Stuart Kauffman and the Grain of the Adjacent Possible

Clarification — 7 October 2026

I got a little too excited here. The grain notation makes an existing relativity explicit. Kauffman and Roli already define function through the whole sustained and discuss nested Kantian wholes. The question here is how those relations are tracked across grains when organized continuations conflict. Kauffman and Roli, §6

The ethical comparison concerns the Continuance Principle's claim about pro tanto harm. A competing account can challenge that claim without requiring unconditional preservation of every bearer or reproducing the whole architecture of Modal Path Ethics. Establishing a reduction requires showing how the rival's own normative premises support the relevant harm judgment. Organism-relative success, taken by itself, does not complete that argument.

Longo, Montévil, and Kauffman also distinguish the evolution of biological organization from physical explanations of already-evolved organisms, which they expressly retain. Their unprestatability argument concerns biological observables and niches, not an accidental gap in yesterday's vocabulary. The substantive dispute concerns what follows about entailment across grains. The cross-grain objection raised here requires engagement with that argument in full; it should not be presented as a completed refutation or as an answer to a denial of all physical law. Longo, Montévil, and Kauffman, §1

The questions about bearer, grain, enablement, and the normative bridge remain. These qualifications specify what the comparison has established and what remains open.

Stuart Kauffman's best idea is also extremely easy to give too much authority.

The adjacent possible begins from something almost offensively reasonable.

What can happen next depends upon what exists now.

A molecule can participate in reactions that were unavailable before the molecule existed. A biological structure can acquire a use that was unavailable before the structure existed. A new organism can alter an environment, creating a niche that changes what can survive there later.

The present does not simply occupy one point inside an eternally furnished menu. It helps build the next menu.

Kauffman has been developing versions of this argument for decades. In one formulation, the actual opens an adjacent possible: the set of things available through some next transition. Biological evolution makes the problem stranger because new structures create new uses, niches, interactions, and relevant descriptions. Kauffman's favorite example is the swim bladder: once lungs existed, a structure derived from them could eventually participate in buoyancy regulation. Before organisms with the relevant structure existed, “swim bladder” was not sitting around as an available biological move waiting for somebody to click it.

Modal Path Ethics has recently become considerably friendlier to this argument.

Patañjali and the Grain of Change

Pariṇāma, causal loci, and what actually continues; The revenge of Catherine Malabou

Modal Path EthicsAidan Lawson

Patañjali and the Grain of Change asked what actually undergoes transformation when one material history contains continuation, damage, fusion, branching, and termination at different causal scales. Its answer was to make grain earn its place through historically integrated, nonredundant successor structure.

That now creates an immediate problem for Kauffman.

Adjacent to what?

The Adjacent Possible Needs a Bearer.

Imagine a living system.

Actually, don't.

That phrase has already hidden too much.

Imagine:

• a molecule;
• an autocatalytic reaction network;
• a cell;
• a multicellular organism;
• a symbiotic consortium;
• a lineage;
• a population;
• an ecosystem.
  • All of them can participate in the same material history.
  • They do not therefore possess the same continuation.

A molecular pathway may remain available while a cell dies.

A cell lineage may continue while an organism does not.

An organism may survive while a psychological organization collapses.

A colony may persist while constituent zooids terminate and are replaced.

A lineage may expand through an ecological transition that contracts another lineage.

There is no obvious reason one global object called THE ADJACENT POSSIBLE should flatten these distinctions.

The current Modal Path Ethics locus criterion is deliberately harder than that.

An extant locus is a diachronically integrated causal organization with a nonredundant successor structure at the relevant level of analysis.

That criterion does not begin by asking whether the analyst can name a level.

It asks whether an organization exists there.

• Does it persist through change?
• Do its components and relations participate in maintaining or transforming it?
• Does its current organization constrain what can follow?
• Does analysis at this level reveal successor relations that disappear when the components are considered separately?

Those questions make the adjacent possible indexable.

For a causal grain G at time t, we can speak provisionally of something like

AP(G,t)

the presently available frontier of continuation at that grain.

That is already enough to make Kauffman's picture more precise.

• The adjacent possible of a bacterium is not automatically the adjacent possible of its host.
• The adjacent possible of a host is not automatically the adjacent possible of the ecosystem through which the host continues.

One history.

Several real organizations.

Several futures.

The future has a subscript.

The Whole Has Multiplied.

Kauffman has an answer ready.

He calls living organisms Kantian wholes.

The phrase comes from Kant's treatment of organized beings: the parts exist through their relation to the whole, while the whole exists through the organized activity of its parts.

