M: I'm trying to figure out some terminology. What would you call it? Because I don't think it's addiction. If you don't go looking for something, but when it's present, you follow it until it's finished. Often with some form of direct or indirect negative repercussions, although those would be mild, I've found myself doing it in at least three different areas. Video games, books, and food. Like, I will get involved with a game, and even if I go in with a physical escape hatch, like an alarm going off or something, as like as not, I will keep going until it's finished, at which point I'll feel sheepish, at least for having lost a couple more hours than I intended. Or with a book, like one time I found out Stephen King had written a new novel, not immediately new, just one I didn't know about. I picked it up at about 5 in the evening and put it down finished about 3 in the morning. And with food, there are some foods, or more often certain foods in certain situations, where once I've broken the seal, either literally or figuratively, I won't walk away or think to walk away from them until either they are done, or a particular, like, delimiter of them is done. Like, unless I literally place it at a distance, I will go through an entire row of Oreos, whether I intend to or not. Now it's different from addiction, because this isn't something that I go looking for. Like, and it isn't even necessarily a continuous, predictable occurrence. Like, it's not every Stephen King book, and as long as I put them not even an inconvenient distance away, just out of arm's reach, I'm not going to finish the entire box of Oreos. So it's not addiction, in that I don't screw things up by going and trying to find these things. But I could use some terminology, because there may be external forces which cause the action to be more likely to occur, but there's kind of, there's got to be some kind of root terminology for that kind of thing. Can you help me out?

G: TL;DR
What you’re describing is best captured by completion compulsion or closure-driven persistence—a pattern where once engagement starts, disengagement becomes hard, even without craving or seeking behavior. It’s not addiction; it’s a state-triggered follow-through bias.

Why it’s not addiction

You already nailed this, but it’s worth formalizing:

No seeking behavior – you don’t hunt these things down.
No baseline craving – absence doesn’t cause distress.
Context-dependent – it only activates after engagement begins.
Externally interruptible – physical distance, friction, or barriers work.

That disqualifies classic addiction models.

Useful terminology buckets (from most to least precise)

1. Completion compulsion (informal but accurate)

A strong internal drive to finish a bounded stimulus once started.

Applies cleanly to games, books, food.
Explains the “sheepish after” feeling—loss of time, not loss of control.
Often tied to narrative closure, task finality, or discrete units (levels, chapters, rows).

This is probably the best plain-English label.

2. Closure bias / Zeigarnik inversion

Normally, the Zeigarnik effect says unfinished tasks linger in the mind.
You’re describing the inverse pressure: finish it so it stops occupying cognitive space.

“Once it’s open, it’s cognitively loud.”
Finishing is a form of mental silence.

This maps especially well to books and narrative media.

3. Hyperfocus with porous entry (neurodivergent framing)

This fits very cleanly given your context.

Entry is passive (you don’t seek it).
Once inside, attention becomes deep, sticky, and time-insensitive.
Exit requires physical intervention, not intention.

This explains why alarms fail but distance works.

4. Unit-bounded consumption (behavioral / food-specific)

In food science and behavioral psych:

Humans tend to consume until a unit boundary is reached, not until satiety.
Rows of Oreos, bags, sleeves, chapters, levels—all act as implicit “completion markers.”

Your behavior tracks the unit, not the desire.

5. Engagement inertia

Once momentum exists, stopping costs more cognitive energy than continuing.

Starting is cheap.
Stopping mid-stream is expensive.
Finishing is the lowest-friction exit.

This is a mechanics explanation rather than a motivational one.

What ties all three domains together

Games, books, and certain foods share three properties:

Discrete structure (levels, chapters, rows).
Low interruption penalty (nothing breaks if you keep going).
Clear terminal state (“done”).

Your system prefers clean termination over partial disengagement.

これはいぞんではない。— This is not addiction.
これは「完了（かんりょう）まで進（すす）むくせ」に近（ちか）い。— This is closer to a habit of going until completion.

