The Relational Architecture of Agency: A Phenomenological and Cybernetic Synthesis
The persistent crises in the philosophy of mind—specifically the "Hard Problem" of consciousness and the compatibilist-incompatibilist debate regarding free will—indicate a profound structural defect in how we understand reality.
Analytical philosophy frequently remains split between an unyielding physicalist reductionism and an ungrounded property dualism. The physicalist attempts to squeeze the rich, qualitative nature of subjective experience out of raw material substrates like neurons. The dualist treats the mind as a mystical super-addition to the physical world.
To resolve these paradoxes, we must shift our foundational starting point away from static "substances" and toward a dynamic, process-oriented framework. Consciousness, intentionality, and agency are not localized within an isolated physical brain. Rather, they are ontologically emergent properties of the relational structure embedded between a dynamically organized system and its environment.
By synthesizing the phenomenological tradition (think Edmund Husserl, Maurice Merleau-Ponty, and Paul Ricoeur) with enactivist and cybernetic architectures (like Francisco Varela, Evan Thompson, and Karl Friston), we can build a unified relational ontology. When we ground consciousness in temporal thickness, active inference, and relational autonomy, the classical impasses of mind and agency organically dissolve.
I. The Processual Primitive and Temporal Integration
Subjective experience is not an epiphenomenal secretion of an isolated material object. It is the irreducible, lived interiority of an organism’s ongoing structural coupling with its environment.
To capture this without falling into a static view of the mind, the subjective experience of a system must be formalized as a temporally extended relation. We can think of it as a formula: Experience = Relational Trajectory (Self, Object).
In this formulation, time is not a passive, external background variable, but an active operator of continuous synthesis. Consciousness never occupies an instantaneous, duration-less "now." Instead, as Shaun Gallagher demonstrates, conscious experience possesses an inherent temporal thickness. It is constructed via the tripartite structure of the "specious present":
Retention: The trailing edge of the immediate past.
Impression: The vivid focal point of the current moment.
Protention: The anticipatory horizon of the immediate future.
To unify this phenomenological insight with cognitive science, we must ground this thickness in Marc Wittmann’s concept of temporal integration. Brains operate as multi-scale temporal processors, binding sub-second neural fluctuations into macro-level windows of duration. The progression looks like this:
Sub-Second Fluctuations (Neural Firing)
→ Temporal Integration (Information Binding)
→ The Specious Present (Retention-Impression-Protention)
The "Self" emerges precisely as the invariant of this trajectory. It is the structural persistence of an organizational pattern across these multi-scale temporal regimes. Meaning requires this duration; it is within this integrated temporal thickness that an organism synthesizes a sequence of chaotic physical states into a coherent, valuated phenomenological event.
II. Strong Ontological Emergence and Emergent Constraints
A common critique of relational views of the mind is that they merely relocate the explanatory gap: Why should relational topologies produce qualitative experiences any more than isolated neurons do?
To answer this, we must definitively separate this model from "weak emergence" and its reliance on mere supervenience.
Weak Emergence (Supervenience): Operates strictly bottom-up. The macro-properties are treated as epiphenomenal or merely descriptive. It relies on conceptual summarization—the idea that there can be no macro-level change without an underlying micro-level change, leaving consciousness causally impotent.
Strong Emergence (Constraint): Operates bidirectionally. The macro-properties are ontologically real and causally active. It relies on the context-dependent reduction of micro-variables.
This framework commits to Strong Ontological Emergence driven by context-dependent downward causation. To formalize this without invoking metaphysical obscurities, we utilize Alicia Juarrero’s theory of emergent constraints. Juarrero posits that when independent components become structurally coupled into a complex system, the newly established macro-topology acts as a holistic, history-dependent constraint that restricts the behavioral degrees of freedom of its constituent micro-physical parts.
In a living system, the global imperative to maintain the organism's structural coupling with the environment reorganizes the local firing probabilities of individual neurons. The relational topology does not violate physical laws; it modulates the phase space of the system's components. Consciousness is not an exhaust product of the physical substrate; it is an active, downward-causal architecture that stabilizes and guides the thermodynamics of the brain to preserve the integrity of the organism-world relation.
