Recently I was talking with Brienne face-to-face, and she noted that a question I’d asked her would be much easier for her to answer if we were talking remotely over a text channel:
Neat thing I learned from Ben Hoffman today: If I imagine that I'm typing at a computer while I'm actually talking to someone in person, I can use my brain better than I usually can in face-to-face conversation. I think the two key thoughts here were, "How would I think about this if I were at a computer with an Internet connection?" and "Imagining seeing the question I'm trying to think about written out in text form.” -Brienne
When I found out that this worked, I thought about what heuristics I was using to generate that suggestion. Here are the ones I initially came up with:
- Find out what's different about the times you can do the thing. Brienne had mentioned that answering my question was hard because we were talking in person but would have been easy over text communication. This heuristic triggered a question from me about what her cognitive process would have been over text. In response, she figured out that she’d be using some sort of visual processing, so I suggested that she just do that right then and there.
- Attention constraints apply to System 1, not just System 2. This suggested that Brienne needed to take a break from whatever subconscious processes she was running as part of the conversation, to free up mental space for a different operation. I think I often do a generalized version of this where I look away from the other person or close my eyes in order to drop my social and situational tracking and look into inner space. This comes at the price of some perceived social awkwardness but lets me think a lot better.
- Different sensory modes trigger different mental modules. This made it seem especially plausible to me that asking Brienne to switch sensory modes would enable her to use cognitive processes that were blocked when she was dealing her voice and mine.
But now I think there’s a broader valuable generalization here, based on Brienne’s recent blog post on hypnotic binding. Brienne writes:
There's a suggestion technique called the hypnotic bind, which everyone heard a bunch when they were five. It looks something like, “Would you rather put away your toys now, or do you want to put them away after dinner?”
Consider what happens in a child's mind when they hear this.
They've been asked a question, so they're inclined to engage their attention in a search for an answer. But the search space for the answer is limited to the space of thoughts that assume they will clean up their toys at some point tonight.
Furthermore, the process of searching for an answer costs them attention, which limits their awareness of the broader desires they feel at the moment. (They want to keep coloring, and they don't want to put away their toys at all.)
So they say, "After dinner."
When this goes as planned, what they are aware of having just experienced is a weighing of options against their values, and then a decision among the options based on those values. When you experience the weighing of options followed by a decision based on your values, it feels a lot like you want whatever it is you've just chosen.
Used as a hypnotic technique, double binding is often about belief and perception of things besides choice. “Do you think you’ll fall deeply into trance now, or will you drift there more slowly as you listen to my words?” Either way, you’re attentive to whatever sensations are consistent with “going into trance”, which is over half of hypnosis right there.
Hypnotic binds don't have to take the either/or form, though. I often use single binding deliberately when I teach: When I pause for questions, I always ask, "What questions do you have?", and never "Are there any questions?"
Since students usually do have questions but often have trouble identifying them on command, directing their attention to the range of thoughts that assume they have questions saves them some work: It leaves more of their cognitive resources available for choosing among the questions that they have.
"Are there any questions?", by contrast, directs attention to the search space of "yes" and "no" - neither of which is itself a question! I always have trouble with this when someone asks me “any questions?”. “Welp, I see no questions in this search space, so I guess the answer is no.”
I’ve sometimes felt a little worried when asking, “What are your questions?” while teaching a class. I’m worried about what I’m doing to the minds of people who don’t have any questions. Occasionally, I’ll respond to this discomfort by clumsily tacking on, “It’s ok if you don’t have any questions,” which explicitly suggests that they don’t have any questions! Which is the opposite of helpful for the people who struggle to identify the many important questions they do have.
Hypnotic binding appears to be an instance of framing a bounded cognitive query so that it’s much easier to search within the framing of the query than outside it. Brienne gives the somewhat adversarial example of tacitly limiting a child’s choices without calling their attention to the fact, and the somewhat cooperative example of helping people look for questions they might have about a lecture, without wasting cognitive effort on the question of whether they ought to have questions, whether their questions are good enough, etc.
I think an important abstraction here is that when you ask your brain a question, it’s often not enough to ask it something that specifies logically what you want - you also have to give it some clues as to where to look for the answer. I call this shaping the query.
Learning how to shape queries efficiently can be useful for assisting other people, but it's also useful in consulting one's own brain. I’ve written about other ways to shape the query for oneself. A key technique is to use positive search terms that are associated in your mind with the thing you’re looking for. As a recent example, Brienne told me about Eliezer’s ambition-calibrating heuristic that if you can’t think of a time you’ve failed in the last 6 months, you’re not trying hard enough things. At first, I couldn’t think of anything - and began to tell myself a story in which I just don’t classify things as failures but instead just think about cases where I’ve redirected my efforts. But then I posed a different query - "what's a recent project I undertook?" - and immediately thought of one where I'd failed, multiple times, in the last 6 months. Because my record of tries is efficiently searchable by project, but not by month or whether I failed.
This is part of why I’m more skeptical than I used to be of attempts to eliminate cognitive biases from our thinking. Specific, targeted debiasing techniques like calibration are great. But it’s important to frame the search in positive terms - look for ways to increase accuracy, rather than ways to decrease bias. If you avoid heuristics that are limited and will give you wrong answers under some circumstances - well, you have nothing else to think with! Just one finite brain, and a bunch of often-useful heuristics. There’s no fully-general tool. But you can learn what the different tools are good for, in order to use them efficiently, and notice when you need to find one that’s not in your toolbox.
I've been thinking of this part of the skill tree as "learning the query language" and have been thinking about it in terms of specific cues in different forms of psychotherapy. It seems like psychotherapy revolves around cues that seem to do *something* across a wide range of people, often to the extent that the discovery of a new general purpose cue leads to the person who found it creating some new named psychotherapy 'school' centered around that particular cue. Examples are the particular queries from IFS, Focusing, REBT, Coherence therapy, CBT, etc. The queries in each of these carry tacit information about the model of cognition they are using.
So maybe something like: if you have some set of useful queries all consistent with a particular model of cognition, then those queries will be found to be useful by the subject if that model of cognition helps them unblend/get unstuck enough from their current model to be able to usefully move around again. 'Movement' here being some felt sense that is obvious from the inside but hard to talk about.
Oh and one way to get a quick read on the cues in psychotherapy thing is with humor: imagine someone imitating a CBT counselor and saying cliche CBT things in an improv game or something. This exposes the schema. At least for me.
I think the exploitability/inexploitability framework is a better match here. Sometimes you can help someone by nudging them to use models they already have. If that doesn't work then you need to help them with something outside their model.
But this feels sort of overly general, like it doesn't constrain my anticipations much. Different models should be better fits for different situations, depending on the type of mental operation the person needs to try. I often have an intuition that someone's problem is a better "fit" for one model than another, e.g. if someone sorta knows something but has trouble remembering it in the face of old triggers, that feels CBTish. while if it looks like someone's S1 doesn't trust them to do a thing for unaddressed reasons, that's more IFS/CT flavored.
>the exploitability/inexploitability framework
Not sure what that means?
By cues I mean that collecting cues seems interesting in that you can very cheaply cue people and observe which cues they are responding to. If you know what model of cognition that cue corresponds to you can play in that frame for a while and the person will likely report afterwards that it was useful. Of course there is the problem of feelings of usefulness vs actual usefulness a la productivity porn, but that's a whole can of worms.
Pingback: Reading, writing, and thinking, with your brain | Compass Rose