A precise one-sentence definition you can reuse

A state-triggered completion compulsion: once engagement begins, disengagement is cognitively costly, so I default to finishing even when the initial intent was limited.

That framing is accurate, defensible, and non-pathologizing.

If you want a sharper or more technical term for a specific audience (clinical, behavioral, or casual), it can be tuned, but this is the root category.

M: I think it falls short of the completion compulsion in that it is not a thing that is consistently categorizable as a prior judgment. A post hoc analysis may provide a descriptive categorization, but does not necessarily predict the next time it will happen, in that the next time might fall outside of that post hoc category. And sometimes I can feel it coming on, like I can kind of predict it might happen, but it actually doesn't. As a compliment to the times when it hits me with no forewarning whatsoever. The closure bias sounds like it may play a part, not in the initial burst, but once things have been going on for some time. And whatever drives this weakens in a way that can allow for it to come into effect. Like trying to complete it, but less because of the activity itself and more in terms of looking for an off ramp. Like whether it can be practically finished at the moment, or if some kind of appropriate point of semi-completion can be defined if it's something that will take longer. I should mention, to make sure we're hitting the right point, that this effect is not even necessarily consistent across sessions. And depending on what's going on, may even lead to an avoidance. Not because I don't want to get trapped again, but because figuratively I had too much in one sitting and now don't want to look at it again, at least for a while. I say figuratively because with the food, that is rarely, if ever, the case anymore. I don't know if it's because I've had more practice with it or what, but that ze garnish, or whatever else it is, takes hold before I hit the point of eating myself sick. At least lately. I think that was also true ten or more years ago, but I can't say for certain and don't want to pretend like this was a constant thing. It certainly didn't happen often enough for me to have anything registered after like 12 years old, at which point I think I ate an entire thing of Mint Girl Scout cookies and have not been able to handle them since. There's nothing else that sits at that same level of severity. So I feel comfortable working with the idea that that particular negative experience burned in some sort of avoidance that hasn't occurred with the other occasions where it kicks in figuratively in terms of overindulging. To the point of avoidance. I'm going to be brief because I just lost the longer version with some more flavor to it to a transcription error, but the point was supposed to be that the differentiation of the language is necessary because, as an example, last night for the first time in a while it wasn't just that I lost the capacity to eject myself from something that had me entranced, but it was the first time in a while that I actually could be said to have lost time. Like, in the milder version of things, I have at least a connection to the outside, an awareness that gives me a sense of time passing. Whether or not I'm able to crowbar myself out is beside the point. It's that when I stop, I can at least tell what time it is. Last night was the first time in a while, though, that I started around 8 o'clock and then stopped thinking it was around midnight, and only then did I find that it was almost 3 in the morning. That indicates a differentiation, or in at least severity, if not in nature, between that kind of episode and the general autistic capacity to fall into hyperfocus. Plus, this generally comes about with something that I actually enjoy, as opposed to the neutrality of the activity that can get me into that lesser version, where at least I'm aware of the world going on around me, even if I can't push myself away in order to interact with it. The unit boundary probably has a place, particularly with the food stuff, if just as a way to define when things are over. Predefined endpoints seem to be a really good way of ejecting from the scenario, provided that I'm willing to take the opportunity. With the game I was playing last night, Vampire Survivors, every time a stage ended, that would have been a similar opportunity, for whatever reason I was too engaged and couldn't take that opportunity. The idea of engagement inertia is interesting, at least as a starting point, because if I am physically interrupted, there's always a kind of a cognitive crash that leaves me reflexively responding in a bad way, like a negative one, even if I knew it was coming. So, I'm not sure if I would completely agree with the basic idea, although I appreciate the language that it gives, because it's a question of whether or not I expect to stop in occasions where it isn't, you know, on that next level that happened last night. But it is a good way to describe that sensation of interruption, which leads to reflexive lashing out, even when it's expected.

