Can Two Minds Share One Experience

The agreement behind the stalemate

Michał’s dialogue with the Thomas Epistemes chatbot provides a useful condensation of his position. Computational description characterizes the organization of cognition, but an account of phenomenal consciousness must address the existence of an actual unified event. His formulation is that computation supplies essence while “physics must supply the esse”—the being or actual existence. Later he identifies the phenomenal act with both a physical event and the highest computational organization involved. The distinction is perspectival rather than a mapping between two separately existing things.

I sympathize with the insistence on an occurrence. Our earlier stalemate concerned his demand that computation be legible to nature. My answer was that digital computation is legible when its content is in the causal nexus: the content-bearing distinctions participate in how the system proceeds. Inference, correction, and downstream use must answer to those distinctions. The computational interpretation is constrained by what would actually happen under relevant changes. An arbitrary assignment of labels to a physical history does not meet that requirement.

But this answer belongs on the bottom part of an ontology ladder. It does not make every effective computation a mind.

A logic engine can be highly effective and support an as-if ascription of understanding without literally understanding anything. Its operations can be real and its results useful while the further conditions for mindedness are absent. Our framework looks for an integrated representational economy: structured vehicles, counterfactual richness, accuracy-responsive dynamics, and internal uses through which representations participate in the system’s own continuing organization. These conditions, developed in Minds by Degrees, concern what makes psychological explanation applicable to the system itself.

A minded system, in turn, can literally understand without undergoing phenomenal experience. Our tentative position on current AI occupies this space: some systems may be minded to a degree sufficient for literal understanding while lacking the further organization that constitutes phenomenality. Neither performance nor a vocabulary of self-description closes that gap. The distinctions are explanatory; they need not all be sharp binary thresholds.

So there are at least three different demands on an account: actual computation, minded understanding, and phenomenal occurrence. Saying that the first does not establish the third is correct. Saying it as though a computational account had already claimed to establish all three obscures where the disagreement lies.

For the rest of this essay, I will take the earlier dispute over computational legibility as background. The more productive question is what our commitments about occurrence imply for the boundaries of experience.

From the multiverse to worldliness

In private notes from November 2025, I described a quale as:

an abstract tangent space to a concrete structure (a HOROR in a brain) and yet it is evidently real

“HOROR” meant a higher-order representation of a representation. Elsewhere in the same exchange I stated the commitment more simply:

I think I’m a platonist because I think abstract things are (can be) real.

This was already a view that required grounding. I rejected mental simples and insisted that the relevant representations were “actually participating in being interpreted.” I did not think a freely invented interpretation could turn any arbitrary object into a subject. What changed was the ontological location of the experience: from an abstract structure grounded in a concrete process to the concrete occurrence itself.

The earlier picture made a subject an inhabitant of the multiverse, individuated through its constitutive abstractions. Its identity could extend across worlds in which those abstractions were realized. My present emphasis is on worldliness: the actual processes that sustain this mind and these experiences. Counting further realizations of an abstraction does not by itself tell us how many subjects or experiential occurrences exist here.

Wolfram’s account of quantum mechanics helps me articulate how this worldliness could be explanatorily shielded from the multiverse. Within that framework, the relevance of the larger computational structure to worldly observers is mediated by the effective physics. Worldliness need not mean confinement to one classical branch. It means that the constitution of a subject is answered through the dynamics that actually sustain it. An abstract specification no longer settles its identity across different realizations.

We have a longer essay on the Wolfram project, so I will retain only the distinction needed here. Whether it succeeds as physics, in particular by explaining a previously unexplained phenomenon, is separable from its usefulness as metaphysics. On the latter question I think it stands on somewhat firmer ground. As a metaphysical framework, it makes explicit a possibility important to this discussion: gravity and quantum mechanics can themselves describe emergent regularities. Their status as physics does not automatically make them a science of grounding itself. Wolfram presents the relation between underlying computation and observer-dependent physical regularities in his account of observer theory.

This remains a disagreement with Michał. If physics supplies the being of consciousness, which physics, and in virtue of what? We can investigate physical conditions without granting a discipline grounding authority by definition. A more fundamental description may be indispensable, but calling it physical does not establish its explanatory priority over every organizational description.

