Instrumental Rationality 4.2: Creating Habits

https://www.lesswrong.com/posts/vE7Z2JTDo5BHsCp4T/instrumental-rationality-4-2-creating-habits

Contents

**Techniques: Creating Habits

** *[The Techniques section has been broken up into sections: one on creating new habits and one on breaking existing habits. They are split up based on what they do rather than what they are because I think it makes more sense. * *Splitting things up by function allows the given examples for the techniques to be more focused on one of the two uses at a time, which I think makes the explanations easier to understand.] * As we touched upon in an earlier section, the process by which habits form, when simplified, looks roughly something like this:

**Trigger Action Plans (TAPs):

** *[TAPs are a way of framing intentional habit creation by focusing on both a salient trigger (i.e. context cue) and an action (i.e. response). They are well-backed by over two decades of research and build off the standard habit model.] * **Research: ** Trigger Action Planning is a technique by CFAR that is an adaptation of the implementation intention technique by psychologist Peter Gollwitzer. Implementation intentions are also known as "if-then plans" because they take the form of a conditional, not unlike those used in programming, where one thing follows another. Basically it’s a way of chaining together actions and situations. In writing, it roughly looks like: "I intend to do action X when I encounter situation Y!" As an example, a typical implementation intention for eating healthier might look like: "I intend to fill up half my plate with veggies whenever I am at a self-serve restaurant." Compared to many other behavioral change interventions, implementation have a stronger evidence base, and they’re simple to put into practice. In a meta-analysis of over 90 studies involving implementation intentions, we found that they "had a positive effect of medium‐to‐large magnitude ([effect size] = 0.65) on goal attainment" *23. Thus, as one would expect, research that shows people who make implementation intentions have a far higher success rate of achieving their target behavior compared to people who merely hold normal intentions (EX: "I intend to do X") *24. CFAR’s Trigger Action Plan model builds on the implementation intention model and tailors it to the standard habit model, focusing on selecting concrete context cues and specific actions †4. ++++

**Technique:

** TAPs combine what we know about the standard habit model of a context cue paired with a response with the if-then nature of implementation intentions. Trigger Action Planning has us specify specifically when and how we’d like to behave via a Trigger and an Action. It deliberately utilizes the same process that our habits naturally arise, allowing us to intentionally create new habits. A TAP takes the form of: When [Trigger X] happens, I will perform [Action Y]. Schematically, it’s almost identical to the implementation intention setup. But the TAP model stresses a few different factors. Here’s a step-by-step walkthrough of how to make your own TAP:

**TAPs Examples:

** To get more of a feel for TAPs, here are four additional examples of Trigger Action Plans that I use. I’ll admit that they don’t conform exactly to the ideal TAP model, but such is the process of translating from theory to practice. Still, I think they help give a better idea of what TAPs might look like in practice for you.

**To-Do List TAP:

** My memory isn’t perfect, and I often need to remember things that randomly pop into my head that take the form of to-dos. For example, "talk to Daniel", "send this email", or "purchase more pens". To help with this, I have the Todoist app downloaded on my phone, and I use this TAP: **Trigger: **"When I think of something I need to do…" **→ ** **Action: **"then I will open up Todoist on my phone and put it in." After doing this TAP a few times, it’s become far more automatic; once I think of something, my next logical thought is, "Okay, now let me put it into Todoist." While this doesn’t go all the way (after all, I still need to actually do the item), just having this TAP has made it far less likely that to-dos slip through the cracks in my mind. ++++

