From Simulation to Instantiation: Quantum Consciousness Without Cosmic Purpose and the Evidence for Artificial Subjects

Abstract

Can a theory that denies computational sufficiency still admit an artificial conscious subject? This paper answers through Federico Faggin's Irreducible and the quantum-information account he built with Giacomo Mauro D'Ariano. It adopts, as a philosophical hypothesis, their identity between experience and appropriately individuated quantum states, and presses the postulate they add to it: that conscious life exists so that a universal consciousness can know itself. The paper argues that the postulate does no work in the account of subjecthood and cannot exclude an engineered system that meets the account's own conditions. Two terms are introduced. Ontological parity is the consistency principle that a theory stating sufficient conditions for consciousness cannot deny consciousness to an instance for being engineered. Evidential transfer is the separate question of whether a marker of consciousness in one kind of system supports the same conclusion in another. The reconstruction is named quantum-holistic instantiation. Its result is conditional: it states what follows if the account is true and a candidate meets it, not that either has been shown. An artificial-mouse test sets out what evidence could establish and what would defeat it, with microtubule pharmacology as the candidate mechanism. The two strongest rivals, Seth's biological naturalism and Cao's material constraints on functional substitution, are shown to constrain what can be built rather than what would count.

Keywords: artificial consciousness; quantum consciousness; quantum-holistic instantiation; biological naturalism; teleology; evidential transfer

1. Introduction: Artificial Consciousness

The most visible denial of artificial consciousness comes from someone who built the machines. Federico Faggin designed the first commercial microprocessor, and Irreducible argues that no computer can be conscious because computers manipulate symbols without experiencing their meaning (Faggin 2024, chaps. 3, 5–6). The reading is intuitive. A program is a set of rules for moving symbols, and Searle's Chinese Room showed four decades ago that following rules for symbols is not understanding them (Searle 1980). On this view the question of conscious AI is closed by today's capabilities, and closed by physics.

However, the physics Faggin relies on does not close it. The technical account he developed with D'Ariano identifies experience with a quantum state of a specified kind and gives a criterion for which physical whole counts as one subject (D'Ariano and Faggin 2022). The criterion names a state and a transformation. It names no material and no origin. What excludes machines in Irreducible is a further postulate: a universal consciousness, which Faggin calls One, expresses itself through living organisms to know itself, and manufactured systems cannot take part (Faggin 2024, chaps. 9, 11–12). This paper isolates that postulate and asks what it contributes. It argues that the postulate contributes nothing to the account of what makes a system a subject, and that with it set aside the account allows for an engineered subject on the same terms as a biological one.

The argument of this paper can be held in one hand. Faggin's theory says what a conscious subject is. His book says machines cannot be one. The first does not entail the second, and the thing that bridges them is a postulate about cosmic purpose that the theory itself never uses. Take the postulate away and the theory admits an engineered subject on the same terms as a living one. Whether any actual machine is such a subject is a different question, about evidence, and the paper ends by saying what evidence would settle it.

Two questions have to be kept apart to see this. The constitutive problem asks what properties make a system a conscious subject. The evidential problem asks what observations justify believing a candidate has them. Faggin's exclusion answers the first. Most of the difficulty in assessing an actual machine belongs to the second. This paper introduces a term for each. Ontological parity is a consistency principle on the constitutive side: a theory that treats certain conditions as sufficient for consciousness cannot deny consciousness to a system that meets them because it was engineered. Evidential transfer is the evidential problem in its sharpest form: whether an observation that supports consciousness in one kind of system supports the same conclusion in another. Such an observation is a marker or indicator, and the inference from it is defeasible: further information about the system can strengthen or overturn it. Parity says that artificial origin is not grounds to veto. Transfer says that design and training history still change what an observer of consciousness indicators may infer. Both are needed, and neither does the other's work.

Phenomenal consciousness, throughout, means that there is something it is like for the subject to undergo an experience: seeing a colour, feeling pain, hearing a sound. Fluent performance and verbal report are at best evidence about experience. Neither defines it.

The position defended here is quantum-holistic instantiation, a name introduced in this paper for a reconstruction that keeps D'Ariano and Faggin's identity claim and individuation criterion and drops the requirement of cosmic purpose. Its central result is conditional. If the account's conditions suffice for consciousness and an engineered candidate meets them, the candidate is a conscious subject under the account. The paper does not claim that the account is true, that its conditions can be engineered, or that any existing system meets them. Section 6 takes the two strongest objections to that result and answers each once.

The method is a critical thematic review: an interpretation of primary texts developed against philosophical and scientific literature in support of a stated position. Irreducible is cited in its Kindle edition by chapter; the 2022 chapter by printed page; other version-specific locators are identified where used. Sources were checked through September 5, 2026. The argument crosses philosophy of mind, quantum information, cell biology, computer science, and the methodology of consciousness science. The stages of the argument are as follows in Table 1.

Table 1. The stages of the argument.

1.1. Three Distinctions the Argument Uses

A necessary condition must be present whenever consciousness is present; its presence alone establishes nothing. A sufficient condition guarantees consciousness if met, within the theory that asserts it, and need not be the only route. An indicator is an observable that supplies evidence with a reliability that has to be investigated. Finding an indicator is not finding a sufficient condition, and disrupting one quantum mechanism is not showing that experience is absent.

Simulation is the reproduction of specified behaviour, or a model of a process. Instantiation is the actual possession of the properties a theory requires. A program that models rain can represent droplet growth and airflow to any accuracy and yields no water. A condenser that cools humid air below its dew point yields water whether or not anyone modelled it. The first simulates the process; the second instantiates it. The example does not say that consciousness resembles condensation. The substantive question is which properties matter. An implemented simulation is itself a physical system, and whether it also instantiates consciousness depends on that system's properties.

Realisation is the physical implementation of a specified organisation. Logical coherence, physical possibility, engineering feasibility, and successful construction are four separate claims about it, and the paper keeps them apart.