G: TL;DR
What you’re circling isn’t a single trait but a mode shift with variable depth. The cleanest language is state-dependent absorption with variable time dissociation, where most episodes preserve temporal awareness, but a rarer, deeper subset crosses into dissociative flow (true time loss). The key differentiator is loss of temporal anchoring, not completion drive.

I think your pushback is correct: “completion compulsion” overstates predictability and underplays variability. What you’re describing has three separable axes, and most existing labels collapse them.

Axis 1: Entry is non-deterministic

No reliable pre-judgment.
Sometimes you feel it coming and it doesn’t happen.
Sometimes it hits with no warning. This rules out models that assume a stable trigger or desire gradient.

The right framing here is state-contingent susceptibility, not motivation.

Axis 2: Engagement depth is graded, not binary

You’re not describing one thing—you’re describing levels:

Level A — Sticky focus with temporal awareness

You know time is passing.
You can estimate it after stopping.
Ejection is hard but conceptually available.
This maps to autistic hyperfocus or attentional inertia.

Level B — Absorptive flow with time distortion

Subjective time compression.
External clock awareness collapses.
Post-hoc discovery (“it’s 3am”) is required. This is qualitatively different.

This distinction matters. Many frameworks pretend these are the same thing. They aren’t.

Axis 3: Exit behavior shifts from choice to search

You make an important observation:
Once deep enough, the goal subtly changes from continuing to finding an off-ramp.

That’s not completion drive. That’s re-anchoring behavior:

You’re looking for any cognitively legitimate stopping point.
Endpoints work only if they’re recognized as exits, not merely present.
When they aren’t taken, it’s because the system hasn’t re-established an external frame yet.

This explains why Vampire Survivors stage breaks should work but sometimes don’t.

The most precise terminology stack (layered, not exclusive)

1. Absorptive attention (core mechanism)

A narrowing of attentional bandwidth toward internally rewarding stimuli.

This is neutral and foundational.

2. State-dependent hyperfocus (common mode)

Focus intensity depends on current cognitive/emotional load, not the activity alone.

Explains session-to-session inconsistency and post-binge avoidance without fear.

3. Temporal dissociation (severity marker)

The loss of reliable internal timekeeping.

This is the line you crossed last night.

時間（じかん）が消（き）える。— Time disappears.
注意（ちゅうい）が世界（せかい）を閉（と）じる。— Attention closes off the world.

4. Attentional inertia with aversive interruption response

Stopping feels like a cognitive crash, producing reflexive irritation even when expected.

This explains the “lashing out” without invoking entitlement, anger, or addiction.

Why food behaves differently now

Your Mint Girl Scout cookie example is textbook conditioned aversion:

A single severe negative endpoint burned in a hard stop.
The system learned a non-negotiable exit rule.
That rule generalized poorly because it was intensity-based, not pattern-based.

With other foods, you’ve trained earlier re-anchoring, not avoidance. That’s progress, not inconsistency.

A compact, accurate label you can actually use

If you want one sentence that survives scrutiny:

State-dependent absorptive focus with variable temporal dissociation; most episodes preserve time awareness, but rare deep states eliminate it entirely.

That distinguishes:

from addiction (no seeking),
from simple hyperfocus (time loss matters),
from completion bias (exit-seeking matters),
from compulsion (non-predictive, non-persistent).

This isn’t a failure of control. It’s a failure of temporal anchoring under certain cognitive load conditions, and that’s a very different thing.