**Conversation TAP:

** In conversations, I’m not always on top of things. It can be easy for me to zone out, as I think of something, and then all I hear are meaningless word-shaped sounds. In the past, I’d often want to ask the other party to repeat what they just said, but by the time I finished deliberating whether or not to ask, more words would have been said, and I’d be even further behind in the conversation. Now I have a TAP for these sorts of situations: **Trigger: **"When I hear sounds that should probably be be words but which feel like they have no meaning…" **→ ** Action: "then I will immediately respond, ‘Sorry, my bad! I didn’t quite understand. Could you say that again using different words?’" In my experience, people are largely forgiving if you’re genuinely trying to stay focused on the conversation, and I’m all for little hacks that make communication better. A related issue for me is when someone says something that I understand on a S2 level, i.e. I know what all the words in the sentence mean, but I still don’t have a good idea of what they mean by such a sentence. In these types of situations, I have a similar TAP: Trigger: "When someone says something that doesn’t give me a clear mental picture…" → Action: "Then I will immediately say, ‘Sorry, I’m not sure I follow. What’s an example of that?’" Hopefully the Trigger here makes sense. I often think in terms of quick mental pictures. As the other person is speaking, I’ve got some visuals that correspond to the words they’re saying. It’s not exactly like I’m imagining a mental movie as they’re speaking, but it’s a little similar. Therefore, if someone says something that doesn’t lead to a mental image, this is a good sign for me that I probably don’t understand as much as I thought I did. Hence the Action in response to ask for more examples, which can hopefully make their point more concrete. ++++

**Journaling TAP:

** This TAP is a good way to remind myself to journal, and I’ve successfully been using a variant of it for several years; I’ve been journaling daily since about 2014. It’s an example of piggybacking in that it uses the ending of an existing routine as the trigger. Trigger: "When I finish brushing my teeth…" → Action: "I will go and write down today’s events in my journal."
Brushing my teeth is already an invariant in my schedule, so that makes it a good Trigger to build new routines off of. Like I mentioned earlier about TAPs being used to initiate action, the Journaling TAP is used to get myself in a position to start writing. Once I’m automatically in position, my more conscious thinking can take over, dictating what I actually write. The TAP is used to make sure I get there in the first place without much effort. ++++

**Actually Practicing: Make Your Own TAP

** Going over a technique like TAPs in your head can be useful, but the trick is to really use it in the real world. Thus, it’s very, very strongly recommended that you take the next 10 minutes to try out the following exercise:

**Systematic Planning:

** *[Systematic Planning is a synthesis of interventions based around planning and monitoring. It focuses on additional ways to increase habit strength and frequency by building off the TAP model.] *

**Research:

** Systematic Planning draws from both two interventions: action planning and active monitoring. Action planning, as the name very clearly informs us, is a form of planning *26. It involves trying to answer questions like "What barriers would prevent me from carrying out my intended behavior? How can I remove those barriers?" In essence, it’s a very top-down approach to figuring out behavior change. When you’re action planning, you’re literally just planning for the action. You’re trying to make the task as frictionless as possible for Future You by identifying potential problems and making contingency plans. Given that we’ve already covered how to plan better, we can apply those lessons here. All our previous tools like Murphyjitsu and Reference Class Forecasting once again come into play. (Five second summary: Murphyjitsu asks us to imagine the most common failure modes for our plans by first assuming that they’ll fail. Reference Class Forecasting says we should rely on past information to make more accurate predictions.) An example of action planning might look like this: You’d like to stop watching television. So you ask yourself whether or not you’d be surprised if you went all of tomorrow without watching television. Your gut says no because Game of Thrones is on. So you decide to hide your remote and check your surprise level again. Maybe this time you’d be more surprised if you found a way to still watch television. Maybe not. Either way, the point here is that action planning allows you to perform these constant feasibility checks to see if you’d really be able to perform the target behavior. Active monitoring refers to the act of checking in to track your progress on a habit and seeing how far along you are. Both action planning and active monitoring take a sort of outside view, where you’re evaluating things without taking part in them. In contrast, TAPs are about being able to respond in the moment; they provide more of an inside view. Compared to action planning and active monitoring, TAPs are more reactive than evaluative. For example, someone doing active monitoring on their vegetable-eating habit might spend some time every day tracking their meals. They might look at how many meals had vegetables, keeping track of their progress over time. Both action planning and active monitoring have been shown to increase uptake of desired behavior. Action planning has been shown to be effective in several studies, including longer-term effects *26. And a meta-analysis of monitoring in 138 studies showed that it had a small-medium effect size (0.4) on actual goal attainment *42. Basically, we have the unsurprising result that making plans and writing things down helps make it more likely that people actually get things done. ++++