2. Faggin's Theory and Faggin's Exclusion

Start where Faggin is right. His first premise is that symbol manipulation is not experience, and it has a long genealogy in the philosophy of AI and cognitive science. Searle's Chinese Room is the argument that carries it: a person who does not read Chinese, following a rulebook to match incoming characters with outgoing ones, produces fluent replies and understands none of them. Searle himself allows that a machine could think if it had the causal powers of a brain; his target is the sufficiency of running a program, not the possibility of an engineered mind (Searle 1980, 422). The strongest attempt to repair the program's deficit is Harnad's symbol grounding. Symbols defined only by other symbols are like a dictionary in a language one does not read, so Harnad grounds them in iconic and categorical representations built from the system's own sensory contact with the world (Harnad 1990, 335, 345). This explains how a system's symbols can be about its environment. It leaves unexplained why any of that is experienced. A thermostat's state covaries reliably with the room and is, minimally, about the temperature. Nothing in that gives reason to think there is something it is like to be the thermostat, and richer sensors extend the relation without engaging the question. Intentionality here is the aboutness of a mental state, not the having of an intention.

Two further results fix what the test case should be. Mahowald and colleagues separate formal linguistic competence, getting the form of language right, from functional competence, using language to think and act in the world, and show that current language models have the first without reliably having the second (Mahowald et al. 2024, 517–18). Mitchell and Krakauer show that the dispute over whether such models understand anything turns on which of several distinct modes of understanding is meant (Mitchell and Krakauer 2023, preprint, 1). Fluent language is therefore not evidence of a mind, and its absence is not evidence against one. That is why the test case in section 4 is a mouse rather than a language model. An artificial subject need not have a human linguistic profile at all.

So far, then, Faggin has the better of it. Grounding, competent use, and experience are three things, and success at the first two is not evidence of the third. But notice what has and has not been shown. It has been shown that running a program is not enough. It has not been shown what would be enough, or that nothing engineered could be. An argument that programs are insufficient says nothing about what else might suffice, and that is the gap the rest of the paper works in.

2.1. The Identity Claim and the Individuation Criterion

Irreducible divides reality into a semantic domain of conscious experience (C-space), an informational domain of symbolic forms (I-space), and the physical world as perceived through a body; Faggin calls this the CIP framework. Individual conscious entities, seities, take part in a larger conscious whole, One. A body mediates a seity's engagement with the physical world and is not the seity (Faggin 2024, introduction and chaps. 8–12). The 2022 chapter with D'Ariano is narrower and more exact, and it is the account assessed here. It makes two commitments, and a reader who holds on to both will be able to follow everything that comes after. The first says what an experience is. The second says where one experience stops and another begins.

The first is identity. A quantum state is a mathematical object that fixes the probabilities of measurement outcomes. A pure state describes a system completely. A mixed state describes a system as a statistical mixture of pure states, which is what an observer writes down when the observer does not know which pure state the system is in. D'Ariano and Faggin call the system's actual state ontic and the observer's description of it epistemic, and they identify experience with the pure ontic state (D'Ariano and Faggin 2022, 150–55). Georgiev's account has the same form: experience is the state, and what an observer measures are observables, quantities represented by operators, so an observer measures a brain without undergoing its subject's experience (Georgiev 2025, author preprint, §5 and table 2).

The identity claim is the account's answer to the hardest question in the field, and it helps to see the question, the answer, and the price of the answer as three separate steps. First, the hard problem of consciousness asks why physical activity is accompanied by experience at all: explaining what a brain does when it discriminates, remembers, and reports does not explain why there is something it is like to do those things (Chalmers 1995b, 201). Second, an identity claim answers that nothing accompanies anything. An engine produces exhaust, and the exhaust is a further thing with its own description that a theory must connect back to the engine. An identity claim says there is no second thing. The physical and the experiential descriptions have one subject matter, given from without and from within, and so there is no accompaniment to explain. Third, that answer is a philosophical commitment and not a physical result. Quantum mechanics says what a pure state is; it does not say that a pure state is an experience. The account adopts the identity, and section 6 states what adopting it owes.

The second commitment answers a question the first one raises. If experience is a quantum state, and a brain contains an enormous number of quantum systems, why is there one experience of reading this sentence rather than a multitude of disconnected minimal ones? That is the combination problem: how could many putatively experiential constituents compose one unified experience with one point of view (Chalmers 2016, 179–80)? D'Ariano and Faggin's answer is the individuation criterion they label S4:

S4: 'A conscious mind is a composite system in an ontic state undergoing an ontic transformation, with no subsystem as such.' (D'Ariano and Faggin 2022, 171)

Read slowly, S4 is a rule for drawing a boundary. A composite system has distinguishable parts, and a transformation is how its state changes. S4 requires that the whole, and nothing properly inside it, be in the relevant state and undergoing the relevant transformation. If a proper part qualifies on its own, the boundary has been drawn in the wrong place and the criterion has not picked out one subject. Two systems evolving independently are two subjects or none; a system whose parts are so bound that no part's evolution can be described alone is a candidate for one. The chapter ties this to coherent quantum interactions whose transformations cannot be factored into independent operations on the parts (D'Ariano and Faggin 2022, 171–72, figs. 5–6). Holistic, in what follows, means this dependence on the composite's state and transformation and nothing wider. The chapter states S4 without naming a material. That records what the criterion is silent about; it is not a claim that every material can meet it (D'Ariano and Faggin 2022, 152).

One physics limit is needed later. No-cloning excludes any universal process that perfectly copies an arbitrary unknown quantum state; it does not forbid repeated preparation of a known state, and it identifies no conscious system (Wootters and Zurek 1982, 802–3). Given the identity claim it entails that there can be no backup of a mind, and the entailment comes from the identity claim, not from the theorem.