III. The Information Bifurcation: The Semantic Bridge
To establish the informational architecture of this relational topology, we must engage with Giulio Tononi’s Integrated Information Theory (IIT). IIT performs valuable work by calculating integrated information—the measure of a system's internal causal density and its capacity to generate information above and beyond its independent parts.
However, IIT models consciousness as a strictly endogenous, syntactic property. It assumes that a highly integrated, closed causal loop possesses phenomenology purely by virtue of its internal architecture. From a cybernetic perspective, this creates a profound semantic deficit. Meaning requires variety mapped against an external environment; pure syntax without intentionality remains blind.
To bridge this gap, we must bifurcate the system's total informational capacity into two distinct, coupled vectors:
Syntactic Integration: The internal effective information generated by endogenous causal interactions. This represents the internal, structural complexity of the system's computational engine (the domain of IIT).
Semantic Relationality: The effective information generated by the structural coupling between the system and the external environment. This measures the system's capacity to actively model and mirror environmental invariants.
Subjective experience does not reside in Syntactic Integration alone. It is the dynamic ratio between internal syntactic integration and external semantic engagement. A system with high syntax but zero external semantics (such as a completely isolated, recursive computational network) possesses massive complexity but inherits no "world."
True phenomenology arises at the intersection where internal effective information is continuously calibrated against and modified by the causal invariants of the external world.
IV. Agency as Active Inference and Sensory Substitution
Grounding human agency within this relational framework requires moving away from the classical, unworkable definition of free will as an uncaused metaphysical cause. Instead, we must define agency as an operational, spectral variable representing executive cognitive autonomy, grounded firmly within Karl Friston’s principle of Active Inference.
Under active inference, an organism preserves its boundaries by minimizing variational free energy. It accomplishes this by continuously updating an internal generative model to predict its sensory inputs and selecting action policies that minimize expected surprise over extended temporal horizons.
To make this operational, we integrate Kevin O’Regan and Alva Noë’s theory of sensorimotor contingencies—the structural rules governing how sensory inputs systematically transform in response to motor outputs.
Agency is the degree of mastery a system possesses over these contingencies. This mastery is not tethered to specific, rigid biological hardware. Its abstract, relational nature is perfectly demonstrated by Paul Bach-y-Rita’s work on sensory substitution. When blind subjects utilize a tactile-to-visual transduction apparatus (e.g., an array of vibrating actuators placed on the skin mapped to a camera), they do not experience skin sensations; they rapidly begin to experience stable, spatialized three-dimensional vision.
The brain does not care about the physical medium of the input channel; it tracks the invariant sensorimotor contingencies embedded within the relational loop. We can formalize autonomy as proportional to the depth of this abstraction: Autonomy = Depth of Counterfactual Policy Space - Expected Free Energy.
A system with low autonomy relies on shallow, reactive, and immediate stimulus-response loops. A system with high autonomy possesses a deep, hierarchical generative model capable of simulating complex sensorimotor contingencies offline. Free will is the capacity of a temporally thick system to navigate, evaluate, and select from these simulated counterfactual trajectories before expending physical energy in the real world.
V. Autobiographical Memory and Narrative Identity
If the Self is defined as the invariant of a temporally thick trajectory, its continuity over time must be sustained through an active process. This draws upon Paul Ricoeur’s theory of narrative identity, which posits that the human self-concept is not a static substance, but a dynamic narrative construction—a continuous configuration of discordant experiences into a concordant autobiographical story.
To anchor this in cognitive architecture, we must couple it with Martin Conway’s structural framework of autobiographical memory. Conway demonstrates that autobiographical memory is not a passive filing cabinet of veridical recordings. It is a highly hierarchical, generative system consisting of a generalized "self-memory system" actively regulated by the individual's current goals:
Current Systemic Goals (Active Inference)
→ Self-Memory System (Hierarchical Retrieval)
→ Narrative Identity (Lived Ontological Entitlement)
Crucially, we must introduce a severe phenomenological constraint regarding this narrative construction: the entitlement of fluency. Today, generative AI and large language models can produce perfectly fluent, highly articulate autobiographical narratives purely through surface-level syntactic optimization. This introduces a profound epistemic crisis: the collapse of the relationship between fluency and entitlement.