Kauffman and Andrea Roli's recent origin-of-life work tries to make that architecture materially explicit. Living systems are described as open thermodynamic reaction networks exhibiting catalytic closure, constraint closure, and spatial closure. Collectively autocatalytic sets can form first-order Kantian wholes; such systems can combine into higher-order wholes. Cells then become molecular autonomous agents capable of thermodynamic work, sensing, orientation, decision, and action.

There is a real achievement here.

Reciprocal organization is definitely pretty important.

The cell is not usefully described as a bag containing independent molecules that happen to occupy nearby coordinates.

Its organization performs work.

Destroy the relations while leaving many parts physically present and the characteristic continuation of the cell disappears.

Modal Path Ethics calls this evidence.

Kauffman's framework starts getting into trouble when the organization is asked to tell us which whole owns the analysis.

Because the wholes multiply.

Kauffman and Roli themselves construct higher-order wholes from lower-order autocatalytic organizations. Their vocabulary explicitly includes nested Kantian wholes.

Now put the problem under pressure.

• A cell is a whole.
• An organism is a whole.
• A symbiotic relation may form another whole.
• A multicellular organism contains huge populations of cells whose local continuation can diverge from organism-level continuation.
• A tumor can acquire horrifyingly competent organization.
• A parasite may sustain itself by degrading the continuation-generating structure of its host.
• An ecological association can display integrated dynamics that no individual organism carries alone.

So which whole gets to be the whole?

There is no scandal in answering:

Several.

That is exactly what grain is for.

The mistake comes later, when one organizational criterion quietly becomes the bearer criterion for every question downstream.

A Kantian whole identifies a powerful form of reciprocal organization.

It does not select the one privileged causal grain by itself.

The whole has multiplied.

This is already annoying.

Locushood Is Harder Than Self-Maintenance.

This is where Modal Path Ethics should resist the temptation to steal Kauffman's category and rename everything.

• Self-maintenance is impressive.
• It is not the universal criterion of continuation.

A person can undergo catastrophic transformation while remaining the bearer of a history that no longer reproduces the former personality.

An institution can continue while replacing almost every member. A language can continue through transmission across bodies that eventually disappear.

A legal order can acquire historically integrated successor structure without metabolizing glucose. A relationship can possess characteristic continuations, dependencies, repair routes, and failure modes while lacking a membrane.

A damaged ecosystem can remain causally integrated despite having no centralized process by which the “whole” constructs all of its constituent parts.

Modal Path Ethics therefore asks something more general:

At what grain has a historically integrated causal organization acquired characteristic successor structure that neighboring descriptions fail to preserve?

• Self-maintenance can help answer that.
• Constraint closure can help.
• Catalytic closure can help.
• Reproduction can help.
  • None receives universal jurisdiction.

Persistence itself is sometimes the wrong question.

Patañjali and the Grain of Change developed the Cross-Grain Transition Principle precisely because one material event can constitute continuation at one grain and termination at another. It also introduced the official “No Ontology Photoshop” rule: the analyst cannot zoom until the desired answer appears. The grain has to be earned through actual causal organization before we know what conclusion we want from it.

Kauffman's wholes are therefore welcome to apply for locushood.

They do not receive tenure automatically.

Function Needs a Subscript.

Kauffman makes another elegant move.

Once a Kantian whole exists, the function of a part can be distinguished from its other causal consequences.

The heart pumps blood.

It also makes sounds.

And moves surrounding material.

Plus it produces heat.

Kauffman identifies the functional subset as the causal consequences that sustain the organized whole. He uses this to argue that biological function becomes objective once the whole whose continuation is being sustained exists.

Again:

Useful machine.

Now make that whole-relative dependence explicit across grains.

A causal consequence can sustain:

• the cell;
• the organism;
• the tumor;
• the parasite;
• the colony;
• the genotype;
• the surrounding ecology.

These can conflict.

So let us write the whole-relative index explicitly.

F_G(x)

The function of x for organized continuation at grain G.

This does not make function imaginary.

Quite the opposite. It tells us exactly what relation has to be real.

• If a process helps sustain an integrated successor structure at the organism grain, that is a genuine relation.
• If the same process undermines another locus at another grain, that relation is genuine too.

Nothing requires one to disappear because another is analytically convenient.

That seems fair.

Good for Me.

Then Kauffman gets ambitious.

Living cells, on his account, do more than persist.

They sense.

They categorize their worlds.

They orient toward relevant features.