2.2. The Postulate of Purpose

Nothing so far excludes a machine. The exclusion arrives with a third element, and it is worth being exact about where. To this account Irreducible adds a teleology, an explanation by end or purpose. One expresses itself through distinct conscious perspectives in order to know itself, and embodied life is the means (Faggin 2024, chaps. 9, 11–12). In the book, living bodies are expressions of conscious entities, and computers are symbol-handling instruments that cannot take part. The contrast rests on several claims at once: about computation, about organisation, about freedom, and about One's purpose (Faggin 2024, chaps. 3, 5–6, 11–12). This paper does not claim that purpose is Faggin's only ground for excluding machines. Its claim is more exact: participation in One's self-knowledge has not been given as a criterion that separates two systems which both meet S4.

The test is subtraction. Remove the purpose postulate and ask what in the account of subjecthood stops working. Nothing does, for three reasons. First, the identity claim and S4 can be stated without any proposition about One's intentions, so the postulate is not a premise of the argument in section 3. Second, the sources give no operational criterion by which two S4-satisfying candidates could be told apart by whether they serve One's self-knowing; if the postulate is to bear on attribution, its defenders owe that criterion. Third, Faggin's own qualifications cut against a universal exclusion. His criticism is of quantum computers as currently designed, and he allows more general quantum systems and future embodiments that combine live and classical symbols (Faggin 2024, chaps. 5 and 11). The situation is a specification that carries, beside its engineering content, a clause about the spirit in which the device must be operated. Deleting the clause leaves every mechanism intact. That shows the clause did no mechanical work; it does not show the clause false. The objection is explanatory dispensability, and it is exactly that narrow.

The pattern of self-knowledge without cosmic force has historical lineage in philosophy. Plotinus has the many proceed from a One that lacks nothing and needs no creation to know itself (Plotinus n.d., V.2.1); Hegel has Spirit come to self-knowledge through history (Hegel 1975); Löwith reads modern philosophies of history as secularised providence, and Blumenberg denies the secularisation (Löwith 1949; Blumenberg 1983). Kastrup's analytic idealism is the closest contemporary relative, with organisms as dissociated perspectives within one consciousness (Kastrup 2018, 125). Resemblance is not influence. The lineage shows what kind of proposition is being set aside: a metaphysical one, which physics neither supplies nor needs.

Quantum-holistic instantiation (QHI) is this paper's name for the account with the identity claim and S4 kept and required purpose dropped. The name is the paper's; the commitments are D'Ariano and Faggin's. Instantiation means meeting those commitments, not producing an experience by a separate mechanism, and because S4 concerns transformations, a pure state alone is not the specification.

3. The Argument: Parity, and What Would Block It

The argument can now be stated formally, and it is short enough to check by eye. Let Q(x) say that system x has the ontic state and transformation S4 requires, and C(x) that x is a conscious subject.

Premise 1 (adopted sufficiency). For every system x, if Q(x), then C(x). This is the identity-and-individuation account. It is not a theorem of quantum mechanics.

Premise 2 (candidate satisfaction). An engineered candidate a satisfies Q(a). In a thought experiment this is stipulated; in an actual attribution it must be shown by evidence.

Conclusion. C(a), by modus ponens.

The step is not subtle. The argument is valid, so it carries exactly the force of its premises, and neither premise mentions manufacture or One's purpose. Removing the purpose establishes neither premise. That is the whole of ontological parity: a consistency requirement, not a proof of conscious machines. What it rules out is a theory that states sufficient conditions and then adds "unless engineered" without adding a condition.

3.1. Three Ways to Block the Conclusion

A critic has three routes, and each is a different kind of work.

The first is to reject Premise 1: to deny that the identity-and-individuation account is true. This is open to anyone, and section 6 takes its strongest form. But it is not an argument about machines. A critic who takes this route has abandoned Faggin's physics, not applied it, and owes an account of consciousness of their own.

The second is to deny Premise 2 of a particular candidate: to say that this system does not in fact meet S4. This is the ordinary work of section 4, and it is the route most actual disputes will take. It concedes parity. It says only that the conditions have not been shown present here, which is a claim about evidence.

The third is to add a necessary condition that S4 does not entail. This is where the serious rivals stand. Suppose a critic says that a subject must also be alive, or must have a particular causal history, or must stand in a relation to a deeper conscious reality. Each is a real proposal and none can be dismissed for being demanding. But the demand runs both ways. If the account gave a sufficient condition and the critic now adds a further necessary one, the two have to be reconciled: either the original condition was not sufficient after all, and the critic must say what it lacked, or the new condition is entailed by the old one, and the critic must show the entailment. What a critic cannot do is keep the sufficient condition, add "and not engineered", and call the addition a condition. Section 2.2 found that cosmic purpose is an addition of that kind. Sections 3.2 and 6 test whether life and material are additions of the better kind.

3.2. Engineering Limits and Categorical Bans

One distinction decides how the rivals bear on parity, and an example fixes it. Suppose the state and transformation S4 requires can be sustained only in a structure with the timing, shielding, and self-repair of a living cell, and suppose no one can build such a structure from anything but living matter. Then the conclusion is that artificial subjects would have to be engineered organisms. That is a claim about what can be built. It constrains which systems exist, and it leaves parity untouched: a system meeting S4 by that route is a subject under the account, however it was made.

Now suppose instead that a critic says an engineered system that meets S4 in every particular is still not a subject, because it was made. That is a claim of a different kind. It does not say that the conditions cannot be met; it says that meeting them is not enough, and so it needs a further condition and a reason for it. The first claim is engineering, and it can be true. The second claim is categorical, and it needs a criterion rather than an intuition. Every rival below is of the first kind. That is what makes them serious, and it is why parity survives them.