Chalmers’s term psychophysical laws invites a related reservation. In his classic proposal, such laws supplement physical theory by relating physical processes to experience. That is a substantive proposal, but its terminology already suggests both the form of the explanation and the vocabulary on one side of it. An identity or an account of constitutive organization might require us to revise that division. This is an objection to treating the framing as compulsory, not a claim that a law could never express a constitutive relation. Chalmers, “Facing Up to the Problem of Consciousness”.

I have moved closer to Michał about actuality without accepting that physics, by its name alone, settles what actuality consists in. This also reverses which boundaries I take an experience or a subject to be able to cross. In my former Platonic account, an experience was constitutively individuated by its relation to a subject’s phenomenal space. An experience belonging to one such space could not also belong to another: even if the same vehicle participated in both, its constitutive relations to two distinct subjects would individuate two experiences. Yet the subject itself, individuated at the level of its constitutive abstractions, could be realized across multiple worlds. Experiences were exclusive to subjects; subjects need not be exclusive to worlds.

The present account locates identity in concrete worldly occurrences while retaining the constitutive importance of regulation. If two subjects’ compatible regulatory involvement sustains one experiential occurrence, their distinctness need not multiply that occurrence into two. What was formerly excluded by the way experiences were individuated becomes a question about realizable dynamics. The reversal is therefore from subjects shared across worlds but experiences exclusive to subjects, toward worldly subjects that might share an experience. This describes the change in my own commitments; neither Platonism nor identity theory, taken as a general label, settles these individuation questions by itself.

What the identity claim commits us to

The early identity theorists provide a useful starting point. Place proposed consciousness–brain identity as a scientific hypothesis, distinguishing it from a claim that mental and physiological descriptions mean the same thing. Smart likewise insisted that the identity was strict: a sensation is the process, rather than an additional item correlated with it. Different descriptions and different routes of knowledge do not alone establish different referents. Place, “Is Consciousness a Brain Process?”; Smart, “Sensations and Brain Processes”.

Our account puts the proposed identity at the level of a concrete dynamical organization. A representational vehicle is the state or process doing the representing; its content is what it represents. The proposal is that certain vehicles, sustained in a particular mode of content-sensitive regulation, are experiential occurrences. Their material may vary, but the constitutive dynamics must actually be present. A description or an input-output imitation of those dynamics does not establish their occurrence.

In Redness Is an Experience, we emphasized the category involved. Redness, in the phenomenal sense, is an experience. We need not posit a further phenomenal property added to an otherwise complete occurrence. Ordinary property language remains available, but it should not silently supply an additional ingredient requiring a second ontological explanation.

Our term acquaintance names the actual regulatory involvement with representational vehicles. It was introduced partly to mark a difference from representation alone: representing something does not ensure that it exists, whereas an actual relation to a vehicle requires that vehicle. But existence of the relata is not enough to prove phenomenality. A thermostat also stands in real causal relations. The stronger proposal concerns a specific kind of cognitive regulation, and identifying it with experience remains a substantive hypothesis.

The regulation is epistemic because it addresses content through its vehicles. It maintains, adjusts, and coordinates the states doing the representing in ways responsive to what they carry and how they function in the cognitive economy. Such regulation can use representations and follow their connections without representing its own regulatory involvement as a further item in experience. Merely supplying blood or keeping a processor cool does not thereby constitute acquaintance. Nor does the proposal require a second internal spectator inspecting a first internal picture.

This permits dreams and imagery without making truthful external perception a prerequisite for experience. The representational vehicle and its regulation can be actual while its represented object is absent or its content false. A dreamed red object need not exist for the experience of red to occur. The framework’s further dynamical commitments are developed in Phenomenal Consciousness as Mode of Being.

It also explains why this is a selective account rather than panpsychism. Experience is identified with particular organizations of activity. Their constituents need not themselves be experiential, and mere causal efficacy does not qualify. Allowing overlapping subjects would not change this: every proposed subject would still have to satisfy the relevant conditions. Feigl already distinguished his identity account, tied to certain configurational aspects of neural processes, from panpsychism’s extension of experience throughout physical reality. We can preserve that selectivity without adopting his restrictive expectations about artificial minds. Feigl, The “Mental” and the “Physical”.