They evaluate environmental conditions as “good or bad for me” and act accordingly. In Kauffman and Roli's 2025 paper, this becomes part of their account of agency and meaning. Elsewhere in Kauffman's recent work, the same sensing-evaluating-acting triad is described as the root of value.

There is real content here.

• A glucose gradient can be pretty important to a bacterium.
  • A toxin can matter.
  • Temperature too.
  • Membrane integrity is great.

A condition that sustains an organized living system differs, relative to that system, from a condition that destroys it.

Yeah. Fair.

Now:

Which me?

The question becomes unavoidable as soon as Kantian wholes are nested.

What is good for a tumor can be catastrophic for its host.

What is good for a parasite may damage the organism through which that parasite continues.

What preserves one genotype in a fused colonial organism may undermine another lineage inside the same larger organization.

What improves one institution's continuation may destroy the community from which the institution extracts its support.

The phrase good for me has identified a bearer-relative valence.

It has not yet produced morality.

Modal Path Ethics recently spent a great deal of time refusing exactly this shortcut.

Harm as Contraction of Reachable Continuation explicitly stops at the bridge.

Descriptive structure does not produce normativity automatically.

The paper therefore states the Continuance Principle openly:

A nontrivial degradation of a structurally significant part of an extant locus's own organized, reachable continuation constitutes pro tanto harm to that locus.

The paper then has to defend that premise like a champion through explanatory fit, limiting cases, exclusions, burden transfer, destructive continuations, repair, and mixed transitions. It explicitly refuses the idea that a descriptive structure has smuggled morality into the room by naming itself carefully enough.

Kauffman's biological machinery can identify a condition whose loss degrades an organized bearer.

The ethical question is whether that degradation constitutes pro tanto moral harm to the bearer.

That is the claim an attempted reduction has to meet. Comparing this harm with changes borne by other loci is a further question.

“Good for me” tells us that something matters to a bearer. It does not tell us how that bearer matters when another bearer enters the field.

That is where ethics starts earning its salary.

The Tumor Is Having a Wonderful Day.

This sounds like a cheap counterexample.

It isn't, though.

The tumor exposes the exact promotion taking place.

A tumor can possess:

• internal organization;
• proliferative success;
• adaptive response;
• resource acquisition;
• characteristic successor states;
• environmental sensitivity;
• repair responses;
• selection among variants.

At some grain, quite a lot can be good for tumor.

Fine. Modal Path Ethics does not need to deny the organization.

That would be ontology by moral preference.

It says something harder.

The tumor's continuation enters a field containing another locus.

Its expansion can be causally supported by degradation borne elsewhere.

Harm as Contraction calls that kind of relation burden transfer when the continuation of one locus is materially supported by contraction imposed on another.

• Now the tumor's success remains visible.
• So does the host's damage.

The moral analysis begins from both.

A theory that reaches value through self-maintenance has not finished the job until it can explain what happens when successful self-maintenance consumes somebody else's continuation.

The bacterium found glucose.

The field has other paperwork.

Enablement Is Worth Stealing.

Kauffman's strongest contribution to this conversation may be smaller than his metaphysics.

Longo, Montévil, and Kauffman introduce enablement because evolutionary possibility does not always look like ordinary law-entailing causation.

An existing structure, organism, niche, or ecology can create conditions under which a later form becomes viable without determining that later form in advance.

Their wider argument is that biological evolution continually changes its relevant phase space: organisms alter niches, new phenotypes become possible, and new biological observables arise.

This is extremely useful.

Modal Path Ethics already has the nearby concept of continuation-generating structure.

The current harm paper defines it as the causally operative organization through which a locus generates, sustains, accesses, or reopens continuations. That can include memory, ecological relations, infrastructure, reproduction, trust, legal standing, repair capacities, and external supports.

Kauffman's enablement gives us a cleaner distinction inside that idea.

A structure can matter to later continuation even where it does not entail one determinate downstream result.

So let us keep the useful bit.

For causal grain G:

Enable_G(E_t,ϰ)

can indicate that organization present in extant field

E_t

contributes to the actual conditions through which continuation class

ϰ

can become available, without entailing that ϰ will occur.

That lets us separate several things that are too easy to collapse.

• Possibility: the description is coherent under some model.
• Reachability: an actual supported route exists for this bearer under present conditions.
• Enablement: current organization helps constitute the conditions under which a later continuation can become available.
• Realization: the transition actually occurs.

Now destroy the enabling structure.

You do not need to name every future that disappeared.