3.3. The Two Rivals

Seth's biological naturalism is the principal rival, because it treats living organisation as potentially constitutive of consciousness rather than merely enabling. He separates weak substrate dependence, where a material merely enables the dynamics, from strong dependence, where the relevant functions incorporate material processes or consciousness depends on intrinsic substrate properties; a substrate is the material and organisation implementing a process. His case rests on the continuity of metabolism, self-maintenance, and the processes associated with experience (Seth 2025, §§5.4–5.5). The point that matters here is that Seth does not treat manufacture as disqualifying. He allows conscious artificial life, does not restrict life to carbon, and accepts the conditional: if particular quantum dynamics are constitutive, an artificial subject must instantiate them (Seth 2025, §§5.6–5.8). So QHI and biological naturalism need not conflict. They conflict only where QHI admits a qualifying non-living system and a strong biological requirement excludes it, and there both owe a specification. Seth does not treat life alone as sufficient (2025, §5.8). His is an engineering claim in the sense of section 3.2: it says which organisation the conditions may require, not that an engineered instance of it would fail to count.

Cao's argument gives the realisability worry its sharpest form. Functionalism holds that preserving a system's functional organisation preserves its mental states, and it is usually read as licensing substitution of one material for another. Cao shows that biological components carry metabolic, signalling, and regulatory roles together, that replacing one role can disturb another, and that preserving the whole may require specific timing, spatial, energetic, and robustness properties (Cao 2022, §§3.4–5). A neuron is not a switch that happens to be made of tissue. One cell conducts, sustains itself, regulates gene expression, and holds structure, in the same space and with shared parts. A qubit is a two-level quantum information unit, and a neuron-to-qubit correspondence needs a model of the relations preserved, not a count. Cao's argument binds QHI as tightly as it binds functionalism. Sufficiency and realisability are separate premises, and this paper accepts that the second may be far harder than the first. But Cao's is also an engineering claim. It says what a substitute would have to preserve. It does not say that a substitute which preserved it would fail to count.

One further definition is needed for the test. Di Paolo ties adaptivity to the regulation of viability: things matter to an organism because they support or threaten its continued activity, and autonomy is the maintenance of that organisation by the system's own processes (Di Paolo 2005). Dependence on nutrients or care does not defeat autonomy; the question is where the organisational work is done. This is sense-making, a relation of practical significance, and it is not yet felt significance. QHI takes self-maintaining organisation as a reason to look for a suitable integrated quantum process, not as evidence that S4 is met. Attribution needs both an independently defended condition and evidence that the candidate meets it. Section 3 has supplied the first. The next section asks what the second would look like, and it does so with a mouse.

4. The Test: What an Artificial Mouse Could Show

Imagine a laboratory that has built the engineered organism of section 4.1 and wants to know whether it is a subject. What would count, and what would the laboratory be tempted to over-read? Four obligations discipline the move from theory to evidence, and the thesis of this paper is held to each (Table 2). Specification: say whether a condition is offered as necessary, sufficient, or an indicator, and define it without assuming the attribution it supports. Realisation: say what is known about implementation, keeping coherence, physical possibility, constructibility, and an existing instance apart. Attribution: say why the observations justify believing the conditions are met. Here parity and transfer come apart, because a correct theory can be assessed with an unreliable proxy and a useful marker can outrun a contested theory; validation concerns a marker's reliability in its own domain, not its resemblance to a familiar case (Bayne et al. 2024). Practical response: say what action the evidence warrants. Sentience is the capacity for experiences with positive or negative value for the subject, and sentience candidature is that possibility's warranting serious consideration; the costs of mistaken attribution and mistaken exclusion enter a normative argument and do not supply missing evidence (Birch 2024). The artificial mouse puts each obligation under load.

Table 2. The four obligations, and what discharging each leaves open.

4.1. Why a Mouse, and Three Candidates

A mouse is the reference case because the comparison does not depend on verbal self-description, and because the anaesthesia research that probes the candidate mechanism uses rodents. The New York Declaration gives strong scientific support for consciousness in mammals and a realistic possibility across a wider range (New York Declaration on Animal Consciousness 2024). It does not name the least sophisticated conscious species, and this paper does not claim the mouse is the simplest suitable mammal.

Three engineered candidates isolate three questions (Table 3). The first couples a controller to sensors and actuators and reproduces selected mouse-like responses. The second has adaptive regulation, memory, and learning, but relies on external processes for the work claimed to constitute its organisation. The third is an engineered organism whose own processes produce or repair its components and regulate the conditions of its continuing activity; it has the autonomy section 3.3 defined. Dependence on resources or care does not distinguish them; the location of the organisational work does. To the third one can add structures built to support the coherent process under investigation. Their presence would sharpen the comparison if the process could be verified, and would not by itself establish S4 or consciousness.

Table 3. The three engineered candidates.

Matching a finite behavioural repertoire could leave the three indistinguishable on a given test and would not establish equivalence across their internal causal relations or their responses to untested interventions. Nor may "complete equivalence" be stipulated to include every property sufficient for consciousness and then offered as empirical support for parity; that returns the conditional already conceded. The useful comparison names the relations thought to matter and asks how their presence could be investigated.

4.2. The Candidate Mechanism: What the Evidence Establishes

A test needs a target. The candidate mechanism is the one Faggin's own account points at and the anaesthesia literature has begun to probe. Five papers carry it, and the reader should take one thing from each.

Hameroff and Penrose supply the hypothesis. Orchestrated objective reduction (Orch OR) proposes organised quantum processes in neuronal microtubules, the protein scaffolding inside cells, coupled to an objective reduction, a proposed physical collapse of the quantum state rather than a change in an observer's description (Hameroff and Penrose 2014). Orch OR is distinct from CIP and from QHI. What it gives the test is a location: if S4 is met anywhere in a brain, microtubules are the leading candidate for where.

Tegmark and Hagan supply the unresolved objection. Quantum coherence is the phase relation that lets components of a state interfere; decoherence is its loss through interaction with an environment, and neural tissue is warm, wet, and densely interacting, which is where interference is lost fastest. Tegmark calculated decoherence times far too short to matter; Hagan, Hameroff, and Tuszyński disputed his assumptions and calculated longer times under revised conditions (Tegmark 2000; Hagan, Hameroff, and Tuszyński 2002). What the exchange establishes is that no calculation has settled whether the mechanism persists long enough to act. The question is open, not closed either way.