Identity theory alone does not tell us that experiences can be shared. It tells us that the answer must respect the identity conditions of the concrete occurrence. We cannot first individuate every experience by a unique owner and then treat the impossibility of sharing as a discovery. Nor can we assume that sharing a piece of tissue is sufficient to share an experience. The relevant unit includes the activity and its constitutive relations.

A phenomenal field does not draw its own boundary

There is an intuition behind the unified phenomenal field that deserves more attention than a quick rejection of the inner theatre. A visible scene has organization: the cup is beside the plate, both rest on the table, a sound comes from behind me. The relations make the scene seem like one connected whole. Perhaps subject boundaries could be read off this organization.

But several relations are being brought together. The cup’s being beside the plate is represented. Their being experienced together is another matter. The causal coordination that sustains the relevant experiences is a third. Even where these relations track one another closely, they need not be identical or equally manifest in experience.

I have the intuition that my thoughts occur in my head when I reflect on where they occur. I am less confident that this location is part of their experience before the reflection. The reflective act may add a representation of location. Likewise, representing the relation that unifies parts of a scene need not reveal the process that unifies their experiencing.

It is useful to call a phenomenal field a graph of experiences, provided we specify the edges. Some might record represented spatial relations; others, experienced co-presence; others, regulatory dependence. A field so understood is an abstraction from actual occurrences and their relations. It is not an additional substance whose phenomenal properties constitute the nodes. We need not abolish the identification of a subject with an organized field, but the field’s boundary must be grounded in the organization rather than inferred from its being called phenomenal.

Take the relation “experienced beside.” Even in an ordinary scene, adjacency is not transitive: a cup can be beside a plate and the plate beside a bowl without the cup being beside the bowl. Taking a graph’s transitive closure produces a reachability relation. It does not establish that every node is experienced directly together with every other, still less that one subject undergoes all of them. The same warning applies when the edges denote pairwise co-experience rather than spatial adjacency.

Suppose A and B share one occurrence, B and C another, and C and A a third. The combined diagram is connected. Nevertheless, the links might all be maintained within the three distinct regulatory organizations, without any organization sustaining the whole as one perspective. A connected component would then overcount unity. A maximal-clique rule would require its own argument; pairwise relations do not automatically constitute a collective act.

The missing information is what the graph was allowed to leave out. A sufficiently detailed graph of the actual dynamics might represent it, but a graph of phenomenally manifest relations need not. The objection is to deriving subjecthood from the wrong choice of edges, not to using graphs.

One occurrence in two cognitive economies

We can now state the sharing proposal more carefully. Imagine two distinct regulatory organizations, A and B, each of which sustains a minded perspective and some phenomenal activity. Their cognitive economies partly overlap. Within that overlap is a representational process that both regulate in compatible ways.

The important possibility is that their compatible contributions jointly sustain one experiential occurrence. Agreement that an object is red would be insufficient. The regulatory demands must be compatible at the level of the ongoing vehicle: its state, timing, maintenance, and relations to other processing.

The metaphor of a center of epistemic gravity helps if used carefully. It adapts Dennett’s center of narrative gravity: a way of identifying the organization around which a pattern coheres, without installing a little observer inside it. Here the organization concerns epistemic regulation. The center does no work over and above the distributed processes whose organization it names.

A and B could share an occurrence while surrounding it with different associations and uses. One subject could connect it with a memory the other lacks. Their total experiences would then differ, even if this occurrence belonged to both. The proposal concerns partial overlap, not the identity of their complete phenomenal lives.

The difficult objection is immediate. If regulatory relations are constitutive of an experience, surely participation in A’s regulation and participation in B’s regulation make two different experiences. How can the relation be constitutive without individuating an exclusive owner?

The answer cannot be that relations cease to matter at the point where sharing becomes convenient. It must be that a single concrete process can include compatible contributions from both organizations, while participating in two larger economies. Constitutive dependence is not necessarily exclusive dependence. What makes the occurrence experiential might be sustained jointly, without every feature of either surrounding subject becoming a constituent of its local qualitative character.