**Technique:

** As a technique, TAPs play the role of answering the question "How can I form intentional habits?" Conversely, Systematic Planning as a technique is about answering the question "How do I make sure I actually use my TAPs?" As a result, Systematic Planning is less about providing an alternative model for forming habits. Rather, it’s about providing additional useful considerations when using the TAP model. This also means that it’s a combination of ideas rather than just one thing. First, though, there’s something interesting to note about how planning and habits, Murphyjitsu and TAPs in particular, are related: **You can run Murphyjitsu when making a new TAP, but you can also make a TAP out of Murphyjitsu. ** An example of using Murphyjitsu when making a new TAP might look like this: Say you would like to check email more on your phone. You make a TAP that looks like [Open phone] → [Check email]. Unsure if your TAP will be successful, you imagine that it’s a week later and you didn’t start developing your habit. Using Murphyjitsu, you ask yourself, "What is the most plausible reason that this TAP didn’t stick?" In response, your internal simulation of events tells you that it’s likely the Trigger wasn’t salient enough. So you update your TAP with a more specific Trigger. An example of making a TAP out of Murphyjitsu might look like this: Say that as you tell your friend what time you’ll meet them at the park, you pause—something about the situation feels odd. Internally, a TAP fires off: [Give a time estimate] → [Imagine one thing that might cause a delay]. As a result, you end up adding an extra ten minutes to your estimate to account for potential traffic. While what you’re doing isn’t exactly the whole Murphyjitsu process, you’re able to get most of the value by turning it into a quick TAP that habituates. So there’s something interesting going on here where you can feed one technique into the other and vice-versa. They complement each other in part because each process involves the other—planning well is a habit, but you can also figure out how to make your habits better if you do some planning. I bring this up to introduce the idea of meta-TAPs, that is to say, TAPs which are designed to help out your other TAPs. As a collection of considerations, I think that Systematic Planning is clearer to understand when the concepts are shown as practical meta-TAPs you can start doing right away. Thus, I’ll be framing Systematic Planning as a series of three techniques: Active Monitoring, TAP Everything, and Murphyjitsu Your TAPs. ++++

**Systematic Planning 1: Active Monitoring

** As we covered, actively monitoring the progress on your habits is a strong way to improve your habits. You’re checking to see whether or not you’ve been executing your TAPs and mentally rehearsing the Triggers. You’re taking the time to sit down and track where you are in regards to learning all of your TAPs. Like most other routine activities, Active Monitoring is probably best done piggybacking off an existing part of your schedule. I’d recommend the mornings as an especially good time, as you can review your TAPs once more before you go off on your day, where you’ll start to see all your potential Triggers. While everyone’s actual monitoring questions might be different, here is a sample set of questions you can feel free to use:

**Systematic Planning 2: TAP Everything

** By now, I hope it’s clear that the TAP framework is a simple and flexible way to frame behavioral change. Yet, thinking of new behaviors in this model isn’t always our default mode of thinking. We’re often thinking about just what we want to do, rather than the when, where and how. Having a structured way to make new habits like the TAP model is good, but you also need to get into a habit of actually using it. Thus, one way to make this process more habitual is to have a TAP designed to look for new opportunities to make TAPs. ("So I heard you like TAPs…") For example, say you’re talking to someone who opens up with a very nice conversation starter: Initially, you might think, "Oh, that was a pretty cool way to start things, I should try that next time." If you’re trying to hunt for practicality, though, you might want to think, "Huh, that phrase was a neat way to start things. Let’s make a TAP out of this. Maybe something like [Next time I meet a nice stranger] → [Say these words]…" This way, you’re turning your desires into actual actionables and habits. The important thing here isn’t to conform exactly to the specifics TAP model, but to get in the habit of viewing things in the "if-then" style, which makes it easier for you to actually execute on the actions you prefer. A "TAP Everything" meta-TAP might look like this: **Trigger: **"When I notice myself thinking, ‘I want to do X’…"
**→ ** **Action: **"Look for a specific Trigger to do X and turn it into a TAP." Basically, I’m suggesting you should make a habit out of making habits. ++++