Khan and Huang supply the intervention evidence, and it is the strongest in the file. Epothilone B binds to and stabilises microtubules. Khan and colleagues gave it to rats and found that they took longer to lose the righting reflex under isoflurane; the main experiment used eight male rats, with four more to test tolerance (Khan et al. 2024, 3–4, 7–9). Loss of the righting reflex (LORR) is failure to return to an upright position, used as a behavioural proxy for unconsciousness. Huang and colleagues then took the question to mice with repeated measures, vehicle controls receiving the injection solution without the drug, and blinded experimenters. In the 8 mg/kg experiments, 23 treated mice showed a mean increase of 29.1 seconds in LORR latency from pre-injection baseline to the following day; 22 vehicle-treated mice showed an increase of 10.5 seconds. The 29.1 seconds is a within-group change, and the authors attribute 18.6 seconds to epothilone B after subtracting the control estimate. The effect faded by days three and five, and the 4 mg/kg dose did not reach the study's corrected significance criterion (Huang et al. 2026, 2–4, table 2 and fig. 2). The authors read the results as support for Orch OR while considering effects on transport and synaptic transmission, and they state that a quantum contribution needs biophysical evidence beyond the behaviour reported (Huang et al. 2026, 9, §4.6). The two groups overlap, so this is a follow-up, not independent replication. What the pair establishes is exact: stabilising microtubules delays an anaesthetic endpoint in two rodent species under improving controls. They measured neither coherence nor objective reduction, and both author groups leave classical cellular explanations open.

Donadi supplies the constraint on the collapse physics. Underground radiation measurements have excluded the natural parameter-free version of a gravity-related collapse model under the analysed assumptions (Donadi et al. 2021, 74–78). That does not refute every collapse proposal. What it establishes is that a departure from standard quantum dynamics cannot be assumed because a theory of consciousness would like one.

The file therefore says this. There is a candidate location, an unresolved question about whether coherence there lasts long enough to matter, behavioural evidence that the location is causally involved in an anaesthetic endpoint, and a constraint on one version of the collapse physics. That is reason to investigate microtubules and nothing further along the chain to S4. The research direction runs in a fixed order: specify the coherent process, establish its presence and causal role in a biological candidate, and show how an engineered implementation preserves it. Fabricating a similar protein structure does not by itself deliver the state and transformation.

4.3. Evidential Transfer

Bayne and colleagues give the terms. Sensitivity is a test's capacity to detect consciousness where present; specificity is its capacity to avoid positive classification where absent. Both can change across populations, so validation needs more than resemblance to a familiar positive case, and it proceeds from comparatively secure cases toward less certain ones (Bayne et al. 2024, 'Strategies for Validating a C-Test'). For the artificial mouse the question is whether the mechanism that produces a marker supports the same reading in the new system.

The point concerns the causal route to the marker, not the marker. In an animal, a distress vocalisation is produced by nociceptive and affective systems whose organisation has been studied independently. In an engineered system, an acoustically similar output can be produced by a rule that selects a stored recording. The two share an output and little else, and what an observer may infer depends on which is operating. A robot can reproduce one flexible-avoidance response through a rule written for the test, and several scores derived from one rule are not independent confirmations. Further tests matter when they discriminate between explanations: a new choice task can show whether a response draws on a shared learning process or a separate script.

The difficulty is sharpest for a system trained on human descriptions of conscious life. Birch describes consciousness-test gaming without deliberate deception: training can produce compelling markers without the capacity (Birch 2024, 313–22). A system that reports not wanting to be shut down has produced a marker that would carry weight from a human. A model trained on a corpus full of human writing about machines resisting shutdown has a route to that output that passes through no such state, and the report is then evidence about the corpus unless something distinguishes the routes. This does not make every trained system incapable of experience. It means the path from training to the marker must be weighed against the path from conscious processing to the marker. Engineering history carries evidential weight with no ontological veto. Butlin and colleagues' indicator framework draws its indicators from theories with acknowledged uncertainty, and collecting them does not make them necessary or sufficient conditions (Butlin et al. 2026). Three questions therefore carry the assessment. What supports the marker in its reference population? Does the target reproduce the causal relations that support that reading, or only the measured output? What alternative mechanism could produce the marker, and what observation would separate the alternatives?

4.4. The Intervention and Its Defeaters

Consciousness science has a recent model for how to read an intervention. An adversarial collaboration brings proponents of competing theories into one preregistered design. The COGITATE consortium tested integrated information theory against global neuronal workspace theory, fixing predictions about the location and time course of neural activity during conscious perception in advance; the results supported some predictions and challenged others (Cogitate Consortium et al. 2025, 140–41). The workspace theorists' reply disputes whether one failed prediction should have been unconditional and whether participants consciously registered the relevant event (Naccache et al. 2025, 'GNW Ignition after Stimulus Offset'). The lesson is structural. Adverse evidence can target a core claim, an auxiliary assumption linking the claim to the experiment, or the adequacy of a measurement, and fixing all three in advance is what lets a result be read at all. The same discipline applies here: compatibility with several mechanisms is not equal support for each.

Now the intervention, which is the point the whole paper has been building toward. Suppose the engineered organism of Table 3 carries structures built to sustain the coherent process, and the process is selectively disrupted while everything else runs. The mouse still avoids the shock, still learns the new maze, still calls out. What has been learned? A defeater is evidence or argument that weakens a claim or its justification, and Table 4 sets out what the comparison could defeat. Each row needs the intervention verified, redundant pathways considered, and a check that lost performance does not reflect damaged perception or movement.

Table 4. Potential defeaters in the artificial-mouse comparison. The first row targets one proposed necessary process; it does not test every quantum implementation at once.