You need evidence that the organization through which later continuations could emerge has been damaged.

This is why the seed-bank case survives Bergson.

It is also why Kauffman gets to keep his wrench.

Enablement Still Has a Grain.

Of course, we immediately make his new tool more annoying.

An ecological relation can enable one organismal continuation and disable another.

A mutation can enable one lineage-level trajectory while degrading organismal function.

A technological institution can enable actions for a state while making exit less reachable for the persons governed by it.

A psychological adaptation can enable life after catastrophe while closing a former mode of relation.

So enablement is not simply:

X enables Y.

It is closer to:

At grain G, under field condition E_t, organization X contributes to the conditions through which continuation class Y becomes available.

That sentence is uglier.

Reality has shown no concern.

The Space Can Change.

Now we get to the part of Kauffman that Modal Path Ethics should take seriously.

An ordinary transition model tempts us to imagine one state space and one transition relation:

S, T

The system moves.

The graph stays politely still.

Biological evolution often refuses that picture.

A new structure can create a new interaction. A new interaction creates a new niche. A new niche makes some later organization viable.

The relevant distinction at t+1 may therefore be something our description at t did not yet need.

We can represent that modestly.

Let

Γ_G,t

be the currently warranted transition topology for grain G at time t.

Then history may give:

Γ_G,t → Γ_G,t+1

The point is not that reality is literally editing a graph in heaven.

The graph is ours.

The causal organization is not.

When history creates new relations that become causally relevant at a grain, the instrument has to change with the field.

That fits remarkably well with the recent Modal Path Ethics cleanup.

• Bergson already forced possibility space out of foundational ontology.
• Patañjali then forced causal grain into the bearer question.
• Kauffman now supplies another pressure:

Do not assume that the transition vocabulary adequate yesterday exhausts the organization that can emerge tomorrow.

Good rule.

There is no need to turn it into a declaration of independence from physics yet.

Ah.

Which Grain Escaped the Laws?

Longo, Montévil, and Kauffman make a much stronger claim.

Their 2012 paper is titled “No entailing laws, but enablement in the evolution of the biosphere.”

The argument moves roughly like this.

• Biological evolution produces new phenotypes, niches, functions, and ecosystems.
• The relevant biological phase space therefore cannot be fully prestated.
  • Without a prestated phase space, they argue, there can be no ordinary law of motion for evolution analogous to one defined over a fixed physical phase space.

They describe this as radical emergence and argue that the evolution of the biosphere is entailed by no law. They also retain physical explanations of organisms that have already evolved. Their dispute concerns diachronic evolution. Those two claims must be assessed separately.

So this is where grain becomes hostile.

Take their claim of unprestatability at a biological grain GB seriously. The question is whether its implications extend to an entailing account at another grain GP.

That extension needs an explicit connecting argument. Evaluating their account therefore requires examining the proposed connection, including what counts as an adequate biological observable and what an entailing account would have to preserve.

A future biological function is already a grain-relative relation between some organized part, whole, environment, and successor structure.

Their claim concerns the unprestatability of the evolving biological phase space. The cross-grain question is what an entailing account at another grain would have to preserve of the biological relations at issue.

The swim bladder illustrates this distinction.

Physics did not need a variable labeled

FUTURE_BUOYANCY_USE_OF_THIS_LUNG-DERIVED _STRUCTURE

before the relevant organism existed.

Why would it?

“Buoyancy function” becomes a biologically significant description when an organized bearer exists for which that causal consequence participates in continuation.

The category can be new at one grain while the material history remains continuous through another.

This is the cross-grain inference that the argument needs to justify. One event can receive different legitimate descriptions at different grains. No single description inherits universal jurisdiction just because it captures something real.

Kauffman can therefore establish:

Our present biological phase space fails to prestate future biologically relevant variables.

Powerful.

He can argue:

Evolutionary theory requires descriptions whose variables and constraints can themselves change historically.

Also powerful.

Then he reaches:

Therefore no entailing law governs the evolution of the biosphere.

Hold on.

The cross-grain connection is the point that still needs to be assessed.

No Ontology Photoshop.

There are two symmetrical ways to cheat here.

A reductionist can say:

Everything is ultimately physics, therefore organism, function, agency, niche, and ecology are fake.

No.

A causally serious higher grain does not disappear because a lower description contains more microscopic detail.

Modal Path Ethics already learned that from causal emergence and has spent considerable time trying not to become very stupid about reduction.

Kauffman's opposite move can fail too:

Our higher-grain biological categories cannot be prestated, therefore lower-grain lawful entailment has failed.