That last qualification bears much of the argument’s weight. If all of A’s organization enters indivisibly into the identity of every experience A has, the proposal fails. If participation in the two organizations necessarily individuates two occurrences even when their regulatory contributions are compatible, it fails. We need a principled account of which dependencies constitute the occurrence and which connect it to a larger perspective. Merely drawing overlapping circles assumes the result.

The idea has precedents. Philip Goff and Luke Roelofs defend the possibility of phenomenal sharing between distinct but overlapping subjects, while distinguishing that from sharing between wholly discrete subjects. Their discussion also warns against defining experiences by their exclusive subjects and then declaring sharing contradictory. Our interest is in the causal conditions that could realize this possibility under a regulatory identity account. “In Defence of Phenomenal Sharing”.

Metastability and larger subjects

Why expect overlap to be unusual or transient? Because distinct regulatory organizations have different ongoing work to do. Even if their demands initially align, attention can move, expectations can diverge, and the shared representation can be recruited into incompatible uses. Agreement about its current content does not guarantee agreement about its next state.

I call the proposed overlap metastable because sharing requires more than a momentary intersection of regulatory demands. Regulation is temporally thick, and on our account this explains the temporal thickness of the phenomenal now. Shared experience therefore requires compatible regulatory involvement sustained over the relevant interval. A passing coincidence would not establish a shared phenomenal now in the first place.

The distinct centers jointly sustain an occurrence while retaining dynamics that can eventually pull them apart. My expectation that overlap will often end in divergence is a further claim, beyond the sustained coordination required for sharing to occur at all. Distinctness does not logically entail conflict: joint regulation might stabilize the overlap, and a theory should allow that possibility rather than require every shared experience to fall apart. Its duration is something to explain from the dynamics.

Consider the shared red occurrence as A’s and B’s regulatory demands diverge. The vehicle cannot necessarily follow both demands while remaining the same experiential process. Several outcomes are possible, but my intuition favors a change in the qualitative character of A’s corresponding representation, rather than a piece of experience disappearing while a nonphenomenal representation remains. For example, the redness might become more imagery-like for A while retaining its perceptual character for B. A’s experience would continue through a qualitative change as the original sharing ended. Its contribution to ongoing inference and action might remain effective through that change.

Such a qualitative change would have to be realized in altered constitutive dynamics, potentially with a different vehicle filling the same node in A’s scene. Whether the original occurrence persists on B’s side or gives way to two successors would depend on how essential A’s contribution had been. The favored possibility is experiential continuity with qualitative divergence; the account leaves other outcomes open.

There is a further possibility that the metaphor of shared experience can hide. Coupling might establish an encompassing regulatory organization. A and B could become components in the sustenance of one larger subject. We must then ask whether the smaller subjects disappear, persist alongside the larger one, or retain only some of their former organization.

These are distinct claims about organization and experience. Two systems sharing an occurrence need not thereby sustain a larger subject. A larger subject need not automatically extinguish smaller ones. And parts of a subject need not be subjects themselves. Each inference requires evidence about the conditions actually maintained at each scale.

A binary-star image can help us imagine multiple centers of influence and an organization sustained around their combination. It cannot tell us whether the hemispheres correspond to the stars, whether the whole brain corresponds to one star, or whether there are experiential centers at both scales. Those assignments are precisely what the regulatory account must establish. The metaphor supplies possibilities for thinking, not a physical model of consciousness.

This is where our departure from Integrated Information Theory matters. IIT’s exclusion postulate selects a maximal complex among overlapping candidate substrates. Our account does not impose that exclusion in advance. An encompassing regulatory unity could coexist with smaller ones if their constitutive organization survives. Conversely, the coupling could destroy the smaller unities for concrete dynamical reasons. We want the causal account to decide, rather than a rule that only one overlapping candidate may qualify. Albantakis and colleagues, “Integrated Information Theory 4.0”.

Allowing shared occurrences is, in this respect, less metaphysically burdensome than either my former Platonic account of individuation or IIT’s exclusion rule. In the former account, a change of subject entails a different experiential token because the subject enters into the token’s identity. IIT faces a related burden when the maximal complex changes: it must justify why the previously qualifying complex loses its phenomenal status. Neither commitment alone establishes an abrupt change in qualitative character, but both require an account of experiential continuity across changes in subject demarcation.