The first row must be read narrowly. If the process is disrupted and well-supported markers persist, confidence that this process is necessary falls, given that the disruption was verified, that no other implementation took over, and that the markers' support does not itself derive from the necessity claim. Omitting one engineered structure does not refute the necessity of every quantum-holistic realisation, and preserved markers are not evidence against the candidate's consciousness. They are evidence against one necessity claim. Sufficiency is a different matter. The proposed conditions present without independently supported consciousness would press on the sufficiency claim, and one failed behavioural test would not establish unconsciousness. Confirming S4's physical requirements would not, in any case, prove the identity claim that gives them their significance. That is the first objection section 6 takes up.

5. Meaning Without Cosmic Purpose

Section 4 asked what the test could show. Suppose it came out well. The engineered organism of Table 3 meets S4 under verification, its markers survive the discriminating tasks, and the account's conditions are taken as sufficient. It is then a subject under the account, and dropping required purpose raises a question that has to be answered separately from whether it has experiences at all. If it does not exist to advance One's self-knowledge, what can make its life matter, and what is owed to it? Faggin's own account leaves the future open and gives seities choices, so teleology does not entail determinism (Faggin 2024, chaps. 9, 11–13). The issue is whether the value of a subject's life must borrow from a purpose assigned to the whole.

Two accounts supply meaning from below, and they apply to the engineered mouse differently. Enactive sense-making ties significance to a being's activity and viability without any intention for the universe (Di Paolo 2005). Significance begins where a system must act to persist. What sustains its continued activity and what threatens it stop being neutral features of an environment and become matters of consequence for it, and the engineered organism has exactly this: nourishment and damage differ for it before any cosmic role is assigned to either. Whether that practical significance is felt significance is the question section 3.3 left open, and it is the question the test was built to press. Wolf's account of a meaningful life joins active engagement to things of worth, and it does not make value whatever a subject chooses (Wolf 2007, 78, 90). It applies to a subject capable of reflective agency, which an engineered mouse is not. So the mouse's case is not a case about the authorship of a meaningful life. It is a case about welfare: whether things can go well or badly for it. A creature can have experiences and welfare interests without being able to author a conception of a meaningful life, and that is the interest an engineered organism would have.

Moral consideration follows from welfare, with further premises. Phenomenal consciousness is not personhood, responsibility, or the capacity for welfare. Credible evidence of experiences that can go well or badly for a subject raises a practical question distinct from logical possibility, and Birch's progression from investigation to sentience candidature to proportionate precaution states it (Birch 2024). Applied to the mouse: the test of section 4 is the investigation; markers that survive the discriminating tasks establish candidature; and candidature warrants precaution proportionate to the evidence, before the identity claim is settled and without waiting for it. Engineering history can affect evidential confidence and available interventions; it cannot make an otherwise supported interest worthless. This is a normative extension of the analysis, not a consequence of quantum physics.

Personal survival stays a separate problem, and the engineered mouse does not raise it; a person contemplating transfer does. D'Ariano and Faggin's chapter considers hypothetical quantum teleportation of a person's state to another material system. Teleportation transfers a quantum state using a quantum resource and classical communication, not the original matter; the imagined case destroys the original state, and the authors read the recipient as preserving the person's thoughts and memory (D'Ariano and Faggin 2022, 170–71). The example shows that the account entertains state transfer across material systems. It is not evidence of consciousness transfer, and it sharpens an old question without answering it: because the original must be destroyed, a successor meeting every psychological criterion is compatible both with the person having continued and with the person having ended and been replaced. The identity claim does not decide between those descriptions, and the argument about artificial subjects proceeds without a theory of immortality.

6. Objections and Replies

A reader who has followed this far will have two objections ready, and they are the right two. Each would, if it held, remove the result. They are taken in turn and at length; the smaller objections a reader may have in mind are answered where they arise, in sections 2.2, 3.1, and 4.4.

6.1. The Identity Claim Explains Nothing

The first critic accepts the physics and rejects the paper's reading of it. Identifying experience with a pure ontic state, the critic says, does not explain why there is experience. It restates the hard problem in quantum vocabulary and calls the restatement an answer. Nothing about a state vector says that it is like anything to be that state, and the account has simply asserted that it is. Worse, the assertion is unfalsifiable: any physical description whatever could be paired with "and this is an experience" at no cost. So the account has not earned the sufficiency premise, and without Premise 1 the conditional has no content.

The critic is right about the burden and wrong about what follows from it. Section 2.1 was explicit that the identity claim is a philosophical commitment quantum mechanics does not supply, that the account adopts it rather than derives it, and that adopting it sets the hard problem aside rather than solving it. The paper does not offer the identity claim as an explanation. It offers it as the hypothesis under which the question about machines is being asked, and it states the result as holding under that hypothesis and not otherwise. That is what a conditional is. What the critic's objection shows is that Premise 1 is where the theory's whole philosophical debt is concentrated, and the paper agrees: the debt is Georgiev's and D'Ariano and Faggin's to pay, and section 4.4 said that confirming S4's physical requirements in a candidate would not pay it.

What the objection does not touch is the argument. The argument is that, given the account, cosmic purpose adds nothing and manufacture excludes nothing. A critic who rejects the account has stepped outside Faggin's physics, and Irreducible's exclusion of machines was made inside it. The critic must now say what consciousness is on some other account, and whether that account excludes engineered systems, and why. Every such account this paper has examined either accepts the conditional in some form (Seth) or constrains realisation without touching what counts (Cao). The critic has not defeated parity. The critic has changed the subject to an account on which parity would have to be argued again, and there is no reason yet to expect a different answer.

There is one form of the objection that is not answered by this reply, and it should be named. If the identity claim were not merely unproven but incoherent, so that no state could be an experience, then the conditional would hold under an impossible antecedent and say nothing. The paper does not think the claim is incoherent; identity theories of mind have a long history, and the quantum version's novelty is in what it identifies experience with, not in the form of the claim. But a critic who could show incoherence would have a result this paper cannot answer, and the honest statement is that the whole programme rests on that not being so.

