The Mind at the Boundary

A calendar alert returns a promise you had forgotten. The words appear on a screen, but the remembering is larger than the note. Your body changes before the whole conversation comes back. The person you are about to meet gives the promise its weight. The place ahead narrows what you can still do about it.

Where did the thought happen?

The brain participates in every part of this scene. The note, body, other person, and approaching place also carry different parts of the condition that makes the memory available now. The thought becomes consequential through the relationship among them.

This is the relationship I want to explore. Intelligence may be better understood as something sustained across active boundaries: an interior reconstructed from limited signals, corrected through action, and made meaningful by a history shared among organism, artifacts, other people, and environment.

The promise suggests a larger unit than either the alert or the brain: a loop in which memory, body, artifact, another person, and place continually constrain what becomes available to thought.

signal → interpretation → action → environmental consequence → revised understanding

Chris Fields and Michael Levin give this reciprocal loop its clearest conceptual form.² Gravitational holography supplies a controlled physical precedent for a different question about boundary encoding.¹⁴ Orchestrated objective reduction offers a contested answer to the unresolved question of how one conscious event occurs.³ Their evidence is unequal and their mechanisms remain distinct. Together they clarify three operations that a theory of mind may need to address: co-computation, encoding, and actualization.

Three operations at the boundary

Co-computation: how a system and environment build consequential states together

Fields and Levin begin with a system and an environment whose internal states are too large for the other to reconstruct completely. Their shared boundary carries a limited set of actionable signals. Each side acts on the other through those constraints.²

Their 2026 manuscript is a preprint, and its generalized “holographic screen” belongs to quantum-information language rather than gauge/gravity duality. Its immediate contribution is the reciprocal loop.

An organism changes its environment. The changed environment stores, transforms, and returns consequences. A beaver’s dam reorganizes the river that shapes the beaver’s later choices. Writing preserves a trace outside the nervous system. Language carries distinctions that no individual speaker created alone. A model’s apparent intelligence depends on data, tools, interfaces, institutions, users, and the world in which its outputs acquire meaning.

On Fields and Levin’s account, the environment participates in cognition because it changes what the bounded system can remember, predict, and do. Offloading becomes a general strategy for making future encounters more tractable.²

The calendar promise has this structure. The screen preserves a trace, the body assigns urgency, the place provides evidence, and another person supplies a history of expectation. The memory returns through the loop.

Encoding: how a boundary can carry an interior

Gravitational holography offers the strongest physical result and the narrowest domain. In certain highly structured theories, a lower-dimensional quantum system without gravity can be mathematically equivalent to a higher-dimensional description containing gravity. Allan Adams, Paul Chesler, and Hong Liu developed controlled examples in which turbulent fluid behavior at the boundary corresponds to the dynamics of an asymptotically anti-de Sitter black hole in the bulk.¹ Ryu and Takayanagi relate boundary entanglement entropy to the area of a bulk minimal surface.⁴

The useful principle is precise: collective variables available at a boundary can, under special conditions, encode a much richer interior description. The boundary and bulk belong to one formal correspondence. Ordinary cell membranes, interfaces, and brains have no demonstrated gravitational dual.

That limitation still leaves a productive question. Could a cognitive system construct a rich internal world from the much smaller set of variables available at its interfaces? Holography gives us a controlled precedent for asking what a boundary can encode; it does not answer the cognitive question.

Actualization: how one event becomes present

Orchestrated objective reduction addresses another layer. Roger Penrose proposes that quantum superpositions involving sufficiently different mass distributions, and therefore different spacetime geometries, become physically unstable. Stuart Hameroff proposes that microtubules organize quantum processes relevant to neuronal function, with objective reduction producing discrete conscious events.³

The mechanism remains unverified. Hameroff and Penrose review microtubule and anesthesia experiments they consider relevant, while those results leave the proposed chain from neuronal quantum processing to gravity-driven collapse and consciousness undemonstrated.³ Independent experiments have also constrained the simplest parameter-free implementation of the Diósi–Penrose collapse model.⁷

We use Orch OR to isolate an explanatory target that survives the fate of its mechanism: how an integrated field of alternatives becomes this event, present now. A theory of cognitive content can explain why an experience concerns a face, promise, sound, or threat. A theory of occurrence must still explain why there is one experienced event at all.

This distinction gives us a cleaner synthesis. Co-computation concerns how content is constructed and corrected through reciprocal exchange. Encoding concerns how a richer state can be represented through limited variables. Actualization concerns the occurrence of one event. Each job can fail independently.

The reconstructed interior

Experience ordinarily feels like a coherent world. The calendar vibration is already part of a place, a body, a relationship, and a possible action. It does not arrive as a list of separate contributions waiting to be assembled consciously.

The same transformation appears throughout perception. Curved retinal arrays become a stable spatial scene. Pressure variation becomes a speaker with distance and intention. Interoceptive signals become hunger, anxiety, effort, or calm within the same experienced moment. Narrow signals participate in a lived interior whose relations matter more than any one channel.

