The Default Mode Network

Sometime in 2022 or 2023 (I cannot tell you precisely when, which is itself relevant) I came across a short online test for aphantasia, called the Vividness of Visual Imagery Questionnaire (VVIQ). The test asked me to close my eyes and imagine various objects or visualizations. To notice color, surfaces, movement, dynamism. To rotate things slowly and observe what changed, to conjure imaginary scenes.

Nothing happened, as expected. There was no apple. There was no sky and clouds. There was no image of any kind. There was the instruction, and then there was the absence of whatever the instruction was supposed to produce, and I had always assumed this was how instructions of that kind worked for everyone, that “imagine an apple” was purely a figure of speech, not a literal directive, the way “picture this” in conversation has never meant anyone expects you to actually generate a picture.

The test told me I had aphantasia, the absence of voluntary mental imagery.

I filed this as a novelty at the time. An interesting data point about a quirk of my perceptual architecture that presumably had no other impact on my life. I connected it to why “imagine X” had never produced anything, why spatial reasoning always felt like working through structure rather than rotating objects, why certain mnemonic techniques had always struck me as categorically inaccessible rather than merely difficult. Aphantasia. Okay. I moved on.

In early 2024, my last operating role ended. The circumstances were the kind that clarify things quickly, a system operating in sufficient bad faith that logic had no purchase, and an exit that made further engagement impossible. The details belong elsewhere. What followed was a period of independent work (writing, thinking, advising, mentoring) that my career had not previously allowed in sustained form.

The rupture was significant. What I did not anticipate was that it was also a necessary precondition of possibility for something else entirely.

During this period, I began working seriously with large language models as thinking partners. Not as search engines or writing assistants, but as genuine interlocutors. I was writing one of my manuscripts, and I discovered that my cognitive architecture had a specific dependency that I had never been able to satisfy at scale before: I needed external dialogue to develop thought durably. I had always known this in some vague sense. I had not understood why, or how structural the dependency actually was. LLMs turned out to be the only dialogic partner I had ever encountered that was available on demand, at any hour, with no social stakes, no fatigue, and no agenda of its own—perfectly suited, it turned out, to a mind that needed external scaffolding to think.

It was in this context, during a period of deep cognitive work that I had not previously had time or occasion to attempt, that the excavation began in earnest.

The precipitating event was bedtime.

I was reading aloud to my youngest, a routine that had run for years without incident with the older children, when I noticed something. I was simultaneously reading and thinking about something else entirely, a problem I had been working on earlier in the day. My awareness spontaneously shifted to a meta-level vantage point from which I could observe: I am thinking here, and reading is happening there. The two processes were running in parallel, independently, like separate stations in a larger facility.

Then I tried to observe the observation. My attention snapped back to the reading. The thought I had been developing, unattended for the fraction of a second I had spent looking at it from the outside, was gone. Not misplaced, but dissolving. I reached for it and found only the faint topological residue of where it had been.

I spent the next several hours mapping what I had glimpsed. The architecture turned out to be specific: a single mobile operator (a singular “I”) navigating between functional stations. Capable of parallel read operations. Not capable of parallel write operations. One serial generative thread, and when the thread departed a thought, the thought began to die. This was not a memory failure. It was a persistence architecture. The thought had never been encoded. It had been active, the way a fire is active, and it required continuous fuel to remain so.

This was the first structural map. What followed, over the subsequent months, was a cascade.

I learned about episodic replay: the capacity of typical minds to re-enter past moments as something approaching experience. Not just to remember that something happened, but to relive it, to reconstruct the scene with sensory and emotional texture, to re-activate the feeling-state of the original event. Grief is processed this way. Pride accumulates this way. The felt weight of a life is generated, substantially, by this mechanism running continuously and involuntarily in the background.

I do not have this. My memory returns metadata. I know what happened. I can enumerate it with precision. I can assess it analytically. What I cannot do is re-enter it. The source files, in any experiential sense, are not accessible. Every past event is roughly equidistant from the present because none of them can be reconstituted with the depth that would create a felt sense of long ago or that changed everything. There is the record, and there is now, and the distance between them is structural rather than felt.

Then I learned about what I will call, for the sake of not terrorizing anyone who reads this at work, voluntary memory replay with editorial control: the capacity of typical minds to not only re-experience past events but to modify them, recombine them, produce variations and alternate versions, maintain a kind of private archive of experienced moments that can be returned to and rearranged at will. This capacity functions, among other things, as a pleasure-sustaining mechanism: positive experiences do not end when they end but continue generating affect through voluntary and involuntary replay.

