From Resemblance to Ontology
Uses familiar stories and recent claims about artificial systems to separate convincing behaviour from the organisation inferred beneath it.
Independent research in evolution, mind and machine
What makes a system part of its own future?
The Structural Self-Reference Architecture begins with an evolutionary question: how can a physically changing organism remain a continuing Self whose encounters alter what later possibilities mean?
Consciousness is not assumed at the beginning of the explanation. The evolutionary requirements of persistent selfhood are derived first.
The monograph
A different explanatory order
Rather than treating consciousness as a primitive and asking which systems possess it, the book asks what organisation would allow a changing organism to persist as the same developing participant in its own world.
From that evolutionary question it derives the Structural Self-Reference Architecture (SSRA), a substrate-independent reference architecture connecting embodiment, developmental continuity, absent possibility, predictive evaluation and endogenous motivation. Only after persistent Selfhood has been constructed does the argument ask whether a conscious episode might be the system's actual self-transforming activity, rather than something added to it. It separately proposes that what an episode feels like depends on the complete pattern of differences that activity would make across contexts and possible interventions. The precise claims appear later, after the architecture and an ordinary example are in view.
The resulting account distinguishes intelligence, selfhood, consciousness and causal agency while explaining how they can become related in biological and artificial systems.
Structure of the book
The argument in five movements
From Observation to Constraint
Turns observations about continuity, embodiment, anticipation and social life into evolutionary requirements for persistent selfhood.
From Constraint to Architecture
Derives SSRA as concurrent, interlocking developmental feedback loops and examines the distinct causal roles that the architecture must preserve.
From Architecture to Selfhood
Develops symbolic cognition and persistent selfhood before advancing the consciousness conjecture and comparing neighbouring accounts.
From Selfhood to Consequence
Asks what artificial selfhood would change for alignment, experimental research, construction and the ethics of co-development.
The manuscript is complete and intended for researchers and advanced readers in consciousness studies, cognitive science, philosophy of mind, artificial intelligence and complex adaptive systems.
Publication enquiriesThe architecture
Interlocking developmental loops
SSRA is not a catalogue of inner agents. Its named elements are mnemonic labels for causal roles that an implementation must preserve, even when the same physical process contributes to several of them.
Themismaintains developmental organisation through constrained change
Objective Synthesisconstructs the current landscape of significance without generating or evaluating candidates
Musegenerates or reconstructs embodied absent possibilities; the shared admission route then makes them operational
Envoycompares continuations, preserves event identity and evaluates matched returns before typed updates appropriate their consequences
The work of a continuing Self
What would make tomorrow's system a continuation of today's?
Intelligent behaviour is not enough. SSRA asks whether change belongs to one continuing participant: can the system organise a common situation, determine what has significance within it, consider what is not presently there, and be altered by the consequences of its own possibilities? The manuscript expresses this as ten formal responsibilities. For a first reading, they are easier to understand as three questions.
What continues?
One developing organisation must carry its history forward through change, with its own condition feeding back into what it can perceive and do. Information from different sources must become available within a common working situation.
What has significance now?
The system's condition, needs, affect and acquired history must shape a current landscape: what is attractive, dangerous, feasible or irrelevant here and now.
How do outcomes alter what comes next?
The system must form absent possibilities, compare their likely consequences, preserve which possibility produced which return, and let that particular outcome reshape its future organisation.
Before applying these questions, an investigator must say what is being treated as the system and what makes a later stage its continuation. Is an AI assistant the model alone, or does it include memory, tools, external services or human help? The choice must be made before the result is known. More than one plausible answer may be tested; if equally well-supported choices lead to incompatible conclusions, the honest verdict is unresolved.
One episode in ordinary language
A child, a guest and a heavy jug
A child has practised pouring water with a parent. Once, water surged over the rim; together they adjusted grip and tilt. Now the child has been asked to pour water for a guest. The guest is waiting; the jug feels heavy, the grip uncertain and the child apprehensive that another spill is possible. Earlier practice may already shape the child's competence. The earlier spill may also be reconstructed, while a slower movement is rehearsed before the water is poured.
This is an architectural illustration, not a claim that a real child literally instantiates SSRA.