One productive hypothesis is that this coherence is a high-dimensional generative reconstruction from boundary-limited information. Memory and bodily state shape what the system expects. Signals constrain that expectation. Action asks the environment for better evidence. The resulting interior remains stable enough to guide behavior while the world retains the power to revise it.

In this model, experience is corrected through a loop:

memory and body state → prediction → experienced world → action → environmental response → revised prediction

The holographic comparison preserves one architectural idea: a limited interface can constrain a richer interior state. The mechanisms remain different. No physical boundary encoding or gravitational dual has been demonstrated in brains. The analogy becomes scientifically useful only if a formal model predicts perceptual organization or correction dynamics better than existing generative approaches.

The environmental part of the loop is more than incoming data. We move our eyes, touch the surface, ask another person, repeat the experiment, or return to the place. Action changes the evidence available to us. A world with history answers differently from one imagined in isolation.

This suggests a division of labor. A candidate actualization mechanism might address why an event occurs. Organism–environment history helps explain what the event means. Coordination across body, brain, memory, and world may help explain why its components appear together. The three questions remain connected without becoming interchangeable.

The missing middle

The largest gap sits between microscopic candidate processes and organism-scale experience. Molecular events, membrane voltage, tissue coordination, neural dynamics, bodily regulation, action, and social response occupy different scales. Evidence at one scale cannot silently become a mechanism at the next.

Bioelectric morphogenesis provides a useful intermediate case. In planaria, a brief manipulation of early bioelectric state can alter later anterior–posterior regeneration, including double-headed outcomes after the treatment has been removed.⁸ The experiment establishes that a transient physiological state can have durable anatomical consequences. Fields and Levin interpret work of this kind as evidence that organism-scale pattern information can be carried in distributed bioelectric dynamics and translated into coordinated cellular action.²⁸

That result does not support quantum consciousness. It supports the narrower possibility that collective physiological variables can coordinate local processes across time and scale.

The next reasoning step must therefore be earned experimentally. Can information be tracked from molecular or cytoskeletal dynamics into membrane behavior, tissue-scale bioelectric patterns, brain and body coordination, world-directed action, and reported experience? Each arrow names a different experiment. A complete theory needs the intermediate steps, not a line drawn across them.

This is where the comparison with holographic fluids remains useful. Compact collective variables can describe dynamics whose microscopic detail is far more complicated. Biology may likewise admit useful collective descriptions. The shared structure motivates a search for the right variables while leaving the physical mechanisms separate.

The seams constrain the synthesis

Four boundaries keep the inquiry rigorous.

First, gravitational holography and the generalized information boundary used by Fields and Levin are distinct constructs. One belongs to a stringent gauge/gravity correspondence; the other describes the limited interface through which separable systems exchange actionable information.¹²

Second, environmental loss of separability and objective reduction describe different transitions. The former concerns coupling between system and environment. The latter proposes elimination of alternatives in a quantum state.²³

Third, cognitive offloading explains capability without supplying a sufficient criterion for phenomenality. Memory, prediction, adaptation, and feedback can appear in systems we would hesitate to call conscious.

Fourth, controlled holographic theories are unitary while objective reduction introduces nonunitary dynamics.¹³ A deeper merger would need a consistent account of that tension, perhaps through ordinary open-system physics, an appropriate open-system holography, or a new reduction law with predictions on both sides of the proposed relation.

These seams do not weaken the project. They identify its actual research program. A good synthesis should show where new reasoning is required.

Generation Alpha at the boundary

The physics leads back to a generation already living inside the design question.

Mark McCrindle uses Generation Alpha for people born from 2010 through 2024, followed by Generation Beta from 2025 through 2039.⁹¹⁰ These boundaries are one proposed convention, not a scientific partition of human character. Pew Research Center warns that generational names and cutoffs are neither precise nor universally agreed, and that the labels can easily turn shared historical conditions into stereotypes.¹¹

Used carefully, Generation Alpha names a useful transition. Its childhood spans the shift from screens that mostly deliver content to systems that answer, remember, generate, recommend, and increasingly act. UNICEF now treats generative AI and AI companions as part of the environment children already encounter, alongside questions of privacy, safety, transparency, accountability, development, inclusion, and preparation for an AI-shaped world.¹²

That changes the theory-of-mind problem. Consider a child working with an AI that answers questions, remembers prior exchanges, imitates emotion, makes mistakes, and appears to understand. The useful educational question concerns the boundary around the agent.

What can it observe? What has it retained? Which goals shape its behavior? What came from a source, another person, the model, or a tool? How did the child’s interaction change the result? Where does responsibility remain human?

The interface may implicitly model an answer. Concealed provenance may encourage a child to treat fluent language as knowledge. Invisible memory can make assistance appear historyless. Constant agreement can present intelligence as a mirror. An inspectable system could support a different practice: trace the result, question the memory, compare it with the world, and notice where human judgment changed what happened.