**Systematic Planning 3: Murphyjitsu Your TAPs

** We touched upon this earlier in the example about how different techniques can feed into one another, and this section just states it a little more clearly. Murphyjitsu and the related set of planning prompts is very useful, and having them fire off when making a TAP can make them more robust and likely to succeed. Thus, we can also make a TAP for this! TAPs work best when the Trigger is clear and the Action is explicit, so most of the common failure modes are because either one or both of the two components are unclear. So an example "Murphyjitsu Your TAPs" meta-TAP might look like this: Trigger: "When I think of both the Trigger and the Action for a TAP…" **→ ** Action: "I will imagine that it’s one week later and I haven’t done my TAP at all. What are the first two failures that come to mind? How can I patch my TAP to fix them?" It’s less important that we plan for all failures and more important that we just do this in the first place, like in Active Monitoring. We can get most of the value by just checking for a few of the most common failure modes. This is where things get interconnected and fun: Once you’ve developed the "TAP Everything"-TAP, you now have a new routine action which would serve as a very good Trigger for your "Murphyjitsu Your TAPs"-TAP. The end goal here would be for "TAP Everything"-TAP and "Murphyjitsu Your TAPs"-TAP to end up becoming chunked together, into one unit. Research in chunking shows that actions habitually done together can end up forming a cached sequence, where one action follows another *27. So there’s some hope that we’ll eventually be able to "chain" habits together, in larger structures similar to how habits themselves consist of a context cue and a response. (Although this is more geared towards motor skills, it at least seems plausible that mental actions could work in a similar way. ) ++++

**Scaling Up:

** *[Scaling Up is one simple way to fight the "intention-action gap", the phenomenon where our desires and actions don’t align. It involves gradually building up an action so that at every step, it’s not too difficult.] *

**Research:

** Scaling Up is a technique intended to bridge the intention-action gap. The intention-action gap is a term used in the research to point at situations where we might fail to take action, despite holding the intention to do so *28. A typical example might be that of someone who wants to exercise more, knows it’s good for them, yet still doesn’t find themselves doing it. Or, the dieter who’d like to eat more healthily, is aware of the benefits, but still isn’t doing so. Why is this important when considering habits? Well, when forming TAPs, we saw earlier that they work best on actions you already *wanted *to do. But what about actions you know are good for you, but don’t really want to get done? This is where finding concrete strategies to bridge the gap are important. First off, it seems good to differentiate between roughly two relevant reasons here that a habit might not stick:

**Technique:

** In short, Scaling Up says if the action you’d like to turn into a habit is undesirable to you, start small and build up. You initiate with a watered down, doable version of the activity, and gradually scale up. The evidence base for this technique comes from a combination of shaping and exposure therapy †5. Shaping is, roughly speaking, the idea of gradually rewarding behavior that gets closer and closer to the target *31. With incremental changes each time, this allows for the building up of even quite complex behaviors. When applied to habit learning, this means eventually doing the hard action you felt aversive to in the beginning. A good example may be something like walking across a balance beam, where you don’t try to go all the way across on your first time. Perhaps you first start out by just standing still and balancing. Then you try to take 3 steps with good form before falling. Then maybe 6 steps. And after that maybe halfway across the entire beam. The point is that every step of the way, you’re working to an intermediate goal which brings you closer to the final step. You’re never biting off more than you can chew. One of the main lessons here is that expanding your expected timeframe for goal achievement can make scaling up to the hard thing more bearable. While this seems to be obvious for things like exercise, I don’t think people always translate this idea into other contexts like studying, which ends up with people doing things like trying to write an essay the night before. The other part of the technique draws inspiration from exposure therapy, the process by which consistently being exposed to aversive phenomenon can decrease the degree of aversion *32. Though this is speculative, it at least seems plausible that consistently successfully doing even an easier version of the hard or aversive action is useful. Doing so could provide additional experiential evidence that said action is not so bad after all, allowing your brain to update the negative feelings associated with the target behavior. If we go back to the balance beam example, walking only a few steps is likely lower resistance. Then, the next time you go to the balance beam, your locally cached feelings of "Hey, that wasn’t so bad last time!" can help make it a little more palatable as well. Taken together, these two parts form the evidence base for scaling up in level as a potentially viable solution for dealing with trying to habituate aversive behaviors. (Remember that there’s far more depth here, and the following technique merely scratches the surface.) As an actual technique, Scaling Up is about finding first a manageable chunk of the actual target behavior you’d like to habituate and then gradually working towards the goal. Step-by-step, it looks like:

NEXT ESSAY