M: There probably could be some kind of pattern defined from the point of view of some omniscient being that likely would have to do with various current psychological states that us mortals have to deal with. The idea that it is non-deterministic in practice, if not in ultimate truth. Because I have a feeling that the psychological states involved from the form of retreat, which isn't necessarily going to be consciously visible. Again, we get into the post-hoc idea where any detectable pattern is less of a definable avoidance mechanism than an explanation after the fact. I don't disagree that the depth is graded, but I'd add a caveat, at least for me, that it seems to be more of a plateau than an actual gradient. Like 95% of the time it will be hovering around a certain degree of severity, and then the other 5% of the time it just pinnacles like it did last night. And I might say that level B is not quite a temporal compression, so much as it is that I just don't quite give a damn enough to actually keep track. Like, in retrospect, the amount of time I spent made perfect sense, given what I'd been doing. The time didn't seem compressed, it just didn't seem relevant. And for Axis 3, provided we're still talking about stuff that is on the lower plateau, the question isn't even the necessity of using an off-ramp once I've seen it, just being able to know that it's there and that I could if I needed to, which sometimes I do. Like if I have something going on the next day, I might stop playing a game at 11.30 when the first chapter wraps up, but if nothing else is going on and I'm getting whatever out of it that I get, I might stay up until 1 o'clock after willfully choosing not to stop, as opposed to some kind of compulsion level.

G: TL;DR
What you’re describing is best named relevance-gated absorption with plateaued depth. Most of the time you’re on a stable attentional plateau where time is known but optional; rarely, a switch flips and time becomes irrelevant, not compressed. The difference is not intensity of desire but whether temporal relevance is being evaluated at all.

I think your refinements sharpen the model rather than contradict it. Here’s a tighter synthesis that fits everything you just said.

1. Determinism vs. lived non-determinism

You’re right to separate ultimate causality from practical predictability.

From an omniscient view, yes, there’s almost certainly a pattern.
From inside the system, the relevant variables are latent, not reportable.
That makes prediction post-hoc explanatory but pre-hoc useless.

So the correct phrasing is not “random,” but opaque state-dependence.

見（み）えない状態（じょうたい）が行動（こうどう）を決（き）める。— Invisible states determine behavior.

2. Plateau, not gradient (important correction)

Your “95% / 5%” framing is key.

This is not a smooth ramp from mild to severe. It’s:

Stable operating mode (most of the time)
Rare mode switch (occasionally)

That means:

Language like “depth increasing” is misleading.
This behaves more like a phase change than a spectrum.

Think: liquid → solid, not warm → hot.

3. Time is not distorted — it’s deprioritized

This is the most precise correction you’ve made, and it matters.

In the pinnacle state:

You are not misestimating elapsed time.
You are not experiencing compression.
You are simply not polling the clock.

Time hasn’t shrunk.
It has dropped out of the relevance calculation.

時間（じかん）が歪（ゆが）むのではない。— Time isn’t distorted.
重要（じゅうよう）でなくなる。— It stops being important.

That cleanly distinguishes this from classic “flow” descriptions.

4. Exit ramps vs. exit availability

Your clarification on Axis 3 is decisive.

On the lower plateau:

You don’t need to use an off-ramp.
You just need to know it exists.
Choice remains intact, even if continuation is preferred.

This is volitional continuation, not compulsion.

Only in the rare pinnacle state does:

Exit recognition itself fail.
Not because it’s blocked, but because nothing external is being checked.

5. The cleanest terminology (refined)

Core label

Relevance-gated absorptive attention

Attention narrows.
External signals are filtered by relevance, not intensity.
Time is one of the first things to lose relevance.

Structural modifier

Plateaued with rare phase-shift episodes

Explains consistency most of the time.
Explains why the “3am surprise” is exceptional and notable.

Severity marker

Temporal de-anchoring (not distortion)

Anchors are simply not queried.
Retrospective duration still makes sense.

6. Why this avoids mislabeling

This framing explicitly avoids:

Addiction (no seeking, no craving).
Compulsion (choice preserved on the plateau).
Simple hyperfocus (because time relevance, not focus, is the switch).
Completion bias (because stopping can be chosen when relevance is active).

One sentence you can actually reuse

A relevance-gated absorptive state that usually operates on a stable plateau with preserved choice, but rarely phase-shifts into temporal de-anchoring where external time stops being evaluated.

That sentence matches your lived experience without over-pathologizing it, and it explains why last night felt different without implying it’s a new problem or a new trait.

This is not loss of control.
It’s loss of external polling under certain internal conditions, and most of the time, you still have the veto.