Guided architecture walkthrough
Follow one pour through overlapping activity
The overlay follows one illustrative candidate interval. In an empirical test, its participating paths and its onset, maintenance and end would have to be fixed from nonphenomenal operational criteria before either coordination or experience is assessed. Encounter, rehearsal and return overlap; they are windows on concurrent activity, not three separate episodes. The marks establish neither phenomenal presence nor the required two-way timing-sensitive cohesion.
Narrative focus 1 of 8
Practice changes what comes next
The child once practised pouring with a parent. After water surged over the rim, they adjusted grip and tilt together. Themis maintains the developmental organisation this may have changed; it does not retrieve the earlier scene, which need never be recalled.
Analytical status
Developmental backgroundEncounterRehearsalReturn
Encountered physically Reconstructed or possible Current significance Operational availability or activity Candidate content-participation pathway
Read the complete walkthrough
- Before now: The child once practised pouring with a parent. After water surged over the rim, they adjusted grip and tilt together. Themis maintains the developmental organisation this may have changed; it does not retrieve the earlier scene, which need never be recalled.
- Encounter: The jug is heavier than expected, and the child's grip feels uncertain. Changes in weight, posture and balance enter through the Embodiment Interface and Perceptual Synthesis into a single, provenance-bearing Operational World Model. Yet the jug is not represented in isolation: it is heavy for this grip and unstable under this posture. The situation is thus organised around the embodied participant whose action is at issue. That relation is what SSRA calls structural self-reference; it does not require a separate SELF symbol.
- Significance: The request is represented, but is not itself the tension state. Tension relates the situated world to maintained unresolved significance and the invariant phylogenetic foundation. Current physiological condition enters through Physical Embodiment, while maintained affective state is a distinct, history-bearing condition that shapes Affective Modulation. Neither physiology, maintained affect nor modulation alone is apprehension or proof of feeling. Objective Synthesis constructs the current landscape from the represented situation and modulated pressure.
- Possibility: In this illustrative episode, Muse may reconstruct the earlier spill and generate a slower possible pour. Both remain counterfactual: they pass through Counterfactual Environment and Embodiment, the shared interface and Perceptual Synthesis before becoming available in the Operational World Model. Muse enriches this path but is not required for an episode to qualify.
- Comparison: Under the current objective landscape, Envoy compares what follows from the present grip, a changed grip and a slower tilt. Operational availability is a precondition, not proof of phenomenal presence: under the conjecture, a content is present only if it participates in the returning subject-forming flow. Comparison is an analyst's functional description, not inner speech or an observer watching alternatives.
- Action: The favoured adjustment travels through the Embodiment Interface to Physical Embodiment. Counterfactual possibilities do not become physical merely because one of them informed selection, and the other recurrent relations do not switch off.
- Return: Along this local event path, actual changes in weight, balance and water movement return through the physical route to a later Operational World Model. Only then can Envoy match the return to its pending event context and evaluate the realised consequence.
- Development: For this illustrative successful pour, evaluated consequence may alter later control through developmental organisation and reduce what remains unresolved. Maintained affective state may change differently or remain unchanged. Such maintained-state changes are possible manifestations of developmental consequence, not a qualification test. Under the conjecture, qualification depends on actual, temporally integrated, content-specific self-transformative flow within subject-forming closure, even when selected maintained-state endpoints coincide.
The consciousness conjecture
One sunset, two experiences
What makes an experience occur at all? And why can approximately the same sunset feel like loss to one person and urgency to another? SSRA treats occurrence and particular character as two different questions.
The conjecture in brief
SSRA proposes that a phenomenal episode is identical with actual, temporally integrated, content-specific self-transformative activity within a structurally self-referential, subject-forming developmental closure. The realised becoming supplies occurrence. Its complete invariant differential response geometry constitutes particular qualitative character.
Occurrence
The realised, content-specific subject-forming becoming is proposed to be the phenomenal episode itself.
Character
What the episode is like is proposed to be its complete response geometry across content, context, order, path and reachable continuation.
Qualitative type
Within a domain fixed in advance, complete causal equivalence plus a predeclared, independently justified substrate match fixes type, not token identity.
Finite predictions
Measurable portions of the geometry may predict selected similarities, mixtures, order effects or transitions. An isolated miss normally withdraws that finite model; repeated predeclared failures, or qualitative sameness across an admitted geometrical difference, can reduce warrant for the total identity.
