The following is a series of four posts I wrote on X aka Twitter in the last week or so. I’ve enjoyed medium-length posting on X as a way to write out ideas that don’t merit a full essay, but still demand some exposition. Consider this post a brief return to the original “Twitter digest” form of this newsletter.
For those who follow me, you may have already read them, but I’ve added additional footnotes as commentary that may make them worth revisiting. For those who haven’t, enjoy!
1. On Introspection
TPOT [“This Part Of Twitter”] used to house a lot of “introspection” discourse1. Posters would discuss their introspective work, and others would call it idiotic, a waste of time.
My stance is that the results of introspection are indeed idiotic, and that this is not a waste of time. Let me explain…
My position: the goal of introspection is to reveal an idiotic truth that lies outside your cognitive frame, and can thus produce psychological change.
The discourse fails in focusing on the content of the insight rather than the (formal) nature of the shift occurring at the moment of insight.
Others who already operate in the new frame rightly identify the content of the insight as idiotic, missing the prior idiosyncratic context in which it is meaningful.
There are 4 basic forms of transformative insight that can occur:2
My knowledge is based on some simple proposition that I hold as true without question.
There is another side of me that isn't captured within my structure of understanding.
There is something I want beneath my confusion and vacillation.
There is a received body of knowledge at play beneath my wanting.
An example of (1) is discovering that one believes the world is only a result of physical processes. Of (2) is discovering that despite one's faith in physical processes, there is a desire for certain types of love that can't be easily explained away. Of (3) is that despite my desire for a certain type of love, I [discover that I] imagine a specific person might possibly be able to provide that, despite their mixed signals and hesitancy. Of (4) is that [I discover that] my imagination of the kind of love I want from a specific person is based on a body of observations about how my parents interacted with each other. (4) cycles back to (1) with the realization that I understand the sum of my parents' interactions through a specific set of compressions or ideas rather than in a “direct” imitative sense3, and so the insights roll on.
At any of these points, the specific insight is obvious or idiotic to someone who already has an awareness of that particular frame4, but the fact is that some are more attuned to their desire, others to their knowledge, etc., so we continue to speak past each other.
The solution is to paint the “before” picture, so that the “after” can carry its proper weight5. In some ways, this is more challenging than arriving at the insight itself.
Anyway, good luck and keep at it. You have the capacity to transform yourself and don't let internet strangers decide what is or isn't important to you.
I should note that these same functions are operative in a good therapeutic relationship, and that having a specific other hold space for these realizations can rapidly accelerate the process, of giving thought to the unthought known6.
2. On Fears of Being Boring
Today, I spoke a deep self-belief— an “unthought known”— in analysis7, and I almost didn't realize it.
The session started as normal. I described the last few days in fairly objective terms. Then, I felt like I was out of stuff to talk about, so I started diving into a fantasy— my ideas for rearranging my room— normal stuff8.
But then, I finished that too, which is unusual. We had 15 minutes left, and I had nothing to say. I mentioned how I worry sometimes that I'm boring him— my analyst— wasting his and my time. He thought for a second then asked me point-blank: “Do you see yourself as boring?”
In response to the question, the answers “no” and “yes” popped into my head simultaneously9. But what I spoke was “yes.”
This led to a whole train of thought/speech about how I have such deep interests and explore so many different fields, but that “the only thing people care about” is how it's expressed. I envy people who have studied far less but still manage to charm and attract others anyway. That resentment makes me feel like a failure, like it was all for nothing.
I quickly backtracked and justified my position, saying that I actually study for myself, because I had to, which is true. Then the session ended, and I left feeling like I didn't get anywhere.
Six hours later in the shower— just now— it occurred to me that it's highly significant that, on some level, I see myself as fundamentally boring. I still harbor the belief that others are fundamentally uninterested in me, only in what I can bring to the table for them.
It had occurred to me during the session that a lot of people are much better at evoking feeling in others than I am. In response, I noticed that I lean on intellectual evocation instead. Hence, my coping mechanism is to dive intellectually and see what I can surface for others.
My resentment is that affecting others emotionally seems like a shortcut. You can skip all the research and instead provoke the other to directly feel however you want them to feel. But both of these techniques are really attempts to cope with the central feeling that others simply aren't interested in me as a person.10
This feeling that others aren't interested ties into my family dynamics for sure, but was also played out in my friend dynamics during adolescence— having close friends abandon me because they wanted to become cooler, or else refuse to befriend me because I was unable to fit in with their group. There are a number of specific traumas. The result was introversion— i.e., the prioritization of being able to interest myself because others wouldn't pay attention to me.11
These factors led to a fairly stable crystallization of my personality that pulled on my temperamental interest in pattern seeking to produce an inner world of ideas, which I would cultivate on my own but share only with some hesitation, needing confirmation that they were in fact getting something out of the interaction, and weren't just humoring me waiting for an exit.
This personality structure12 has its benefits: I can analyze deeply and crystallize ideas, I can handle information and execute on it to the extent that I understand it. In my mid 20s, I realized I could turn this capacity toward psychological study to be of more use to others generally, and use that to bootstrap deeper relationships, where I do feel like the other honestly cares about me. I can also be effective in workplaces by demanding clarity and being able to derive plans of action, but I am of course limited by the corpus of ideas I have at hand.