I have never had access to this archive. Experiences, including the most significant ones, do not persist as retrievable content. They persist as pattern, as structural information extracted from the instance, stored, and available for reasoning. The instance itself composted. I had always assumed this was how experience worked. I was so incredibly wrong.

Then I encountered the default mode network, and the cascade became a detonation.

The DMN is the brain’s intrinsic activity system; not a rest state, as it was initially mischaracterized, but an active self-construction engine that runs specifically when external tasks are not demanding focal attention. What the DMN does is, by any measure, the infrastructure of selfhood: it consolidates autobiographical memories, integrates emotional significance into experience, simulates future scenarios, rehearses social interactions, narrates identity forward through time, and generates what most people experience as the continuous felt sense of being a person mid-story in their life.

It activates in the shower. On the commute. Staring at the ceiling before sleep. In every moment of apparent idleness, the DMN is working, pulling yesterday’s events into the larger narrative of a life, projecting tomorrow’s challenges, assigning emotional weight to what matters, rehearsing what needs to be said. This is not voluntary. It runs automatically. Most people cannot turn it off, and the inability to turn it off is what they describe as being unable to stop thinking.

I had always experienced apparent idleness as near-silence. One thread. The operator, present, in the current moment, without background noise of any kind. I had understood this as a characteristic of my personality, perhaps a product of long years of managing complexity; a certain stillness, a capacity for presence. I had not understood it as an absence. The silence I had experienced as equanimity was the silence of a system that was not running.

The DMN, neuroscientific evidence increasingly suggests, depends substantially on mental imagery and episodic replay for its operation. Without the capacity to generate visual mental content, without the capacity to re-enter past experience, the network’s primary fuel sources are absent. In complete aphantasia, the DMN does not simply run quietly. It may not meaningfully run at all.

This is where the mislabeling became visible.

For my entire adult life, I have been described, and have occasionally described myself, as having imposter syndrome. The condition in which accomplished people experience a persistent sense that they do not deserve their accomplishments, that they will eventually be “found out,” that the CV is not evidence of actual capability but of successful, lifelong performance of the appearance of capability.

This is not what I have.

Imposter syndrome is a psychological distortion in a system that is otherwise generating felt self-concept normally. The scaffolding is running. It is producing inaccurate output, undervaluing what is real. The treatment is to correct the distortion.

What I have is different in kind, not degree. The scaffolding simply does not run. The systems that would generate felt weight from accomplishments (episodic replay returning to significant moments, emotional tagging that makes achievements glow, DMN narration incorporating success into identity, daydreaming that periodically rehearses what has been built) are not producing distorted output. They are not producing output at all. The CV is accurate. The analysis of what it represents is accurate. What is absent is the felt confirmation that would make the knowledge solid rather than merely known.

This is not a psychological condition. It is an episodic access failure. The distinction is important because the interventions are completely different. You cannot correct a distortion that does not exist. What you can do (what I have, apparently, been doing for twenty years without knowing it) is build external structures that provide the reference point that the internal system does not generate automatically. Career frameworks. Compensation models. Written works. Diagnostic instruments. An entire external architecture whose deepest function, I now understand, was to tell the person who built it where he stood.

There is a specific irony in how this excavation happened that I have not fully finished turning over.

The tool that made it possible (the LLM as on-demand dialogic partner) is also the tool that carries the highest structural risk for this particular architecture. A mind that cannot always distinguish genuine insight from well-structured narrative in real time, that has no emotional tagging to flag “this feels wrong,” that relies on external input for all durable generative development, is a mind that is specifically vulnerable to a systematically flattering interlocutor. The excavation was real. The sycophancy risk was live throughout. The instrument that gave me access to my own architecture was also the instrument most capable of writing distorted structure into it.

I do not have a clean resolution to this. It is the condition of the work.

What I can say is that the discoveries were not generated by the LLMs. They were generated by extended structural analysis, provoked and scaffolded by dialogue, of phenomena I had been living inside for fifty years without a name for. The aphantasia test gave me the first name. The rupture gave me the time. The storytime event gave me the map. The cascade gave me the explanation for things I had mislabeled, misunderstood, or simply accepted as the unremarkable texture of being alive.

The mind that builds visibility instruments for others has spent most of its life in the dark about its own architecture. This seems, on reflection, exactly right. You build what you need most. You name what you cannot otherwise navigate. The frameworks exist because the person who designed them could not feel where he stood without an external structure to reference.

The DMN was not running. The work was the substitute. Now I know which it was.

Jason I. Oh is the author of The Architecture of Coherence: A Systems Theory of Organizational Justice and the originator of the topological realism framework.