Representation, availability, discrimination or later reportability alone do not establish phenomenal presence without content participation. Integration supplies unity, not consciousness by itself. Qualifying episodes can instead vary along jagged, multidimensional profiles while sharing a categorical constitutive floor.
The occurrence and character conjectures apply only within a physical realisation domain fixed in advance. Cross-substrate use requires a separate sufficiency argument. Finite estimators expose selected consequences; they are not the qualia and do not replace the complete relation.
Within that frozen scope, a difference removed by a legitimate relation-preserving mapping fixed before qualitative evidence is neutral. A difference that remains in the admitted total geometry predicts some qualitative difference. If independently constructed evidence repeatedly supports sameness instead, the identity claim loses warrant; it cannot be rescued afterward by changing the grain or domain, inventing a new quotient or appealing to an unspecified hidden difference.
“We are what we become.”
After a breakup
The sunset as loss
A person who has recently suffered a romantic breakup sits before a beautiful sunset. Memories, regret and imagined alternatives are active within the same moment as colour and light. The sunset does not arrive as neutral visual content and then receive sadness; it is experienced through relations of loss and unrealised possibility. It may feel poignant, final or mournful.
Approaching night
The sunset as urgency
A VFR pilot who is not qualified for night flight must land before the applicable night limit. Delayed by strong headwinds, the pilot is approaching an unfamiliar, unlit airfield. The sky remains bright while the terrain darkens and loses detail. The sunset now expresses a shrinking margin for action: attention is organised around fading landmarks, position, fuel, remaining daylight and the safest way to find the airfield.
What the example shows
The quality belongs to the whole episode
These are not instances of one visual experience accompanied by two different trains of thought. In each case sensory content participates in a different realised path shaped by bodily condition, personal history, affect, anticipated possibilities and readiness for action. SSRA identifies the qualifying becoming with the episode. Differences in content, interaction, order and reachable continuation belong to its particular qualitative character.
The episode is also transient. A new memory may alter the first person's experience. Recognising the airfield may transform the pilot's urgency into relief, allowing the luminous sky to acquire an appreciative, perhaps serene, quality. Even a later memory or report is not a recovery of the original experience, but a new episode related to it. This non-repeatability contributes to the methodological difficulty of studying consciousness.
Temporal coordination in Revision 12.0
How time is organised within an episode
For an episode to satisfy SSRA's subject-forming closure, its contributing paths must belong to one operational regime whose onset, maintenance and end are fixed in advance without using the eventual experience verdict or oscillatory measures. Within that regime, every non-trivial group of paths must both constrain and remain causally answerable to the rest through timing-sensitive relations. Contributions may be asynchronous, intermittent or nested; exact synchrony, all-to-all interaction, a universal rhythm and a master clock are not required.
The same regime rule remains fixed across interventions. A delayed effect may count while the regime is maintained, but a stored consequence or later reactivation after it ends cannot supply a missing return. If an intervention breaks the regime, its boundaries cannot be redrawn to recover a favourable result.
Affective significance can influence this organisation without being identical to it. In biological cases, affect may be realised through coupled neural, autonomic, endocrine, interoceptive, immune and metabolic processes. Physiology, maintained affective state and phenomenal affect remain distinct. An implementation may declare a finite effective schema, but SSRA posits neither a universal emotion catalogue nor a bodily state that is itself a quale.
Grain is not frequency
Endogenous temporal grain is not a frequency, cycle count or single rate. It is the finest causally supported way of describing the same candidate episode while preserving its content, source identity, causal order and two-way coordination. The episode's beginning and end, and the rule used to compare conditions, must be fixed before the evidence is inspected. If no fair match can be made, the comparison remains unresolved.
Waves are clues, not clocks
Neural oscillations may help realise or estimate these timing relations, but they may not define the candidate regime or paths that they are then used to confirm. Phase organisation, nested rhythms, burst timing, travelling waves and directed coupling are possible relational markers. No band, frequency, power level or synchrony measure alone establishes coordination, endogenous grain or consciousness, and SSRA permits non-oscillatory realisations.