Our account permits an occurrence to continue while the organizations participating in it regroup, overlap, or form a larger unity. It therefore does not require experiences to flicker in and out of existence, or undergo drastic qualitative shifts, merely because the subject boundary changes. Their continuation and qualitative development answer to the relevant constitutive dynamics. Genuine discontinuities remain possible where those dynamics support them; what we avoid is imposing an additional discontinuity through a rule of exclusive ownership. The earlier example illustrates the alternative: A’s experience can change in character as the shared regulation diverges, with the change explained by that divergence.

Permitting this does not populate every nested subsystem with a subject. A subject is not whatever region we choose to draw a boundary around. The organization has to maintain a perspective and meet the further conditions for experience. Nesting is a possibility whose realization must be earned at each scale.

Conjoined twins and divided brains

Conjoined twins make the distinction between sharing information and sharing an occurrence unusually concrete. Tom Cochrane’s “A Case of Shared Consciousness” discusses Tatiana and Krista Hogan, whose brains are connected by a thalamic bridge. He argues that some of their experiences are most likely shared tokens while they remain distinct subjects. His argument goes beyond their ability to report information available through the other twin’s senses; it concerns where the relevant content is processed. He acknowledges that practical empirical methods cannot directly establish the numerical identity of their experiences. His own overview also links the documentary material on which the discussion draws.

The case should not become a shortcut around the difficult question. A channel between two brains could enable one brain to construct an experience from information originating in the other. Alternatively, the experiential process itself could participate in both cognitive economies. Those alternatives could yield similar reports while differing in the organization producing them.

Our account asks which vehicles are sustained, which processes regulate them, and whether the regulation is jointly constitutive. It would also ask whether demands from one economy change the occurrence available to the other, and whether divergence produces replacement or reconstruction. These are research questions suggested by the hypothesis, not findings we can attribute to the twins. Their case motivates the distinction without settling our proposed mechanism.

Split-brain patients approach the issue from a different direction. Instead of connecting two brains, surgery disrupts a major route of communication within one. Yet the number of subjects cannot simply be read from the number of disconnected anatomical regions.

Pinto and colleagues reported that two patients could not compare stimuli across visual half-fields, while their detection and localization responses were not restricted in the familiar textbook manner to one hand or response channel. The authors interpreted their findings as divided perception without two independent conscious perceivers. Subsequent criticism disputed whether unified response control warrants that conclusion and raised alternative explanations, including cross-cueing. The disagreement is precisely about which aspects of unity the tasks establish. Pinto and colleagues, “Split Brain”; “Unifying Control over the Body”.

For our purposes, neither a simple two-subject story nor an equally simple one-subject story should be built into the example. Access, discrimination, report, and regulation can come apart. An account must identify which dependencies remain, which are lost, and which are reconstructed through other routes. Shared control of a response is evidence about organization, but it is not a direct count of experiential subjects.

Even cooperating hemispheres in an intact brain illustrate a conceptual choice. We might be dealing with one encompassing organization whose parts never qualify separately. We might be dealing with smaller unities integrated into a larger one. Or some conditions could sustain more than one scale at once. The framework permits these alternatives; it does not establish a hidden population of subjects in ordinary brains.

Sharing makes these alternatives less mysterious. Suppose, surprisingly, that an ordinary brain cycled between two subjects and one encompassing subject, perhaps with intervals of coexistence. This need not mean that its component experiences repeatedly vanished and were recreated. Insofar as their constitutive dynamics persisted, the experiential occurrences could stay as they were while participating in different organizations of subjecthood. An encompassing subject could include the experiences sustained within the smaller unities, rather than require replacements for them. The total organization of experience could change, including which experiences were jointly available, without every component undergoing a corresponding qualitative discontinuity. This is a possibility the framework makes intelligible, not a claim that ordinary brains actually cycle in this way.

Artificial systems as controlled possibilities

AI examples allow us to vary the architecture explicitly, provided we do not confuse hypothetical constructions with claims about present systems. Begin with our tentative allowance that a current system may literally understand while lacking phenomenal experience. Connecting two such systems does not make their exchanged information a shared quale. It may establish new cognition or even a larger minded organization while still failing the conditions for phenomenality.

