Speculative rationality skills and appropriable research or anecdote

by [anonymous]1 min read21st Jul 201515 comments


Personal Blog

Is rationality training in it's infancy? I'd like to think so, given the paucity of novel, usable information produced by rationalists since the Sequence days. I like to model the rationalist body of knowledge as superset of pertinent fields such as decision analysis, educational psychology and clinical psychology. This reductionist model enables rationalists to examine the validity of rationalist constructs while standing on the shoulders of giants.

CFAR's obscurantism (and subsequent price gouging) capitalises on our [fear of missing out](https://en.wikipedia.org/wiki/Fear_of_missing_out). They brand established techniques like mindfulness as againstness or reference class forecasting as 'hopping' as if it's of their own genesis, spiting academic tradition and cultivating an insular community. In short, Lesswrongers predictably flouts [cooperative principles](https://en.wikipedia.org/wiki/Cooperative_principle).

This thread is to encourage you to speculate on potential rationality techniques, underdetermined by existing research which might be a useful area for rationalist individuals and organisations to explore. I feel this may be a better use of rationality skills training organisations time, than gatekeeping information.

To get this thread started, I've posted a speculative rationality skill I've been working on. I'd appreciate any comments about it or experiences with it. However, this thread is about working towards the generation of rationality skills more broadly.

15 comments, sorted by Highlighting new comments since Today at 4:50 PM
New Comment

Disclaimer: I work for CFAR and might well be biased in this response. That said, at CFAR we don't (re)name techniques or exercises in order to obscure what we teach or to try and grow an insular community - we do so to improve comprehension and skill acquisition.

For instance, we used to call our class on implementation intentions "Implementation Intentions", but we changed the name to TAP (Trigger-Action Planning) because that was much more memorable and provided a better handle for people to use the technique. (Compare "I'm going to make a TAP for that" to "I'm going to set up some implementation intentions for that", or "How many TAPs have you set up recently?" to "How many goals have you created sets of implementation intentions for recently?")

In other words, our goal isn't to hide the source of this material - in fact, we explicitly discuss the source material and provide links to relevant research in our workbooks - but rather to aid ease of use and ease of learning.

As for the more general point, CFAR is quite interested in spreading the information that we teach, developing new techniques, and furthering the art and science of rationality.

Several of our alumni have gone on to teach CFAR material to interested folks in their own area. We've provided multiple scholarships to people attending workshops with an explicit goal of taking what we teach and bringing it to underserved communities. We hold an alumni reunion yearly where CFAR staff and alumni share what they've learned and give mini-talks on the latest interesting developments in or promising avenues for rationality training.

I suspect that, if anything, CFAR would like to be doing more to bring this material to a wider audience - there's only so much time in the day, though!

Hey, I'm one of the founders of CFAR (and used to teach the Reference Class Hopping session you mentioned).

You seem to be misinformed about what CFAR is claiming about our material. Just to use Reference Class Hopping as an example: It's not the same as reference class forecasting. It involves doing reference class forecasting (in the first half of the session), then finding ways to put yourself in a different reference class so that your forecast will be more encouraging. We're very explicit about the difference.

I've emailed experts in reference class forecasting, described our "hopping" extension to the basic forecasting technique, and asked: "Is anyone doing research on this?" Their response: "No, but what you're doing sounds useful." [If I get permission to quote the source here I will do so.]

This is pretty standard for most of our classes that are based on existing techniques. We cite the literature, then explain how we're extending it and why.

Is rationality training in it's infancy?


As far as I understand CFAR plans to publish written material on their website that details their theory of how things sit together. Maybe as soon as the end of this year.

CFAR is quite willing to let workshop attendees be able to spread what they learn to other people. Nobody is encouraged to keep things secret but to spread knowledge. At both LW community camp in Berlin a large chunk of the material was CFAR based.

