The Center For Applied Rationality

This weekend I attended a workshop in Boston organized by the nonprofit Center For Applied Rationality. The workshops are focused on developing better understanding and control of our own decisions and behavior. Heavily leaning on cognitive science research, the CFAR teachers and community have developed a curriculum which seeks to teach and train useful skills, such as “how to make more accurate predictions, avoid self-deception, and use arithmetic in ways that better motivate you to action”.

It’s kind of like if you took the line of thinking introduced in Thinking Fast And Slow, combined it with a couple dozen hand-tested techniques and applications, enough to fill a semester-long college course… and then distilled it all into a weekend retreat.

Confession: I was highly skeptical of the usefulness of this retreat, coming into it. I suspected I knew a fair amount of the techniques to be offered through my own explorations in cognitive science and human rationality 1 — especially since I had read the LessWrong Sequences over the past year, and had made progress integrating the understanding into my life. Even with financial aid, the price point was high enough to make me seriously consider and reconsider my decision to go. Even after I committed, I strongly considered making up an excuse and avoiding the event.

Having done the retreat, I am now highly confident that it was worth the time and money.

The retreat was a four-day experience, with food and housing fully provided (hosted by the most gracious individuals working at the Friendly Crossways Retreat Centerin Littleton, Massachusetts). Most days had packed schedules of various types of lessons and group/pair activities, interspersed with short minute breaks and longer mealtimes.

Image credits to Anna Riedl and Jordan Tirrell

Each retreat is different as the curriculum changes. The overarching structure of this weeked was as follows:

  • Day 1: We learned what I’m calling “rationality primitives”: small pieces of tools and observations which are part of a general toolkit. We set up the System 1 vs System 2 model of cognition, where S1 is the lower, reptile mind-brain, and S2 is the higher level human brain. In a fully functional human being, we want both of these systems working in concert with each other to great success. This is nothing new to anyone who’s read Thinking Fast and Slow or the LW Sequences.One example of what I mean by “rationality primitives” is the idea of a Trigger Action Plan, or TAP. From the linked post, (which summarizes it well and bias-bustingly has no association with CFAR), “[Trigger action planning] sets triggers that later create spontaneous sub-conscious prompts to action.” In jargon, this is taking advantage of locational and situational sub-conscious ‘primes’, in order to prompt desired activity. In practice, this means selecting a desired action: “I’d like to remember to write in my journal nightly”, and linking it to a “trigger”, such as “when I put my plate in the dishwasher after dinner”. Mentally rehearsing “put plate in dishwasher, then go write in journal” sets up a sort of real-life if/then statement which programmers are familiar with. This S1 training tactic is simple enough, and interestingly something I had been doing in my own life without knowing the words/jargon for: I’d paired “do pushups” with “put towel on hook in bedroom after shower” as a way to remember to get those pushups in.

    In general, the primitives from the first day were not altogether too new or foreign, although it’s always fun to learn a couple new names for ideas I’d already been playing with. I soldiered on, knowing the cognitive danger ofthinking I knew everything already.

  • Day 2: We took what we learned the previous day, including the primitives, and combined them into higher order tactics. My cognitive-trap internal battle evaporated as we took the little independent pieces I had some familiarity with and built them into larger-scope cognitive tools and structures within which to build habits and get things done. We had lots of focused time on finding and working out fixes for “bugs” in our lives, and were strongly encouraged to take steps to execute those solutions immediately, rather than even waiting to get back to day-to-day life.The lessons of day 2 arced toward better understanding of urges and motivation, ways to break down large tasks and goals into smaller ones, mentally linking and remembering our goals in a visceral way in order to keep ourselves motivated, and more. One of the common trends I enjoyed was a more graph-y way of breaking down and analyzing ideas or goals (graph in the CS or Neuro sense, with nodes connected by directional arrows or inhibition lines).

    A particularly interesting tool we learned on day 2 was dubbed “Focused Grit”. The technique is when you have a problem to solve, to set a timer andactually spend 5 minutes on the clock solving the problem. It turns out we don’t sit down and genuinely think about solving problems, we tend to allow ourselves to bounce off them and procrastinate. It’s often easy to come up with 1 idea and then be done thinking about the problem. Actually spending 5 minutes is a different thing entirely — especially if you convince yourself (your System 1) that you have to solve the problem in those five minutes, there’s no going back to it afterward, and whatever solution you come up with is the final one.

  • Day 3: The structure for this day was largely taking turns teaching each other various aspects of the techniques learned over the previous days. This served (wonderfully) to cement the understanding and coalesce the initially-seemingly-separate techniques and ideas into a coherent model of understanding. The tone of the retreat was/is such that the techniques being taught are all attempts toward condensing personal experience and cognitive science into an art & science of rationality. The instructors were all very clear that this is an open problem: we do not have all the answers yet, and we should all be working on forming better models and teaching strategies. I must have received the question “What is wrong with this retreat, what would you change?” over 5 times during the weekend, all from the organizers and volunteers. By the 50-hour point in the weekend, it felt like everyone attending had been good friends for ages. Everyone I talked to was interesting and thoughtful, and we had plenty of group silliness during and after the scheduled course-time.

  • Day 4: More life debugging, with a focus on working on problems with partners (Pair Debugging) and groups. We focused on really using and applying the techniques learned over the past days. I had some incredibly personal conversations with a few of the people I paired with and had group time with, conversations I’ll remember for a long time (and longer, thanks to memory tactics and good note-taking habits espoused by this very workshop). We took time to ask ourselves (on the clock, 5 minutes!) a barrage of big, personal, questions: If your life’s narrative was a novel, are you doing what the protagonist would do? If not, why not? What are the things that are truly important, that you want (need) to be optimizing for? And so on. After dinner there were about an hour’s worth of “lightning talks”: quick, 5-minute talks given on various rationality-related subjects, by the participants of the retreat. I gave one titled “Language Hacking”, which focused on the importance of the words we use in daily life (The map is not the territory!), and drew heavily from the work of OG rational linguist Alfred Korzybski. The whole 90-hour weekend was endcapped by tons of socializing, an alumni party in the evening, and shenanigans late into the night.

The community alone is worth it: I had an insightful conversation with every single person I talked to, both during the planned interactions and while socializing.

CFAR is teaching and refining a curriculum based on insights from cognitive psychology and behavioral economics, designed to equip attendees with the tools they need to maximize their productivity and refine their decisionmaking abilities. It’s a great use of a weekend, if you can stomach the cost. Only time (and my own agency) will tell if the retreat has implicated lasting effects in my behavior, though this following week has been wonderfully productive, so I’m hopeful.

1 For those interested in reading materials on the subject, here’s a curated shortlist:
Harry Potter and the Methods of Rationality, aka HPMOR. This is the place to start.
Rationality: AI to Zombies, formerly the LessWrong Sequences. Read these next.
Judgement Under Uncertainty: Heurstics and Biases A classic of cog-sci.
Superintelligence: Paths, Dangers, Strategies. Reading this one currently.
Global Catastrophic Risks. Both educational and terrifying.
MIRI Research Guide. Book list recommended by MIRI

3… 2… 1… Victory!