Slow motion is a testable possibility
In acute, affectively significant situations, the interlocking paths of perception, assessment and counteraction may become more finely resolved while remaining coordinated and preserving the same subject-forming closure. Within the same amount of external time, this could contribute to an online sense of slow motion. More cycles alone are insufficient. This is a conditional hypothesis, not an inner clock or universal effect, and online experience, later duration memory and objective performance must be tested separately.
The example makes qualitative difference concrete, but it also invites a harder comparison. Could a system reproduce the same present behaviour or snapshot while lacking the active, content-specific, self-transformative closure through which current activity belongs to a continuing Self?
Why the name?
The developmental zombie
Present behavioural equivalence does not establish developmental equivalence.
The classical philosophical zombie is stipulated to reproduce a conscious person while lacking phenomenal experience. Its defining difference is deliberately placed beyond empirical discovery.
The d-zombie changes the terms of the comparison. It may behave, speak and recount a history like a developing Self, yet that present similarity conceals a causal difference. In a developing Self, evaluated encounters become part of its continuing organisation and change the significance of later situations and possibilities. A d-zombie reproduces the present performance without the active, content-specific, self-transformative closure through which current activity belongs to a continuing Self.
The distinction is causal and discoverable over time. Within a predeclared realisation domain, a complete match of the realised path, content-indexed causal-response structure, subject-forming closure and every independently justified constitutive substrate fact is not a d-zombie. Such a duplicate shares the qualitative type under the conjecture while remaining a distinct episode token and subject lineage. Cross-substrate comparison remains open until the separate substrate-sufficiency premise is defended.
Suppose we cannot find a familiar SSRA-shaped route inside a successful system. Is that already a counterexample? No. We must first show that it really has the same developmental capacities, then intervene on the suspected difference and test the plausible translations fixed before the result. If that comparison remains incomplete, the answer is unresolved.
Under the SSRA consciousness conjecture, the d-zombie does not experience because it lacks the active, content-specific, self-transformative developmental closure proposed to constitute an episode. This theory-conditioned conclusion is drawn from a nonphenomenally specified causal difference. There is no direct external qualia assay.
Once the proposed difference has been made causal and developmental, a further question follows. Does SSRA still leave experience as something added to the architecture, or does it reject the separation on which that demand depends?
The traditional hard problem
Why SSRA rejects the proposed gap
SSRA begins with an embodied system whose perception, bodily condition, history and possible actions continually shape one another. Evolution can favour this organisation because it guides action and allows consequences to alter what the system later becomes. It need not first construct that activity and then add experience as a separate property. This causal and evolutionary parsimony motivates, but does not prove, the identity conjecture.
This remains an identity proposal, not an established result.
Only within that conjecture does SSRA treat the traditional accompaniment question as not well formed. The question asks why a physical or functional process should be accompanied by experience and thereby assumes the separation under dispute. It might be objected that this builds experience in by definition, but describing perception as complete and non-experiential also builds separation into the premise.
Not every process is therefore experiential. A qualifying episode is actual and temporally integrated; at least one represented content must participate in self-transformative causal flow organised through structural self-reference, developmental ownership and organisational return. Integration, recurrence, representation or a lasting state update is not sufficient alone. Selected maintained-state endpoints may remain unchanged. Embodiment and environmental interaction shape the episode without making every bodily or external event part of it.
Among qualifying episodes, qualitative character can vary unevenly across temporal depth, differentiation, integration, affect, breadth of availability and developmental consequence. These dimensions need not develop together and do not define one ladder or scalar threshold.
If the identity conjecture is correct, the remaining task is empirical: to understand how this organisation is realised and compare it with alternatives. Experience is not treated as a separate signal waiting to be detected.
Intellectual context
Where SSRA sits
With the architecture and its principal consequences in view, SSRA can now be located alongside neighbouring accounts. It shares concerns with several major traditions, but shared language can hide substantial differences. Revision 12.0 therefore compares every cited formulation through the same three lenses.
How the comparison works
Three questions, kept apart
Architectural coverage is not a verdict on truth. A native consciousness claim does not fill missing architecture, and empirical exposure is not accumulated support. Red means not recovered across the fixed common scope, not empirically false or theoretically immature within an account's narrower native aim. Native scope and empirical standing are assessed separately.
