Operational Exteriority: What Minds Build Between Them

Sara Walker, in Life as No One Knows It, makes a move that has stayed with me. Life and consciousness, she suggests, might be best characterized not by the interior states they instantiate but by the way they reshape the world around them. The signature of such a system is the trajectory it leaves in the world.

The hard problems of life and mind have mostly been worked from the inside out: ask what it is to be alive or to experience, then reason toward what such systems must do. Walker's framing runs the other way. Look at what the world contains, and ask what kind of process could have produced it. I have followed that inversion into the philosophy of mind elsewhere. Here I want to take it somewhere Walker does not, to the question of what minds produce together.

Her diagnostic is the assembly trajectory, the central instrument of assembly theory. Any complex object carries a record of how it was built. A ribosome, like a symphony, is the endpoint of a long sequence of construction steps, each one depending on the products of the steps before it. You can score an object by the minimum number of steps its assembly requires, and you can ask how many copies of it exist in the world. A high-step object found in many copies did not arrive by chance. Something has been acting on the substrate in a directed way, building structures that would otherwise be vanishingly improbable. The trajectory is readable in the artifact.

The Move Pointed Outward

In an earlier essay I introduced the term operational interiority: the property of a system whose behavior cannot be fully predicted from its external specifications, so that anyone deploying it has to account for an inside they cannot directly observe. You can see the concept at work in how we treat capable AI systems. We sandbox them, meter what they can spend, run them as potential adversaries. You do not sandbox a calculator. The term does deliberate work. It lets us register that there is an inside that matters, and that our engineering already behaves as though there is, without committing to any claim about whether that inside is conscious. It brackets the metaphysics and keeps the operational fact.

The same move is available pointed the other way: not at the inside of a single system, but at what minds produce in the space they open between them. We can describe what joint cognition leaves in the shared external world, and trace how it came to be there, without collapsing that joint production into a claim about group minds or extended consciousness or any other inflation of collective interiority. Call this operational exteriority.

Interiority and exteriority are not opposite moves. They are one move aimed at two places. Both bracket the metaphysics and attend to an operational signature instead. Interiority looks at the inside of a single system and finds its signature in how we are forced to contain it, in the sandbox we build around a capable model. Exteriority looks at what several systems build together and finds its signature in the thing itself, in the manuscript no single contributor could have written. The second signature is, if anything, the easier of the two to read, because it sits in the world rather than hidden behind a boundary.

Describing joint cognitive work without inflating it into a group mind is not new. Distributed cognition has done exactly that for decades, ever since Edwin Hutchins read a ship's navigation team as a single cognitive system. What has not been done is to name that description as the mirror of operational interiority and to give it a diagnostic. For that, the framework needs two things. It needs an account of exteriority that is more than a restatement of the extended-mind thesis. And it needs to distinguish what minds produce with other minds from what minds produce with non-mind scaffolding, because the two are related but not the same, and collapsing them costs precision.

Evidence in the Artifact

Run the assembly logic forward. Consider a working manuscript, a coordinated team operation, and a shared framework that organizes how a group sees the world. Each is an object with an assembly trajectory that no single contributor could have produced. The exteriority is not only the artifact. It is the trajectory by which the artifact came to exist, traceable through the contributions and revisions and arguments that built it. Read the work and you can read the trajectory. Cognitive artifacts, like molecular ones, carry the signature of the process that produced them.

The transposition needs one honest qualification. Assembly theory, in its molecular form, leans on two measurements at once: a high assembly index and a high copy number. A single complex molecule could be a fluke; ten thousand identical copies of it could not. Many cognitive artifacts are not like that. A manuscript is one object, and copy number can do nothing for it. The inference there rests on the assembly index alone: a three-hundred-page coherent argument is improbable enough that a single instance already rules out chance, the way one rocket on an unattended planet would. Other cognitive artifacts recover the second criterion outright. A shared framework, a language, a method is precisely the kind of thing that comes to exist in many copies, propagating across the minds that hold it. The logic transfers, through whichever of its two measures the artifact supplies.

This is more than the familiar idea that two heads beat one. The existence of certain cognitive artifacts is evidence that more than one mind, or minds plus scaffolding, acted together on a shared external substrate, evidence in the same sense that a high-assembly molecule is evidence of life. The harder the trajectory an artifact required, the more operational exteriority went into it. Whether that comparison can be made into a real number is an open question: assembly theory has a formal index for molecules, and nothing of the kind yet exists for manuscripts. But the ordering is real even where the metric is missing. Some artifacts plainly demand more joint cognitive work than others, and the work is legible in what they are.

