful: Focusing is a counterexample to the lay dual process theory that people sometimes use around here, but not the up-to-date, cognitive-scientific one.

Briefly, the key distinction (and it seems, the distinction that implies the fewest assumptions) is the amount of demand that a given process places on working memory.


Although language is often involved in Type 2 processing, this is likely a mere correlate of the processes by which we store and manipulate information in working memory, and not the defining characteristic per se. To elaborate, we are widely believed to store and manipulate auditory information in working memory by means of a 'phonological store' and an 'articulatory loop', and to store and manipulate visual information by means of a 'visuospatial sketchpad', so we may also consider the storage and processing in working memory of non-linguistic information in auditory or visuospatial form, such as musical tones, or mathematical symbols, or the possible transformations of a Rubik's cube, for example. The linguistic quality of much of the information that we store and manipulate in working memory is probably noncentral to a general account of the nature of Type 2 processes. Conversely, the production and comprehension of language must often be an associative or procedural process, rather than a deliberative one; otherwise you might still be parsing the first sentence of this comment. That's all technically original research and I Am Not A Cognitive Scientist, but I think it should be pretty obvious even from a layperson's perspective.


There's nothing stopping Type 2 from being relatively fast, either; it's just another correlate that doesn't always hold. Trivial example: Have you ever awoken and not been able to make mental sense of what you're seeing for a few seconds? It might take you longer to do that than to perform one transformation of a Rubik's cube while fully awake, even though the former is automatic and the latter deliberate. In general, people sometimes seem to act as if there has never been a judgment that was simultaneously deliberate and fast, because they have come to describe all fast judgments as automatic. Such judgments are plausible in my experience.

See also: Evans (2013).

comment by MrMind · 2017-01-23T09:43:39.924Z · score: 1 (1 votes) · LW · GW

The skill to try to build is understanding what your nonverbal parts actually want

Isn't that the easy part? Just look at what it's doing: if I'm eating a bag of chips instead of working out than it means that my non-verbal part wants to eat a bag of chips. Or there's something else?

comment by Qiaochu_Yuan · 2017-01-23T19:07:38.857Z · score: 0 (0 votes) · LW · GW

The things your nonverbal parts are doing are often bad strategies for achieving reasonable goals, and so there's an inference problem to solve in figuring out what the underlying reasonable goal is. A lot of the things your nonverbal parts do are pica in a metaphorical sense (pica in a literal sense is e.g. eating ice cubes because of an iron deficiency). Your desire to eat a bag of chips, for example, might reflect an underlying goal of getting more salt or fat in your diet, because in the ancestral environment those things were rarer, but if you already have too much salt and fat in your diet then that's not super helpful.

A more pica-like example: suppose you catch yourself watching a lot of TV. Depending on the content of the TV, this might reflect an underlying goal of having more social connection (say if you catch yourself watching a lot of How I Met Your Mother, where the main characters form a tightly-knit group of friends). TV's not social connection, but it sort of vaguely resembles it closely enough to be kind of satisfying but not really. I think this is more what "akrasia" looks like a lot of the time.

comment by lifelonglearner · 2017-01-21T02:09:10.040Z · score: 0 (0 votes) · LW · GW

This appears to be a useful way to think about internal desires in a non-combative way. You're not focused on trying to override parts of yourself, but just listen and try to be an arbiter in the conflict.

comment by lifelonglearner · 2017-01-20T15:17:48.201Z · score: 2 (2 votes) · LW · GW

In my conversations with people from CFAR, I believe this is the view they espouse. They mention this during their discussion of motivation. Elo, Nate Soares, and Andrew Critch have also written about this at some point. (I can dig up links if you want to see how others tackle this.)

As a result, this looks to be a sorta stable fixed point for motivation-related thoughts. Despite knowing this, I often still miss opportunities to debug my internal motivations and source some more sustainable motivation. In general, I'd like to see more people outlining perhaps a mental transcript or a steps that they go through to get the S1/S2 cooperation to actually happen. Easier said than done, as they say.

comment by MrMind · 2017-01-23T09:44:38.138Z · score: 0 (0 votes) · LW · GW

Exactly. Even if I can debug the internal process, that doesn't stop it from happening.

comment by siIver · 2017-01-20T11:55:05.602Z · score: 2 (2 votes) · LW · GW

Yeah, this is pretty much my conclusion, too. If I had read this article a couple of years ago, it'd have helped me a lot.

I'd add that you should still overrule system 1 in some really important and rare cases, it's just not practical for recurring things.