Also too fast.

• The higher-grain category does not get to disappear downward.
• It also does not get to legislate downward.

That is No Ontology Photoshop in both directions.

• Pick the grain because the organization earns it.
  • Follow the continuation there.
    • Then state exactly what your evidence establishes.

That procedure is so much less exciting than announcing the end of the physics worldview.

Science may still survive.

Radical Emergence Can Wait Outside.

Nothing in this argument proves reductionism.

Important.

Modal Path Ethics has no need to rescue a deterministic universe from Kauffman.

• Perhaps future physics will make strong emergence unavoidable.
• Perhaps biology contains principles that cannot be reduced in some philosophically significant sense.
• Perhaps an adequate theory of causal organization will eventually force a metaphysical revision much larger than anything argued here.

Fine. Bring evidence.

The narrower point is enough:

A claim about unprestatable biological organization needs an explicit account of what it establishes across causal grains.

That is a claim about inference discipline.

And Modal Path Ethics is unusually well positioned to make it because its own ethics has already stopped depending upon metaphysical openness.

A reachable continuation can be determined.

It can recur.

It can belong to a cyclic world.

It can be novel or familiar.

What matters is the actual relation between an extant locus, its historically arrived organization, and what it can still continue into from here.

Kauffman's strongest metaphysical conclusion is therefore unnecessary even where his biological insight is excellent.

That is often the healthiest relationship to a philosopher.

• Take the instrument.
  • Leave the coronation.

What Kauffman Found.

After all that, quite a lot survives.

• Kauffman is right to insist that biological novelty creates a problem for any picture in which evolution simply searches a completed inventory.
• He is right that existing structures can acquire new functions.
• He is right that organisms alter the conditions of later evolution.
• He is right that reciprocal organization matters.
• He is right that something important happens when a causal consequence becomes constitutively involved in sustaining an organized living whole.
• He is right that current organization can enable futures without specifying them.
• And the adjacent possible remains a beautiful image for the frontier created where actual organization meets what can become available next.

Modal Path Ethics can use all of that.

It simply has to ask a few more annoying questions.

Adjacent for which locus?

At which grain?

Which organization made the function real?

Which continuation does the function sustain?

What happens when sustaining one locus contracts another?

When “good for me” conflicts with “good for you,” where exactly did morality come from?

And when a future category could not be prestated biologically, which grain has actually been shown to escape entailment?

Those questions do not erase Kauffman's work.

They put boundaries around what it establishes.

• The whole can be real without becoming the only whole.
• Biological valence can be real without becoming moral law.
• Novel function can be real without having existed beforehand as a completed option.
• An evolving phase space gives us a serious problem of representation and entailment whose implications must be argued at the relevant grain.

There is room here for a better machine.

The Grain of the Adjacent Possible.

Patañjali asked:

What undergoes transformation?

Kauffman asks:

What can become possible next?

Those questions meet before either answer is complete.

Something has to undergo the transition.

Some organized continuation has to make the next state adjacent.

Some causal grain has to carry the relation through which a new function, niche, repair route, or mode of life becomes available.

And sometimes the transition changes that grain's own later vocabulary.

That is exactly the kind of future Modal Path Ethics wants to remain able to describe.

We do not need a completed possibility warehouse, every future biological function hiding invisibly inside the present, one metaphysically privileged whole, or organismic self-interest to manufacture ethics.

And the relation between biological novelty and entailment across grains still needs an argument precise enough to bear the conclusion.

Kauffman found a world that keeps making new doors.

Modal Path Ethics is interested.

It would simply like to know which wall the door is in.

The adjacent possible has a grain.

Follow the continuations.

Sources.

Stuart Kauffman and Andrea Roli, “Is the emergence of life and of agency expected?”, Philosophical Transactions of the Royal Society B 380 (2025).

Giuseppe Longo, Maël Montévil, and Stuart Kauffman, “No entailing laws, but enablement in the evolution of the biosphere” (2012).

Stuart Kauffman, discussion of the adjacent possible, exaptation, and the swim-bladder example.

Stuart A. Kauffman and Andrea Roli, “Beyond the Newtonian Paradigm: A Statistical Mechanics of Emergence,” chapter 8 in Evolution ‘On Purpose’: Teleonomy in Living Systems (MIT Press, 2023), §8.10. (https://cris.unibo.it/retrieve/5b7f8b31-75cc-4797-b22a-6f7fb007250f/c008900_9780262376013.pdf)