How I Interrogate My Own Mind

A few weeks ago, I promised a more in-depth breakdown of how I graph and examine my own mind (by virtue of journal entries), and this week, I deliver! Since writing that article and showing my format to more people, I realized how rare the practice of metacognition (examining your own thinking) is.

I like to view my journaling as a pipeline, where the rawest inputs come in from the top; those constitute my relationship with the world - my most kneejerk sensory responses all the way to my most composed musings. Ultimately, this creates an attention circuit where the things I focus on have some background observation and improvement process, no matter what they are.

This, of course, creates a solid series of rich voice data: "how did I think about or respond to something?" If I pick out a point in time for a given journal entry, I can examine my thinking in a certain way, not unlike growth rings in a tree.

Now, each entry by itself is an atomic unit; it's a bit of data that carries no large macro correlation. Examining all of them in aggregate, however, I can find a lot of interesting patterns that I myself was not aware of.

How do I accomplish this? AI, of course! There are a lot of curious reasons people may love or hate AI - speaking as a software person myself, I recognize that an LLM is essentially the art of statistics applied to a natural language dictionary; not unlike our brains listening to someone at a party and figuring out what they meant (or sifting through several interpretations) while loud music plays in the background.

Specifically: I use Claude Code to examine my local files and surface different sources - I think an image would better illustrate how I actually interrogate my mind (this is a chat between me and Opus 4.5):

Because I have rich personal voice data spanning several years, I can conduct metacognition at scale. An LLM is also fundamentally more well-rounded than a singular human mind in how it infers language patterns - it will surface angles to me that I didn't think about before. I love the statistic that humans have at most 100 billion biological neurons by default; finding ways to escape this limitation is one of my life callings.

One example which enriched my journal entries: about half a year ago, I used Google Gemini's CLI tool to go through every month's journal entries, and for each one, create a hero's journey/story narrative arc. You'll notice in the image below there's a big central file to which a bunch of journal entries are linked: that's a summary file. Gemini created the file of its own accord, and after I reviewed/looked at it, I kept it in my journal (albeit separated from individual daily entries, for cleanliness).

One evolution of this practice I've thought about is a wearable microphone or audio pin that will capture my voice in full fidelity, for both transcription and archival purposes; it will serve as another source of rich personal voice data that I can run additional analysis on. As this personal surface area increases, I can do all sorts of in-depth personal self-study and metacognition study.

You might wonder: "Michael, what the hell is the point? Aren't some things better left to chance and discovery? Isn't this time you could spend just living life?" - And you might be right in one sense: after all, life is about discovery.

However, I don't know where technology is going, and I'm speaking as an optimistic technologist. What I do know, however, is that we are entering a future where the world will change far quicker than our own biology (for now), and mastering ourselves/our own systems is how we will maintain our humanity. That, and understanding the narrative our lives take on. The story we tell ourselves is important, too - and knowing how we want that story to evolve.

A favorite quote of mine:

What I will absolutely encourage if you have interest in this kind of system: adopt Obsidian (or some other editing format that doesn't lock you to a cloud platform), and start collecting data on yourself. Journal, see what the world gives you and how you respond to it. The Gemini CLI I mentioned above is free (though not as robust as Claude Code in my experience). Your eyes will open to your own system in ways you didn't think about. Interrogate and analyze your own data.

Metacognition in 2026 onwards is going to become so important, especially in our digital age - DePaul University has already released curriculum specifically on it. If you want an even nerdier reason: cybernetics is the field of circular feedback and recursion (not just the fancy robot stuff the term evokes in present science fiction). I can absolutely see these fields converging - what happens when our 100 billion neurons can and will eventually control quadrillions more?

Anyways, that's all for this week: reply with what you enjoyed or what you want to hear more of! If you want help setting this up, just reach out. :)

Be well,

Michael Kirsanov