I can share observations/thoughts about some similar experiences.
Sometimes I lie.
But let me back up: for a long time, I’ve been aware of two different modes in which I speak—one is fluent and “real-time”; the other is slow, halting, and feels somewhat like wearing mittens, somewhat like trying to squash high-dimensional objects down into a lower-dimensional space, somewhat like.. tasting alternatives for the right connotations and inflections. Sometimes this second mode is evoked when some part of me becomes concerned about doing PR with the person I’m talking to, or if its a topic where I care about nuance or precision. I’d estimate that ~80% of my speech is in the fluent mode.
A few years ago (a few years into my meditation practice) I realized that speech in the fluent mode is, in some, way non-conscious. Sometimes I could become aware of speech just… unspooling through me, without intervention from… “me”. The part of me that calls itself “me”. The “global work space” of consciousness. Sometimes it was sophisticated speech! This has never happened in mittens-mode. Needing to sieve up a new/novel/unusual semantic meaning from the depths of language-space seems to require… “me”.
Once, fluent-mode upset someone and I tried to explain this to them. It… wasn’t well-received.
More than 99.99% of the time, I’m entirely happy with what, fluent mode utters. Very rarely, though (probably on the order of once every year or two) it will just lie. “How do you know X?” “Oh, through my dance community” <I actually met them through OkCupid>.
This tends to happen when I sense that the other person will get a negative emotional charge out of the truth, and/or if I have some negative emotional content around it. @AnnaSalamon, you gave me the model I use for thinking about this years ago: the parent who on picking up the phone angrily asks “Why don’t you call more often?!” and then wonders why their child doesn’t call more often. If a person has shown me systematically in the past that they will have a strong negative emotional reaction to the truth, then I powerfully learn their revealed preference that in some situations they hate the truth. I also learn an upsetting reciprocal lesson about myself and find myself in an awkward to back out of situation.
These moments are particularly vexing because they’re extremely difficult to “train against”—I have to be caught unaware, and in the right mental state for this to happen.
(Conversely, sometimes the babble system which Scott dubs the guf isn’t generative enough in real-time. There have been various babble exercises posted here in the past year; other ideas I’ve had include learning to rap (badly! Badly! I don’t expect to learn to rap well!) and trying comedy or theatre improv (Keith Johnstone’s Impro seemed to be recommended a lot in rationalist circles a few years ago). One thing I wonder about: will training the guf to be more generative increase the risk of the guf make utterances I wouldn’t fully endorse?)
I can share observations/thoughts about some similar experiences.
Sometimes I lie.
But let me back up: for a long time, I’ve been aware of two different modes in which I speak—one is fluent and “real-time”; the other is slow, halting, and feels somewhat like wearing mittens, somewhat like trying to squash high-dimensional objects down into a lower-dimensional space, somewhat like.. tasting alternatives for the right connotations and inflections. Sometimes this second mode is evoked when some part of me becomes concerned about doing PR with the person I’m talking to, or if its a topic where I care about nuance or precision. I’d estimate that ~80% of my speech is in the fluent mode.
A few years ago (a few years into my meditation practice) I realized that speech in the fluent mode is, in some, way non-conscious. Sometimes I could become aware of speech just… unspooling through me, without intervention from… “me”. The part of me that calls itself “me”. The “global work space” of consciousness. Sometimes it was sophisticated speech! This has never happened in mittens-mode. Needing to sieve up a new/novel/unusual semantic meaning from the depths of language-space seems to require… “me”.
Once, fluent-mode upset someone and I tried to explain this to them. It… wasn’t well-received.
More than 99.99% of the time, I’m entirely happy with what, fluent mode utters. Very rarely, though (probably on the order of once every year or two) it will just lie. “How do you know X?” “Oh, through my dance community” <I actually met them through OkCupid>.
This tends to happen when I sense that the other person will get a negative emotional charge out of the truth, and/or if I have some negative emotional content around it. @AnnaSalamon, you gave me the model I use for thinking about this years ago: the parent who on picking up the phone angrily asks “Why don’t you call more often?!” and then wonders why their child doesn’t call more often. If a person has shown me systematically in the past that they will have a strong negative emotional reaction to the truth, then I powerfully learn their revealed preference that in some situations they hate the truth. I also learn an upsetting reciprocal lesson about myself and find myself in an awkward to back out of situation.
These moments are particularly vexing because they’re extremely difficult to “train against”—I have to be caught unaware, and in the right mental state for this to happen.
(Conversely, sometimes the babble system which Scott dubs the guf isn’t generative enough in real-time. There have been various babble exercises posted here in the past year; other ideas I’ve had include learning to rap (badly! Badly! I don’t expect to learn to rap well!) and trying comedy or theatre improv (Keith Johnstone’s Impro seemed to be recommended a lot in rationalist circles a few years ago). One thing I wonder about: will training the guf to be more generative increase the risk of the guf make utterances I wouldn’t fully endorse?)