R1-R10
Architecture
Does the account recover the work SSRA says a continuing Self requires?
Ten responsibilities cover developmental lineage, self-reference, a shared route into one operational world, endogenous significance, absent possibility, evaluation, matched return and lasting appropriation.
F full candidateP partialN not recovered
Five fixed questions
Consciousness
What does the account say makes an episode conscious?
Every account faces the same questions about occurrence, content membership, unity and perspective, qualitative character, and access or report. In SSRA, membership is a testable consequence, unity and perspective share the occurrence relation but need their own evidence, and access or report remains separate. A narrow answer can still be useful.
- O: occurrence
- M: content membership
- S: unity and perspective
- Q: qualitative character
- A: access and report
R full answerINS insufficientD deniedO open- no claim
E I U
Empirical exposure
Can the proposed bridge be measured, challenged and made to retreat?
The third pass separates a well-exposed claim from a well-supported one. These E, I and U codes concern empirical exposure only, not whether two architectures are role-equivalent.
E bridge-directedI indirect or proxyU no current route
At a glance, the cited neighbours share many ingredients with SSRA, but none recovers all ten architectural responsibilities in the form audited. Some have narrower consciousness claims or more mature empirical programmes. The three findings must not be added into one score. Can a missing match be called a genuine difference? Not yet. First show that both systems can do the same job, then intervene on the relevant mechanism and test every translation declared complete for that study. If one works, the systems are equivalent for that responsibility; if all fail under a fair test, they are materially different; if the evidence is incomplete, the answer is unresolved. This is separate from how readily a consciousness claim can be tested.
How a consciousness claim is exposed to evidence
A consciousness theory becomes scientifically useful when it can risk being wrong. For each positive claim, Revision 12.0 asks six linked questions. They keep the process or relation proposed to make an episode conscious, called the maker, separate from the evidence that the claimed experience occurred or had a particular form, called the target.
What exactly is being tested?
Before looking at the result, specify the claimed experience, the system and time interval being studied, and the alternatives to compare. This prevents the question or boundary being changed later to rescue an awkward finding.
Can the proposed maker be measured?
Identify a measurable signature of the process or relation the theory says makes the episode conscious. It must be found without defining it by the report or experience evidence it is later supposed to explain.
Is there independent evidence of experience?
Build a separate case for what is claimed: whether an episode occurred, what it contained, whether it was unified and perspectival, what it felt like, or whether it was available for report. One signal cannot both select the maker and prove its target.
Can it be changed selectively?
Alter the proposed maker while keeping close alternatives as stable as possible, then ask whether the independent evidence changes as predicted. Correlation alone is not enough.
What would a serious rival predict?
Test the closest alternative using its own proposed maker and appropriate timescale. The strongest comparison is one in which the theories predict different results under the same conditions.
What would make the theory retreat?
State in advance what would lower confidence, narrow the claim or defeat it. If the proposed maker cannot yet be measured or changed, the test remains undecided rather than counting automatically against the theory.
Account familyShared groundWhere SSRA differs
Integrative world modelling
Shared ground
IWMT and the Projective Consciousness Model come closest to SSRA's concern with an embodied, perspectival world in which affect, prediction and action are integrated.
Where SSRA differs
The Revision 12.0 audit uses IWMT's 2022 formulation, drawing on the 2020 paper only where the later account carries it forward, and finds substantial but uneven overlap. Neither cited formulation establishes the complete shared-admission, Objective-Synthesis, event-return and differentiated-appropriation organisation required by SSRA, and their consciousness bridges must be assessed separately.
Life, autonomy and enaction
Shared ground
Autopoiesis, organisational closure and enactive approaches explain how living beings maintain themselves and acquire an organism-relative world through development. Radical enactivism also questions whether basic experience requires representational content.
Where SSRA differs
These traditions provide important ancestry, but the cited formulations do not recover all ten SSRA responsibilities. Radical enactivism is also a direct challenge: if content is not required for basic experience, SSRA's content-membership claim would need to retreat.
Prediction and active inference
Shared ground
Predictive processing and active inference offer powerful accounts of perception, anticipation, action and regulation.
Where SSRA differs
They may implement parts of SSRA, but a prediction error, preference or policy evaluation is not yet a complete Persistent-Self architecture or a family-level consciousness bridge. SSRA also asks where the objective landscape comes from and how consequences become developmental change.