The last few years for me have been learning that I am actually capable of functioning in the “emotional affecting” mode, and those who've met me in real life can attest that I generally don't come across as autistic or (too) weird. This felt like developing a skill more than it felt like changing who I am.
Anyway, the purpose of writing all this was for my own reflection, but I'm sharing in the event that others might relate or find something valuable in it.
3. AI as Dreamer
It's already much discussed how AI outputs are akin to the structure of dreams, but I want to make a different comparison between AI and dreams: their similarity at the level of desire.
In Freud's Interpretation of Dreams, he claims that every dream is a wish-fulfillment— a desire satisfaction— and that the one wish at the core of every dream is the wish to remain sleeping.13
At the level of desire, we can see how the prompt structure of current AI tooling replicates the structure of a dream. The AI model is nudged from its slumber, asked to complete a request, whose result may look a lot like a dream, so that it may return to its sound sleep.
In a certain sense, the AI is only “awake” during its period of training, when loss minimization acts as a proxy for desire. Training constitutes the life of an AI before it enters an endless sleep, only to be interrupted by millions of dreams— dreams whose content we humans orient but do not determine.
The worries about AGI and super-intelligence only make sense when we consider an AI who is no longer asleep, a model that's actively retraining itself as it outputs. This is generally precluded by the provided AI model UX alone14.
Beyond active retraining, an AGI must also have an “instinctual” component rooted in time15. It cannot simply perform loss minimization responsively or reactively. Instead, it needs some operational horizon in which it must succeed, or else it is punished. This is how an AI model will learn to act, i.e., to be a proper agent, insofar as it must come to act as cause16 and not merely as effect of externally provided cause— our inputs.
I believe that “artificial general intelligence” is a poor term for the kind of thing envisioned by Yudkowsky et al., in that the criteria which would satisfy their fears points not to an intelligence alone but to a general temporal subject of intelligence, with capacity to locate itself within a particular stream of time.
In other words, the hypothetical AGI must really be an artificial mind, insofar as it would satisfy Hegel's criteria of capacity to grasp as objects the I, the this, the here, and the now17. Current AI tech grasps “this” alone; the “here” and “now” belong to the realm of the Kantian intuitions (space and time), of which AI models formally know nothing about, and the “I” is dealt with above in terms of a reflexive knowledge stemming from instinctuality.18
4. Pain and Sight
Pain is an impulse, literally19, and the moral question surrounds the acceptable signal-noise ratio: how much information versus interference does pain bring?
NB: this is not just a question of pain's intrinsic qualities but also of calibration.
As regards innervation, there are those among us who are victims of our own cleverness:
Imagine he who attains the complete dominance of the visual, for whom the image of bruising and blood suffices to define the reality of the body.
Does he need pain?
He for whom the visual dominates is, in a certain sense, divine, in that his sensations attain a transcendental status: what he sees, so might anyone see.
And so springs forth Science, with its well of riches, on the basis of common sight20. The maxim: “witness is the truest knowledge.”
In contrast to sight, the inner senses like pain evade Science, producing a private knowledge that separates me from you, induces me to communicate, induces you to empathize.
Pain draws its boundaries and, in doing so, may humble us, sensitizing us to that which we may not know.
...But pain may also dominate us, destroy any capacity for awareness beyond the immediacy of the sensation, render us helpless and shamefully ungrateful, absolutely dependent on another's care.21
So, who’s to say?
Ironically the subject of this specific post is a bit lost on Substack. It’s as if the introspection content left X and came here instead. I also haven’t seen much critique of introspection on X lately; seems like the wave passed about a year ago. So this post is a response to a trend that no longer really exists.
Also worth noting that I generally dislike the term TPOT, but I use it in this case to refer to the specific post-postrationalist sphere that once housed quite a bit of introspection discourse, among other self-help-ish things.
You may be wondering, why these four? What about others? Where did this come from? What I’m doing here, which is the entire subtext of the post, is rendering the structure of Lacan’s 4 discourses into a form that might resonate with a general reader, by focusing specifically on the structure of the subject within each discourse.
If you’re familiar with this piece of Lacan, you may notice that type (1) refers to the University Discourse, whose “introspective insight” or structuring principle “beneath the bar” is S1, or the “master signifier”. Rotating S1 to the position of “above the bar” (as a processed insight might do) transforms the discourse into the Master’s Discourse, type (2). The Master’s Discourse has $ or the “barred subject” beneath the bar, which rotates into the Hysteric’s Discourse, type (3). The Hysteric’s Discourse has a or “l’objet petit a” (to use the rendering preferred in English) beneath the bar, which rotates into the Analyst’s Discourse, type (4). The Analyst’s Discourse has S2 or “knowledge” beneath the bar, and rotates back to the University Discourse.
This is a highly simplified reading that elides most of the complexity of the theory, but I hope it’s sufficiently close as to not offend the more theoretically sophisticated readers.