This is boundary literacy. It shifts AI literacy beyond operating a tool toward understanding the relationship around it: capability, hidden state, provenance, dependence, authority, and consequence. UNICEF’s child-centered AI guidance places transparency, explainability, accountability, privacy, development, inclusion, and skills inside that relationship.¹²

Learning also has to remain visible. The OECD’s 2026 review finds that general-purpose generative AI can improve task performance without producing corresponding learning gains, while purposeful educational use can support critical thinking, creativity, collaboration, and sustained learning.¹³ The design question is therefore larger than whether the child reached the right answer. Did the encounter strengthen understanding and judgment that remain available when the system is absent?

What later generations inherit

McCrindle proposes Generation Beta for 2025–2039, followed by Gamma for 2040–2054 and Delta for 2055–2069.¹⁰ These later names are planning conventions, not evidence about the personalities or values of people who have barely begun, or have yet, to exist.

Their value is the time horizon they expose. Generation Alpha is encountering AI as a new participant in learning and social life. Children born into the proposed Beta years may encounter AI memory, agents, synthetic media, and automated decisions as ordinary infrastructure. Those born later may inherit systems whose assumptions were established decades before they could inspect or contest them.

Each cohort inherits more than devices. It inherits defaults about what machines may observe, which memories persist, how sources remain visible, when human approval is required, whose objectives shape an interaction, and whether refusal remains practical. Choices that feel like interface details in one decade can become institutional habits in the next.

This is where cognitive habitats become the design consequence. A cognitive habitat is the complete environment through which people remember, interpret, decide, and act. It can be digital, physical, biological, or social: a family’s shared memory, a classroom, a research environment, a garden, an XR space, or an AI system connected to tools and institutions. The common design question is what relationship the environment makes possible.

Three commitments follow.

Preserve the living sources of thought. Summaries, diagrams, and model outputs should remain connected to evidence, prior interpretations, disagreement, and editable artifacts. Holographic error correction provides a limited design metaphor here: within an appropriate code subspace, a logical bulk observable can sometimes be reconstructed from suitable boundary regions.⁶ A resilient knowledge environment should likewise make important meaning recoverable through several traces rather than one brittle summary. This is ordinary information design informed by a physical precedent.

Make reciprocal effects visible. A system that remembers for us is shaped by, and may retain, our language, attention, corrections, and preferences. The interface should reveal that exchange, especially when the institution operating the system has objectives of its own.

Evaluate the whole loop. Model accuracy remains useful, but the larger test is whether the arrangement improves understanding, resilience, judgment, and agency across time. A better answer can still produce a worse cognitive habitat if people lose the ability to inspect how it was made, revise what persists, or decide what follows.

Generation Alpha makes these commitments immediate. The generations after make them durable. Systems designed now will help teach future people where knowledge comes from, where authority lives, and whether their own judgment still matters.

What the synthesis makes testable

The speculative edge becomes useful when resemblance turns into explicit mapping.

At the theoretical level, information geometry offers a possible bridge. In a specific perturbative holographic setting, boundary quantum Fisher information is identified with bulk canonical energy.⁵ Penrose’s proposal also depends on differences between mass distributions.³ A serious program could ask whether Penrose’s gravitational self-energy can be derived from, related to, or shown incompatible with an information-geometric distance between logical states in a holographic code. A toy model should establish the relation before any biological claim follows.

At the biological level, the missing arrows need causal measurements. Candidate studies would track whether a molecular or nonclassical variable predicts changes in membrane voltage, whether those changes alter tissue-scale coordination, and whether the resulting dynamics predict behavior or conscious-state transitions after classical cellular effects are controlled.

At the systems level, the experiment is close at hand. Compare a model acting alone with the same model embedded in transparent memory, source retrieval, tools, other agents, embodied context, and human review. Measure answer quality alongside source recovery, correction behavior, resilience after failure, decision quality, and the person’s understanding of what the system did.

The hypothesis is practical: intelligence attributed to the model may partly belong to the larger arrangement. Changing the boundary should change the capability. Making the boundary inspectable should change the person’s agency within it.

The promise returns

Holography shows that a boundary can participate in the description of an interior under precise physical conditions. Cognitive offloading describes systems and environments constructing consequential states through limited reciprocal exchange. Orch OR proposes a physical account of how one event becomes actual, while its mechanism remains contested.

Together they reveal a structured question rather than a finished theory. How do limited interactions across many scales become one coherent, meaningful, and actionable experience?

In the framing developed here, the forgotten promise returned through the relation among the calendar, a body that registered urgency, another person whose expectation gave it meaning, a place that supplied context, and an action still available to take. Each element carried part of the arrangement through which remembering became real.

That is also the standard for the environments we build. The practical question reaches beyond whether a model produced the right answer:

Did the complete arrangement help the person remember, understand, judge, and act with greater agency?

The next science of intelligence may depend on how precisely we can answer that question across boundaries. It leaves a practical question open: what kinds of cognitive habitats make people and the larger systems they inhabit more capable while preserving their ability to inspect, revise, or refuse the relationship?