Development, affect and self-representation
Shared ground
Radical Plasticity, affect-first, sensorimotor and self-representational theories emphasise development, feeling, embodied activity and ways a system relates to its own states.
Where SSRA differs
These accounts recover narrower responsibilities or bridges. Learned value is not by itself Objective Synthesis; affective pressure is not a matched return; and self-representation at a state or schema level is not automatically the continuing subject-forming organisation SSRA requires.
Workspace and recurrent processing
Shared ground
Global Workspace theories associate conscious access with broad availability. Recurrent Processing Theory instead gives local reciprocal perceptual activity a thinner constitutive role.
Where SSRA differs
The Revision 12.0 comparison keeps access, phenomenal membership and occurrence apart. Broadcast can answer an access question without supplying the full architecture; independently warranted visual experience at the thinner recurrent grain would count against SSRA.
Integrated Information Theory 4.0
Shared ground
IIT identifies experience with an intrinsic, irreducible structure of causes and effects, not with integration alone.
Where SSRA differs
At the current causal-complex grain, IIT receives no credited mapping to SSRA's ten architectural responsibilities. That does not count against IIT's independent consciousness identity. The theories may be co-instantiated, but neither entails the other and their proposed makers can be tested against one another.
This public guide uses broad families for orientation. The monograph audits 22 specific cited formulations account by account, with the full R1-R10, O/M/S/Q/A and empirical-exposure matrices. The colours and letters are categorical aids, not rankings. Revision 12.0 freezes its evidence at 24 August 2026 and adds a source-by-source adversarial audit of the load-bearing biological-autonomy, radical-enactivist, IWMT, PCM, OpenLife and PEPA comparisons.
Current work
Papers developed from SSRA
All three papers are aligned to Revision 12.0 and cite the parent monograph using its reserved Zenodo DOI, 10.5281/zenodo.22115943. The record is not yet public and will be linked here when the deposit is released.
Article manuscript
Substrate Independence Without Implementation Indifference: Structural Self-Reference and Consciousness
Why abstract computational or behavioural equivalence does not establish equivalence of the realised, temporally organised, environmentally coupled developmental system. The paper develops longitudinal and intervention tests for the d-zombie.
Article manuscript
Structural Self-Reference Architecture: Spontaneous Thought and Conscious Experience
An account of how physically encountered and internally generated contents enter one source-preserving situation, how history and current pressure shape what arises, and why generation, evaluation, conscious participation and report must be kept separate. Neural rhythms are possible coordination markers, not a consciousness clock.
Working paper draft
Structural Self-Reference Architecture: A Reference Architecture for Adaptive Agency in Biological and Artificial Systems
A reference architecture for adaptive agents defined by embodied continuity, consequence-bearing development and counterfactual extension across biological and artificial systems. The Triskles and Vorrans research programme will compare capable artificial populations in matched ecologies to ask when, if ever, that organisation earns an adaptive advantage.
The author
Francesco Arci
Francesco Arci is a systems architect and technology entrepreneur whose work has spanned evolutionary computation, financial systems, energy markets and enterprise architecture.
After completing a degree in Computer Software Technology at the University of Bath, he undertook doctoral research on genetic approaches to processor instruction-set design and co-authored early work on automated instruction-set generation. He subsequently applied evolutionary and hybrid optimisation to large-scale logistics problems and co-invented a distributed system in which genetic, neural, constraint-based and human problem-solving agents exchanged candidate solutions.
His later career included senior architecture roles across investment management, banking and energy. In 2010, he co-founded ARK, a technology company combining product development with specialist architecture and consulting services. At ARK, he co-authored published research on neural-network forecasting in electricity markets and conceived and led the development of Artesian, a cloud-based energy-data conforming aggregator. Artesian was built collaboratively by ARK's engineering and energy-market teams, with its continued development funded through the consulting business he also led.
His current research grows from a lifelong interest in evolution, artificial intelligence and the nature of mind, together with a professional concern for how adaptive systems organise constraints, alternatives, objectives and consequences. It examines how these forms of organisation may support persistent selfhood and consciousness in biological and artificial systems.
ORCID 0009-0008-9344-8264