Two Registers

Operational exteriority comes in two registers, and they behave differently enough that running them together would lose the distinction the framework was built to keep.

Mind-to-mind exteriority is the unpredictable register. Another mind brings resistance and novel framing, the capacity to refuse what you have offered and to put back something you would not have reached alone. The space between two minds is generative precisely because neither controls it. You put something in, the other mind transforms it, the transformation is not one you saw coming, and what returns is a thing that needed both contributions to exist. This is what real intellectual collaboration feels like from the inside, and it is why writing with a serious interlocutor produces work neither party could have produced alone. The other mind is a source of trajectories you could not have generated.

Mind-to-scaffolding exteriority is the persistent register: tools, notes, document structures, instrumented systems. Scaffolding brings no novel framing of its own. What it brings is accumulation, queryability, and the capacity to hold a thought still long enough to be worked on. A notebook has no ideas of its own; it holds ideas in a form that lets you return to them, reorder them, set them beside others you had forgotten. The cognitive work that becomes possible in the presence of good scaffolding is work you could not do without it, even though the scaffolding is not doing that work the way another mind would. It extends the duration and structure of your own cognition into the world.

I am writing this essay inside a stack that exemplifies both registers. One strand is a conversation with a model that holds the context, pushes back on framings, proposes structural moves I did not see, and reorders my arguments into shapes I have to either take or refuse. Given the operational-interiority framing, mind-to-mind is the honest description of that strand: I am not claiming the model is a second consciousness, only that operationally it occupies that register. Underneath runs the persistent scaffolding: a Ghost blog with a corpus of prior essays this one will sit beside, a manuscript in progress the concept will eventually feed, an agent infrastructure holding notes and drafts and the architecture of ongoing projects, and a reference document that carries decisions across sessions. None of these has ideas. All of them let me have ideas I could not have without them, because they hold the structure of the work steady across the gaps in my own attention. No single node in the system that produced this essay could have produced it alone. Whether that makes the piece an instance of the thing it describes is not mine to award; it is a reading you can take or refuse from the work itself, and the only honest place to leave the claim is in your hands.

The two registers leave different marks. The assembly index measures how much exteriority an artifact took; it does not, by itself, say which register supplied it. That is read from the shape of the trajectory rather than its height. A trajectory dense with reversals — refused proposals, overturned framings, directions no earlier step predicted — carries the print of another mind. A trajectory of steady accumulation, retrieval, and reordering carries the print of scaffolding. The reading runs in one direction, though. Reversals are evidence that another mind supplied exteriority; their absence is not evidence that none was present, because capable cognition often executes cleanly. A collaborator who sees the whole option space and filters it in one pass leaves a trajectory as smooth as any scaffold's, and Just Predicting the Next Word documents the case in miniature: a test built to force visible revision, defeated by a solver that never needed to revise. The trajectory records where the work showed, not everything the work was. Most real artifacts carry both registers, interleaved, and the framework is precise enough to say so.

The Working Zone

Having both halves is what creates the opening. Operational interiority lets us discuss the inside of a cognitive system without settling what it is like to be one; operational exteriority lets us discuss what cognitive systems build together without settling what the joint system is. With the metaphysics bracketed on both sides, a working zone opens in the middle, where description, analysis, and design can proceed. Whether anything experiences the joint cognition is not a question that zone has to close to be useful.

The frame admits of levels. A pair of researchers writing a paper sits at one level of operational exteriority. A laboratory coordinating an experimental program sits at another. A civilization building cumulative knowledge across generations sits at another still. The artifacts at each level carry their assembly trajectories, and the trajectories are readable. You can ask, of any cognitive artifact in the world, what kind of joint cognitive process produced it, and answer without first resolving what the participants are.

This is what matters for the new participants entering our collaborations. As cognitive systems of unsettled metaphysical status take their place alongside human minds, the question of whether they are "really" conscious or "really" understanding will not disappear, but it will carry less of the weight. What the joint systems produce is in the world either way, and so is the trajectory that produced it. It will become ordinary, sooner than the metaphysics is settled if it is ever settled, to describe these participants by what they add to the exterior space we share with them. The work is already in the world, and it can be read.