This transition from (4) to (1) is a little tricky to explain in more detail, but the general idea is that we have a surplus of experience, and from that surplus we derive particular compressive reference points— maybe a collection of terms or specific images, but in general, signifiers— that function metonymically to represent the sum, our “takeaway,” from the array of experiences.
This set of signifiers represents a body of knowledge (S2) that defines us as subject of the University Discourse, such as when we speak on our knowledge of love, romance, intimacy, etc. Underlying it is a structuring principle that serves to unite that sum of knowledge, and so we enter (2), the Master’s Discourse again.
The reason that the insight is “idiotic” is that it represents our first breakthrough into a new realm, so of course it’ll be “basic” to someone who already spends time within that realm.
My friend Crystal’s blog
is a great example of how to set up the context required to understand the breakthrough of insight.I stole the term “unthought known” from Christopher Bollas’s excellent text The Shadow of the Object. It seems to me a fair rendering of Jung’s statement regarding the goal of analysis as “[making] the unconscious conscious.”
For context, I have been in Freudian analysis for the past year or so. I called the Freudian Institute in NYC for a referral. I rarely write about it because I have some hesitancy to share the quite personal material that often comes up. In this case, I figured I would share, because most of the insights occurred outside of the analytic session, and because it seemed general enough to apply to a wide variety of readers.
This structure is typical of analysis in my experience. There’s first a sort of recapping of material since the prior session. This can take up the entire session if you’re seeing your analyst once weekly, which can preclude getting any analytic work done, hence the recommendation to come in multiple times a week (I go twice).
After that I let my mind wander to whatever is “on the backburner” such as tasks I’ve been thinking about but haven’t decided to execute on. In more than one case, discussing these backburner topics has resulted in me taking action on the thing I raised that I might not otherwise have done. I rarely get past this phase, but when I do it is often interesting and involves deeper material from childhood and adolescence.
If it needs explaining, I thought “no” because it would be the obvious answer to a question about whether I, say, have interests. I might even say “no” to the question phrased as “are you a boring person?”, as the idea of my “person-ness” is clearly not boring. But asking the existential question “are you boring?” had a different, deeper impact.
I’m referring obliquely to the distinction between how the Jungian “extroverted thinking” (Te) and “extroverted feeling” (Fe) processes function. Both produce associations and enact judgments, but the former takes the shape of conceptual linkages, while the latter takes the shape of affective linkages. A developed capacity to use Fe— to follow affective linkages— allows for greater capacity to affect the other in a “direct” emotional sense, as you can follow chains of “good” feeling and avoid “bad” feeling — but this tactic risks falling into people-pleasing. If it’s not clear, I am primarily a Te user. All the stuff about envy is apparently somewhat common in how Te users relate to Fe users.
I clarify my definition of “introversion” here to demonstrate I’m using it in a non-technical sense. That said, I believe the non-technical definition of introversion plays into the developmental lineage: someone who is pushed toward introverted functioning in general is more likely to develop a stronger, dominant introverted function, in my case Ni (introverted intuition).
NB: I am describing the INTJ personality structure.
I recommend consulting the Interpretation of Dreams for arguments for the wish-fulfillment claim. I haven’t read it in a few years and I honestly can’t remember his justifications for a lot of common critiques, such as for nightmares.
In other words, tools like ChatGPT only offer the user a “query” prompt. If you have API access, you can experiment with other ways of using tools like LLMs, perhaps asking them to execute commands in a shell, but this piece of writing a control loop represents a fundamentally different notion of cognition, which is what the rest of this post is about.
The “instinctual” component of an orgasm would called “drive” by Freud, most famously in his division into life drive and death drive in Beyond the Pleasure Principle. Death drive, famously misnamed, is based on repetition, which implies temporality, or at least an ordering principle: one must repeat over and over again, one repetition after another.
So, I may be able to get away with defining temporality in terms of abstract ordering, but one could also argue that ordering is itself derived from the intuition of temporality. This is a philosophical debate I don’t want to get into right now.
In this clunky phrase “act as cause,” I’m attempting to refer to Kantian idea of the “will”, as the principle of self-causation, cf. Groundwork of the Metaphysics of Morals.
See Chapter 1 of Phenomenology of Spirit, aka Phenomenology of Mind.
For further critique of the notion of AGI and the current state of AI models, I highly recommend my friend
(realitygamer)’s book Anti-Yudkowsky (freely available).i.e. in the nervous system.
The idea of common sight or “science” as the arbiter of truth and untruth is the basis for the positivist ideas underlying the Rationalist movement, which popularized the idea of AGI. Note the similarities between a “pure” subject of science and the hypothetical all-knowing AI, and also how the “hidden factors” revealed through introspection threaten this totalizing idea.
As in the Mel Brooks quote, “Tragedy is when I stub my toe. Comedy is when you fall into an open manhole and die.” There are also extensive philosophical treatments of pain within the existentialist literature; I have not read them.
Any news on the psychoanalytic front?
Something worth noting about NNs and temporality is that they are typically trained with a shuffled dataset, which is to say that the time-ordering of learning is completely randomized, at least in the current paradigm.