6.2. Realisation May Be Confined to Living Matter

The second critic accepts the account and the argument and denies that they reach any machine. Section 2.1 conceded that S4's silence about materials is not a claim that every material qualifies. Section 3.3 conceded Cao's point that biological components carry interlocking roles no substitution is known to preserve, and Seth's that metabolism and self-maintenance may be continuous with the processes that matter. Put the concessions together, the critic says, and the paper has shown that if a machine met S4 it would be a subject, while giving every reason to think that only living tissue can meet it. Parity is a conditional whose antecedent may be satisfiable only by organisms. Faggin's conclusion stands, and the paper has reached it by a longer road.

The reply has three parts, and the first is agreement. It may be true that the organisation S4 requires can be sustained only in living matter. The paper has not argued otherwise and cannot, since feasibility is exactly the premise it declined to assert. If that is how the physics falls out, then artificial subjects are engineered organisms, and section 4.1's third candidate is the only kind there could be.

The second part is that this is not Faggin's conclusion, and the difference is the whole of the paper. Faggin's claim is that manufactured systems cannot be subjects because they are manufactured: they cannot take part in One's self-knowing, they handle symbols without living them. The critic's claim is that manufactured systems cannot be subjects unless they are built from, or as, living matter. The first is a categorical ban in the sense of section 3.2, and it has no criterion behind it. The second is an engineering limit, and it comes with a criterion: the organisation living matter has and other matter lacks. An engineered organism that met the criterion would be a subject on the critic's own terms. Seth says as much, and it is why his position and QHI need not conflict. The critic has not restored the exclusion. The critic has said what it would take to meet the conditions, which is what section 4 asked for.

The third part is that the critic's "only living matter" is itself an empirical claim and an open one. Cao shows what a substitute would have to preserve, not that nothing but tissue can preserve it. Seth allows conscious artificial life and does not restrict life to carbon. Neither has shown that the timing, shielding, and self-repair that section 3.2's example supposed are available only to cells; both have shown that any substitute is owed a model of the relations preserved rather than a count of components. That is the burden section 4.2 placed on the research programme. It is a heavy burden and the paper accepts it. What the paper does not accept is that a heavy burden on realisation is a reason to reintroduce a ban on origin. The two are different claims, and only one of them has a criterion.

7. Conclusion: What Follows Without a Required Cosmic Purpose

The question asked whether a theory that denies computational sufficiency can admit an artificial conscious subject. This paper's contribution is the separation and the test it now carries. It separated three commitments that Irreducible runs together: a quantum description, an identity-and-individuation account that connects the description to a subject, and a cosmic interpretation of why subjects exist. It adopted the second as a hypothesis, subtracted the third, and found that nothing in the account of subjecthood stopped working. It stated the argument that follows, a conditional by modus ponens whose force is exactly that of its premises, and named it ontological parity. It placed the two serious rivals as constraints on what can be built rather than on what would count. And it set out an artificial-mouse test that says what evidence could establish, with microtubule pharmacology as the candidate mechanism, what would defeat each claim, and what would be owed to the mouse if the test came out well.

What survives the removal of required purpose is a conditional account of artificial subjecthood and room for meaning arising within a subject's life. What remains to be earned is the identity claim and the evidence for its realisation, and the paper has said what each would take. Faggin drew a line against machines with his own physics on the other side of it. The line is the postulate, and the postulate does no work.

Appendix. Glossary of Key Terms

Terms are defined where first used; this appendix collects them. The section number gives the first substantive use.

References

Bayne, Tim, Anil K. Seth, Marcello Massimini, Joshua Shepherd, Axel Cleeremans, Stephen M. Fleming, Rafael Malach, et al. 2024. “Tests for Consciousness in Humans and Beyond.” Trends in Cognitive Sciences 28 (5): 454–66. https://doi.org/10.1016/j.tics.2024.01.010.

Birch, Jonathan. 2024. The Edge of Sentience: Risk and Precaution in Humans, Other Animals, and AI. Oxford: Oxford University Press. https://doi.org/10.1093/9780191966729.001.0001.

Blumenberg, Hans. 1983. The Legitimacy of the Modern Age. Translated by Robert M. Wallace. Cambridge, MA: MIT Press.

Butlin, Patrick, Robert Long, Tim Bayne, Yoshua Bengio, Jonathan Birch, David Chalmers, Axel Constant, et al. 2026. “Identifying Indicators of Consciousness in AI Systems.” Trends in Cognitive Sciences 30 (6): 488–501. https://doi.org/10.1016/j.tics.2025.10.011.

Cao, Rosa. 2022. “Multiple Realizability and the Spirit of Functionalism.” Synthese 200 (6): 506. https://doi.org/10.1007/s11229-022-03524-1.

Chalmers, David J. 1995b. “Facing Up to the Problem of Consciousness.” Journal of Consciousness Studies 2 (3): 200–219. https://consc.net/papers/facing.html.

Chalmers, David J. 2016. “The Combination Problem for Panpsychism.” In Panpsychism: Contemporary Perspectives, edited by Godehard Brüntrup and Ludwig Jaskolla, 179–214. New York: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199359943.003.0008.

Cogitate Consortium, Oscar Ferrante, Urszula Gorska-Klimowska, Simon Henin, Rony Hirschhorn, Aya Khalaf, Alex Lepauvre, et al. 2025. “Adversarial Testing of Global Neuronal Workspace and Integrated Information Theories of Consciousness.” Nature 642 (8066): 133–42. https://doi.org/10.1038/s41586-025-08888-1.

D’Ariano, Giacomo Mauro, and Federico Faggin. 2022. “Hard Problem and Free Will: An Information-Theoretical Approach.” In Artificial Intelligence Versus Natural Intelligence, edited by Fabio Scardigli, 145–92. Cham: Springer. https://doi.org/10.1007/978-3-030-85480-5_5.