When it comes to the question of whether to speak about mindfulness or againstness, it reminds me of Jay A. Labinger:

The language you use to talk about something influences the way you think about it. If the chemistry you’re talking about is truly something new, then a fight over terminology may be quite an important part of getting to understand that chemistry better.

When I search for a definition of againstness on LW I found:

"The relevant CFAR unit is called "againstness." You can think of sympathetic dominance as related to (being?) a sensation of "againstness," e.g. when you get angry during a heated argument your feelings are directed against the person you're arguing with. "

Certainly the word mindfulness is not about getting angry. Being mindful allows you to recognize that you are angry but if you focus on teaching mindfulness the focus is different. The results will also be different.

Hopping is a much shorter phrase then "reference class forecasting". If you want to get people to actually do it instead of talking about it, it useful to have short terms for concepts.

If you want people to adopt a new behavior that they currently don't do, it's useful to have a new word for it. If you just use old words people nod that they agree and then don't change behavior.

[-][anonymous]5y 4

I disagree. CFARs(and the sequences) tendency to rename and reinvent things is significantly harmful as it makes it much harder to use the existing bodies of research, books, and anecdotes to enhance your practice of them.

The mission of CFAR isn't to summarize existing research but to reinvent the wheel. To go on to build better wheels. For that mission the act of taking techniques apart and reinventing them is vital.

For that mission it's helpful if people at workshops are in beginner's mind and don't say: "I already know what you are talking about because I already know label X".

Take the word 'rational' as use on LW. The word exists outside of LW. That means it should be easy to understand LW'lers when they use the term? No, because the term means something different on LW than it means outside.

Naming isn't easy. Of course there are bad names both in the sequences and in CFAR, but that doesn't mean that it's always a bad choice to use a new name. Inventing new terms makes it easier to actually align the listener and speaker and both understand the word the same way.

[-][anonymous]5y 3

That's a silly mission. There's already a ton of AWESOME wheels out there from psychology research, the self help movement, religious practices, and others. The short term goal shouldn't be to build a new wheel, it should be experiment and figure out which of the existing wheels actually role. Once that's done, you can go about inventing new techniques, but starting from scratch is just a silly way to go about it.

If someone is asking a lot of money just to teach old techniques under a new name, that... doesn't seem right. Even if the students are told in the textbook that the technique is actually old and known -- are they told it before they pay for the lessons? Are they told it openly at the beginning, or is it just a tiny footnote on page 98?

On the other hand, if the technique is not exactly the same, and CFAR believes that the modifications provide a significant bonus, then maybe the new name is deserved. Perhaps also if the old name is so inconvenient that people have problems remembering it and connecting the label with the referent. Also, sometimes there exist so many versions of some old technique, that the label doesn't connect to anything specific (ask all those people who "meditate" what precisely they do; everyone seems to use this word however they wish).

Having to reinvent old techniques seems like not doing your homework. (How many other old techniques are you going to reinvent in the next five years? How much funding do you need for that?)

I can imagine CFAR having good reasons for what they do. But I can also imagine them to suffer a "rationalist" arrogant version of Dunning–Kruger effect; something like: "Oh, I am so rational that I absolutely do not have to pay attention to the domain experts, because there is absolutely nothing I could learn from them, it's all just mumbo-jumbo... now let me think and invent my own mumbo-jumbo which will be more rational, because a more rational person - albeit a total beginner in the domain - invented it". (I am working at the frontiers of science / You are doing pseudoscience / They are superstitious.)

I would need more specific data to make a conclusion here. Looking forward to a public textbook.

If someone is asking a lot of money just to teach old techniques under a new name, that... doesn't seem right. Even if the students are told in the textbook that the technique is actually old and known -- are they told it before they pay for the lessons?

That assumes that the main point of the lesson is to teach specific techniques. That doesn't seem to be the case from what I understand by talking to Val about CFAR's strategy at the LW community camp in Berlin (I wasn't at an actual CFAR workshop).