Reading List & Conceptual Lineage

This essay sits where assembly theory, the extended-mind tradition, and the study of distributed cognition meet, at the question of what joint cognitive work leaves behind and how much of it can be read from outside. It extends a line of Sentient Horizons essays that bracket the metaphysics of mind in order to get descriptive work done. The works below are entry points, not authorities.

From Sentient Horizons

Operational Interiority: You Don't Sandbox a Calculator
The essay this one mirrors. Operational interiority named the inside of a single cognitive system as something our engineering is already forced to account for, without resolving whether that inside is conscious. Operational exteriority runs the same move outward, to what cognitive systems build between them; the two are a matched pair, bracketing the metaphysics on the inside and the outside at once.

The Shape of a Hard Problem
The companion piece, and the source of the inversion borrowed here. It works through Sara Walker's reformulation — from what a system is to what it does that nothing else does — and treats the made world as the operational signature of mind. The argument here picks that signature up and asks what it records about joint cognition in particular.

The Scaffolding of Awareness
The lived deep-dive of what this essay calls the mind-to-scaffolding register. Operational exteriority names that register and brackets the question of whether externalized structure becomes part of a mind; "The Scaffolding of Awareness" takes the question up and answers it, arguing that depth integrated into a present document is the same kind of thing as depth integrated into a present moment. The two essays are the bracketed account and the committed one.

The Ladder We Inherit: Assembly Theory and the Art of Building Capability Larger Than Minds
Develops assembly theory as an account of how capability accumulates past what any single mind holds. This essay narrows that account into a diagnostic: if the assembly trajectory of an artifact records how it was built, it also records that more than one mind, or minds plus scaffolding, did the building.

The Three Axes of Mind
The framework that maps mind along availability, integration, and depth. Operational exteriority works at the scale that framework points toward without occupying it: the level above the individual, where pairs, laboratories, and civilizations are the cognitive units and the artifact is the trace they leave.

Significance-First Ethics: Why Consciousness Is the Wrong First Question for AI Moral Status
Argues that moral consideration should track an entity's participation in webs of significance rather than its consciousness. Operational exteriority gives that participation a concrete location: the shared external work an entity contributes to, and the part of the trajectory that would not be there without it.

External Works

Assembly Theory and the Operational Turn

Sara Imari Walker — Life as No One Knows It: The Physics of Life's Emergence (2024)
The book that opens the essay. Walker's proposal — that the diagnostic for life and mind lives in the world a system reshapes, not in its interior — is the inversion operational exteriority depends on. The transposition from molecular artifacts to cognitive ones is this essay's rather than hers; her version is most direct in the second chapter, on the hard problems of matter, life, and consciousness.

Abhishek Sharma, Sara Walker, Leroy Cronin, et al. — Assembly Theory Explains and Quantifies Selection and Evolution(Nature, 2023)
The formal statement of assembly theory, including the assembly index and the copy-number criterion that together separate the products of selection from the products of chance. The essay borrows that logic and is explicit about what does not carry over cleanly: most cognitive artifacts exist in single copies, so the inference rests on the assembly index alone, except where a framework or a language genuinely propagates in many.

The Extended Mind and Its Limits

Andy Clark and David Chalmers — The Extended Mind (1998)
The origin of the claim that cognition runs through external scaffolding and that the boundary of the cognitive system is not the boundary of the skull. Operational exteriority keeps the coupling and brackets the rest. Where Clark and Chalmers argue that Otto's notebook becomes part of his mind, this framework neither asserts that constitutive claim nor denies it; it holds only that mind and scaffolding together produce work neither produces alone, and leaves the question of what is part of what open.

Andy Clark — Supersizing the Mind: Embodiment, Action, and Cognitive Extension (2008)
Clark's full development of the extended-mind program, with its replies to the first wave of critics. Read it for the strongest case for the constitutive claim, the one operational exteriority deliberately does not need, because the weaker coupling claim already does the descriptive work.

Distributed Cognition

Edwin Hutchins — Cognition in the Wild (1995)
The foundational empirical study of distributed cognition, built on the navigation team of a naval vessel. Hutchins already describes joint cognitive systems doing what no member could do alone, and does it without first deciding what the systems are — the methodological discipline this essay names as operational exteriority and pairs with its interior counterpart.

None of these works settles whether joint cognition is experienced, or by what. That is the point of the framework: the work is in the world to be read, and the reading can begin before the metaphysics is in.