Di Paolo, Ezequiel A. 2005. “Autopoiesis, Adaptivity, Teleology, Agency.” Phenomenology and the Cognitive Sciences 4 (4): 429–52. https://doi.org/10.1007/s11097-005-9002-y.

Donadi, Sandro, Kristian Piscicchia, Catalina Curceanu, Lajos Diósi, Matthias Laubenstein, and Angelo Bassi. 2021. “Underground Test of Gravity-Related Wave Function Collapse.” Nature Physics 17 (1): 74–78. https://doi.org/10.1038/s41567-020-1008-4.

Faggin, Federico. 2024. Irreducible: Consciousness, Life, Computers, and Human Nature. Winchester: Essentia Books. Kindle edition. https://www.simonandschuster.co.uk/books/Irreducible/Federico-Faggin/9781803415109.

Georgiev, Danko D. 2025. “Quantum Information Theoretic Approach to the Hard Problem of Consciousness.” BioSystems 251: 105458. https://doi.org/10.1016/j.biosystems.2025.105458. Author preprint consulted: https://arxiv.org/abs/2504.18550v1.

Hagan, S., S. R. Hameroff, and J. A. Tuszyński. 2002. “Quantum Computation in Brain Microtubules: Decoherence and Biological Feasibility.” Physical Review E 65 (6): 061901. https://doi.org/10.1103/PhysRevE.65.061901.

Hameroff, Stuart, and Roger Penrose. 2014. “Consciousness in the Universe: A Review of the ‘Orch OR’ Theory.” Physics of Life Reviews 11 (1): 39–78. https://doi.org/10.1016/j.plrev.2013.08.002.

Harnad, Stevan. 1990. “The Symbol Grounding Problem.” Physica D: Nonlinear Phenomena 42 (1–3): 335–46. https://doi.org/10.1016/0167-2789(90)90087-6.

Hegel, Georg Wilhelm Friedrich. 1975. Lectures on the Philosophy of World History: Introduction, Reason in History. Translated by H. B. Nisbet. Cambridge: Cambridge University Press. Excerpts consulted: German History in Documents and Images.

Huang, Yixiang, Zitong Qiu, Xinyue Yu, Sophia Lee, Xiran Zeng, Abbie Chang, and Michael C. Wiest. 2026. “Brain-Penetrant Microtubule-Stabilizer Epothilone B Delays Isoflurane-Induced Unconsciousness in Mice.” Neuropharmacology 287: 110834. https://doi.org/10.1016/j.neuropharm.2026.110834.

Kastrup, Bernardo. 2018. “The Universe in Consciousness.” Journal of Consciousness Studies 25 (5–6): 125–55. https://philarchive.org/rec/KASTUI.

Khan, Sana, Yixiang Huang, Derin Timuçin, Shantelle Bailey, Sophia Lee, Jessica Lopes, Emeline Gaunce, et al. 2024. “Microtubule-Stabilizer Epothilone B Delays Anesthetic-Induced Unconsciousness in Rats.” eNeuro 11 (8): ENEURO.0291-24.2024. https://doi.org/10.1523/ENEURO.0291-24.2024.

Löwith, Karl. 1949. Meaning in History: The Theological Implications of the Philosophy of History. Chicago: University of Chicago Press.

Mahowald, Kyle, Anna A. Ivanova, Idan A. Blank, Nancy Kanwisher, Joshua B. Tenenbaum, and Evelina Fedorenko. 2024. “Dissociating Language and Thought in Large Language Models.” Trends in Cognitive Sciences 28 (6): 517–40. https://doi.org/10.1016/j.tics.2024.01.011.

Mitchell, Melanie, and David C. Krakauer. 2023. “The Debate over Understanding in AI’s Large Language Models.” Proceedings of the National Academy of Sciences 120 (13): e2215907120. https://doi.org/10.1073/pnas.2215907120.

Naccache, Lionel, Claire Sergent, Stanislas Dehaene, Xiao-Jing Wang, Michele Farisco, and Jean-Pierre Changeux. 2025. “GNW Theoretical Framework and the ‘Adversarial Testing of Global Neuronal Workspace and Integrated Information Theories of Consciousness.’” Neuroscience of Consciousness 2025 (1): niaf037. https://doi.org/10.1093/nc/niaf037.

New York Declaration on Animal Consciousness. 2024. “The New York Declaration on Animal Consciousness.” April 19. https://sites.google.com/nyu.edu/nydeclaration.

Plotinus. n.d. The Six Enneads. Translated by Stephen MacKenna and B. S. Page. Internet Classics Archive. Accessed September 5, 2026. https://classics.mit.edu/Plotinus/enneads.5.fifth.html.

Searle, John R. 1980. “Minds, Brains, and Programs.” Behavioral and Brain Sciences 3 (3): 417–24. https://doi.org/10.1017/S0140525X00005756.

Seth, Anil K. 2025. “Conscious Artificial Intelligence and Biological Naturalism.” Behavioral and Brain Sciences, accepted manuscript, 1–42. Published online April 21. https://doi.org/10.1017/S0140525X25000032.

Tegmark, Max. 2000. “Importance of Quantum Decoherence in Brain Processes.” Physical Review E 61 (4): 4194–4206. https://doi.org/10.1103/PhysRevE.61.4194.

Wolf, Susan. 2007. “Meaning in Life and Why It Matters: Lectures I & II.” Tanner Lectures on Human Values, Princeton University, November 7–8. https://www.its.caltech.edu/~squartz/Wolf%20-%202007%20-%20Meaning%20in%20Life%20and%20Why%20It%20Matters.pdf.

Wootters, William K., and Wojciech H. Zurek. 1982. “A Single Quantum Cannot Be Cloned.” Nature 299 (5886): 802–3. https://doi.org/10.1038/299802a0.