Oh, I am so rational that I absolutely do not have to pay attention to the domain experts, because there is absolutely nothing

CFAR does happen a bunch of scientists who are domain experts in their advisory council. They read scientific papers. I don't think it's fair to say that they "pay no attention to the domain experts".

albeit a total beginner in the domain - invented it

What makes you think that the CFAR folks are total beginners?

There's already a ton of AWESOME wheels out there from psychology research, the self help movement, religious practices, and others.

Are you really suggesting CFAR should pick up techniques from religious practices and import them as is without reinventing them and stripping out all the paranormal stuff?

[-][anonymous]5y 1

In terms of religious practices, I think CFAR could play the same role that EG The Center For Mindfulness does at UMASS Medical Center. They figure out which wheels from zen practices actually role, and which don't.

[-][anonymous]5y 3

Towards trainable mental skills for domain-neutral high-performance cognitive reappraisal.

Blessed with the capacity for cognitive reappraisal, one is constantly confronted with some degree of freedom over the emotion they experience in a given of clarity. How does one decide upon an emotion?

Therapeutic considerations dominate literature on cognitive reappraisal, however, performance considerations take a share of pie too. To illustrate the latter, sports psychologists have identified certain emotions as higher performance and lower performance emotions in sport. Some of the results are counter-intuitive and partially incompatible with the therapeutic research. Importantly, it appears that student's performance emotions in the class differ from athletes. This makes it difficult to generalise about a general theory of performance emotion which can be applied to any arbitrary situation arising, from a political negotiation, to editing a Wikipedia page, to conducting a semiotic analysis in one's mental space.

Alas hope is not lost. Yerkes–Dodson law is a generalised 'theory'/law of learning predicted from stress (anxiety) response. Perhaps if we stratified the stress response of students in classrooms, athletes on the track, and other performance scenarios, then mathematically transformed the data to model the stress/anxiety to performance relationship, we may be able to classify emotions based on their impact on human performance in task where novelty, unpredictability, self-inefficacy or a threat of negative social evaluation can be predicted. I've been constructing exercises for myself and experimenting with this in cold approach, but this craft's community doesn't like that crowd, apparently. It's a fun exercise in theory of mind, IMO!

anxiety risk management

I've been toying with the idea of a risk management framework for rationalists.

The question is, how do you align your willingness to take risks with its ability to do so?

I frame risks in terms of threats to mental well-being, and figure anxiety is good catchall. Then, I practice defensive pessimism to identify what I'm willing to lose before my anxiety level rises to a point of internal dissent. Contrasting my current state to that counterfactual makes me grateful and therefore positive and optimistic. Then, I gamble all of that over a diversified portfolio of risky activities with the highest potential rewards I can muster. I try to convert ~70% of these rewards into non-property gains e.g. learning, happiness, relationships then reinvest the rest in future gambles in so far as my baseline anxiety tolerance level hasn't rises or fallen. Finally, I don't explain myself, or rationalise about past decisions. Why?


edit: hyperlinks fixed. Thanks for telling me. I added the www.'s originally thinking they were missing, but didn't test them out.

edit 2: for people this is useful for, I recommend trying out other techniques associated with generalised anxiety disorder and sports (performance) psychology:

To combat the previous cognitive and emotional aspects of GAD, psychologists often include some of the following key treatment components in their intervention plan; self-monitoring, relaxation techniques, self-control desensitization, gradual stimulus control, cognitive restructuring, worry outcome monitoring, present-moment focus, expectancy-free living, problem-solving techniques, processing of core fears, socialization, discussion and reframing of worry beliefs, emotional skills training, experiential exposure, psychoeducation, mindfulness and acceptance exercises

I try to use these myself!

You have two links ("cognitive appraisal" and "Yerkes-Dodson law") to www.en.wikipedia.org, which doesn't exist; they should go to en.wikipedia.org.

[-][anonymous]5y 0

This is interesting - is self-reporting a reliable way to measure stress levels, or would cortisol testing be the only way?

These days heart rate variance is likely the best way to measure stress levels.