Now stipulate two artificial systems that do have the relevant phenomenal dynamics. There are several different ways to connect them.

They could exchange messages. One receives a report about the other’s experience and constructs a representation of it. The report may be accurate, intimate, and detailed without making the reported occurrence a constituent of the recipient’s experience.

They could maintain synchronized copies of a state. Exact agreement in the represented values would still leave two concrete occurrences. Copying a vehicle is not sharing its numerical identity. How close the resulting experiences would be qualitatively would also depend on their constitutive context.

They could both access one memory location or run on the same hardware. This introduces actual resource sharing, but the shared resource might only provide information for two separately sustained phenomenal processes. A jointly readable buffer need not itself be an experiential vehicle.

Finally, they could participate in the ongoing regulation of one process, with both contributions constitutive of its experiential mode. This is the candidate for sharing. It is much more demanding than common access: the dependence must run through the actual maintenance of the occurrence, rather than terminate at a representation of its output.

Even here the subject count remains open. The arrangement might preserve two distinct regulatory economies with a shared experiential part. It might establish a third, encompassing economy. It might transform the original systems so thoroughly that their former subjecthood no longer persists. The architecture would have to discriminate these cases by how it sustains regulation, self-location, and the availability of the shared occurrence within each economy.

These variations clarify why the ontology ladder matters. We can increase cooperation, information exchange, or computational efficacy while remaining below phenomenality. We can also alter phenomenal organization without simply increasing any of those quantities. What matters is which concrete relations the alteration establishes.

What would count against the proposal

The account needs more than permissive language. Its strongest vulnerability is the individuation of a local occurrence within a relationally constituted whole. If we cannot identify a shared experiential process without arbitrarily subtracting essential aspects of A or B, then we have not described one experience in two subjects. We have described overlapping physical support for two experiences.

Likewise, the claim that distinct regulatory centers tend to destabilize sharing must be earned. If their contributions stabilize a common process indefinitely, the expectation of eventual divergence would be wrong. If every arrangement capable of sustaining the common process necessarily forms one exclusive subject, then partial sharing between continuing subjects may be impossible on this account. Dropping an exclusion axiom does not guarantee that the dynamics permit overlap.

Conceivability cannot decide these questions. We can assign the same vehicle one content in A and a different content in B on a diagram. But the assignments may make incompatible demands on its actual transitions. The vehicle has one concrete history. If one process requires it to remain in a state while the other requires it to leave that state, writing both requirements down has not supplied a realization. Compatible joint regulation needs a realizable history too; agreement between descriptions is insufficient.

This is where my identity commitment limits the thought experiments. Some apparently conceivable assignments may not be metaphysically possible once the constitutive dynamics are specified. Conversely, the intuition that an experience must have exactly one owner cannot by itself rule out an arrangement whose actual organization sustains overlap. Both permissive diagrams and prohibitive intuitions need to answer to the proposed occurrence.

Nor does the account yet explain away every version of the hard problem. Identifying experience with a mode of regulation is a hypothesis about what experience is. It does not derive feeling from the mere fact that something is regulated. Its value must come from how specifically it connects phenomenology with organization, where it draws distinctions, and whether those distinctions withstand investigation.

The question to carry forward

Michał’s insistence on actuality has become easier for me to appreciate as I have moved away from individuating subjects through abstractions across the multiverse. The remaining task is explanatory: to identify which actual process an experience is and how its constitutive dynamics determine its unity and boundaries. An appeal to physics assigns the explanatory task to a discipline; it does not by itself perform it.

That work also bears on what counts as one subject. A subject need not be an additional phenomenal object sitting behind its experiences. It can be the organized process through which they participate in a perspective. The unity of that process is real if its constitutive dependencies are real; it need not be visible as another item in the experienced scene.

If one occurrence can participate in two such organizations, distinct subjects can overlap. If their coupling sustains an encompassing organization, a larger subject becomes a further possibility. Whether it replaces or coexists with the smaller ones is then an empirical and metaphysical question about what continues to be sustained.

The imagined shared redness leaves us with a precise task. Establish that there is one occurrence. Establish that two perspectives remain. Then show how their regulatory involvement can constitute the first without abolishing the second. That would make mind melding intelligible without defining the participants out of existence at the moment they meet.