I have tried stack tracing my own brain on things like ugh fields and they mostly come out of procedural uncertainty. On the other hand, cravings have so far resisted any attempt at deconstruction.
Does anyone have a sharper inner view?
The second taxonomy along with dropping of the intention to fix is what worked well on cravings for me. I told myself that this is a long term project and so the need to alter cravings could come later, I just wanted to see the process clearly. The most obvious were cravings for certain foods, where the event was a mental image along with some sort of echo of the taste/texture followed immediately by associations of where the closest place I could obtain it are.
Hmm. Sounds like Open Monitoring (meditation) would have this as a side-benefit. Hypothesis: increasing non-judgmental awareness would lead to increased self-regulatory capacity. Perhaps consistent practice of any discipline that improves awareness would have a spillover effect?
"A few minutes a day for 3 weeks caused a noticeable effect that has endured."
As in you set the intention to catch this sort of thing, and then that ended up happening a few minutes each day?
Or you spent several minutes each day deliberately practising it somehow?
Deliberate practice. Though there were some free flow through spontaneous ones after the first week or two.
Would you mind giving an example or two of how you've done deliberate practice? I always have trouble finding ways to exercise mental skills when I'm not being forced to use them.
At first I set a 2 minute timer, chose one particular type of frame and just noted each mental event that matched. So Mental imagery was just noting "image" each time I caught a mental image. Sometimes they are much too fast or changing as you try to catch them, that's fine. There was no sense of rushing to keep up or worry if one didn't arise for awhile (though that sometimes seemed to warrant noting as "blank image"). I increased slowly to 5 minutes and did the other types of frames. After some comfort was built I tried multiple types at once, "image", "talk", "image", "feeling" etc. just noting as they arose. I did this for 1 month.
I used various habit type loading things to maintain the habit for a month, then practiced it more sporadically since then. The benefits seem to be permanent.
This resonates; I think it points at a similar "looking deeper" thing I've been doing, but you've given it a label, which is nice. Thanks for the post.
In my experience, applying this to ugh fields results in similar Focusing-esque shifts.
Related: The 5-Second Level; Attention control is critical for increasing/altering motivation
Epistemic Status: sharing a hypothesis that's slowly been coalescing since a discussion with Eliezer at EAG and got catalyzed by Anna's latest LW post along with an exercise I have been using. n=1
Mental phenomena (and thus rationality skills) can't be trained without a feedback loop that causes calibration in the relevant direction. One of my guesses for a valuable thing Eliezer did was habitual stack traces causing a leveling up of stack trace resolution i.e. seeing more fine grain detail in mental phenomena. This is related to 'catching flinches' as Anna describes, as an example of a particularly useful phenomena to be able to catch. In general, you can't tune black boxes, you need to be able to see individual steps.
How can you level up the stack trace skill? Triaging your unwillingness to do things, and we'll start with your unwillingness to practice the stack trace skill! I like 'triage' more than 'classify' because it imports some connotations about scope sensitivity.
In order to triage we need a taxonomy. Developing/hacking/modding your own is what ultimately works best, but you can use prebuilt ones as training wheels. Here are two possible taxonomies:
Note whether it is experienced as
Note whether it is experienced as
Form the intention to practice the stack trace skill and then try to classify at least one thing that happens. If you feel good when you get a 'hit' you will be more likely to catch additional events.
You can try this on anything. The desire for an unhealthy snack, the unwillingness to email someone etc. Note that the exercise isn't about forcing yourself to do things you don't want to do. You just want to see more clearly your own objections to doing it. If you do it more, you'll start to notice that you can catch more 'frames' or multiple phenomena at the same time or in a row e.g. I am experiencing ambiguity as the mental talk "I'm not sure how to do that" and as a slightly slimy/sliding away sensation followed by aversion to feeling the slimy feeling and an arising distracting desire to check my email. Distinguishing between actual sensations in the body and things that only seem like they could maybe be described as sensations is mostly a distraction and not all that important initially.
These are just examples and finding nice tags in your own mentalese makes the thing run smoother. You can also use this as fuel for focusing for particularly interesting frames you catch e.g. when you catch a limiting belief. It's also interesting to notice instances of the 'to-be' verb form in mental talk as this is the source of a variety of map-territory distinction errors.
There is a specific failure worth mentioning: coming up with a story. If you ask yourself questions like "Why did I think that?" your brain is great at coming up with plausible sounding stories that are often bullshit. This is why, when practicing the skill, you have to prime the intention to catch specific things beforehand. Once the skill has been built up you can use it on arbitrary thoughts and have a sense for the difference between 'story' and actual frame catching.
If other people try this I'm curious for feedback. My experience so far has been that increasing the resolution on stack traces has made the practice of every other mental technique dramatically easier because the feedback loops are all tighter. Especially relevant to repairing a failed TAP. How much practice was involved? A few minutes a day for 3 weeks caused a noticeable effect that has endured. My models, plans, and execution fail less often. When they do I have a much better chance of catching the real culprit.