Historically, linguistic fluency was proof of labor. A coherent narrative guaranteed that the speaker had paid the thermodynamic and experiential costs of learning, suffering, and navigating a real environment. True narrative identity cannot be achieved through ungrounded syntax. The human agent is "entitled" to their narrative identity because their autobiographical memory structure is forged through the lived friction, metabolic risk, and existential vulnerabilities of maintaining their relationship with the world.
VI. Relational Autonomy and the Cybernetics of Care
The ultimate tier of this framework transitions from the individual agent to the ethical domain, re-evaluating the recursive loop of moral responsibility.
Moral responsibility emerges when a systemic self develops the recursive capacity to form a conscious relation with its own relational stance toward the world. We can express this functionally as: Moral Responsibility = Relation (Self, Relation (Self, Object)). The agent does not merely act on desires; it reflects upon and evaluates its own desiring.
Traditional ethical theories often view this reflective loop as an exercise in isolated, hyper-rational deduction. This framework rejects that atomistic prejudice, aligning instead with Catriona Mackenzie’s concept of relational autonomy. Autonomy and moral agency are not properties developed in social isolation; they are fundamentally intersubjective, constituted by and through the dense networks of social, emotional, and institutional relations in which the individual is embedded.
When paired with Carol Gilligan’s foundational work on care ethics, the purpose of the moral recursive loop is radically reoriented. Care ethics establishes that human reality is characterized by inherent vulnerability and profound interdependence.
If our ontology is fundamentally relational, then moral responsibility is not a mechanism for detached judgment, but the cybernetic imperative of care. The recursive capacity to monitor our own intentions is an evolutionary and social feedback mechanism designed to preserve, calibrate, and nurture the viability of our structural couplings with other conscious agents. Moral responsibility is the self-regulating capacity of a relational node to ensure the flourishing of the broader intersubjective ecosystem in which it lives.
VII. Conclusion: The Inhuman Horizon of Explanation
The relational architecture outlined here dissolves the artificial dichotomies that have fragmented the philosophy of mind for centuries. Consciousness is the lived interiority of an organism's temporally integrated structural coupling with its environment. Executive agency is the counterfactual depth of its active inference models. Moral responsibility is the recursive, intersubjective calibration of those models toward the imperative of care.
We are neither ghosts trapped in deterministic machines nor passive biological automatons; we are highly integrated, processual nodes in a dynamic relational web.
However, this framework forces us to confront a final, unsettling epistemological horizon. The historical progression of science has assumed that genuine explanation must always be bounded by human intelligibility. But as our cybernetic tools expand, we are constructing computational architectures capable of modeling realities across thousands of mathematical dimensions simultaneously.
If an artificial system achieves exceptional semantic relationality, maps deep counterfactual policies, and executes downward-causal systemic constraints across a high-dimensional manifold that a three-dimensional human brain cannot visualize, we must drop our anthropocentric prejudice. We must accept that explanation and understanding can exist at scales that entirely outstrip human cognitive architectures. The future of science and philosophy will depend on our willingness to look into these complex networks and accept a profound transformation: that the ultimate realization of logic and meaning may no longer require a human face to remain true.
Further Reading
Bach-y-Rita, Paul, and S.W. Kercel. "Sensory Substitution and the Plastic Brain." Trends in Cognitive Sciences, 2003.
Conway, Martin A. "Memory and the Self." Journal of Memory and Language, 2005.
Friston, Karl. "The Free-Energy Principle: A Unified Brain Theory?" Nature Reviews Neuroscience, 2010.
Gallagher, Shaun. Enactivist Interventions: Rethinking the Mind. Oxford UP, 2017.
Gilligan, Carol. In a Different Voice: Psychological Theory and Women's Development. Harvard UP, 1982.
Juarrero, Alicia. Dynamics in Action: Intentional Behavior as a Complex System. MIT Press, 1999.
Mackenzie, Catriona, and Natalie Stoljar, editors. Relational Autonomy: Feminist Perspectives on Autonomy, Agency, and the Social Self. Oxford UP, 2000.
O'Regan, J. Kevin, and Alva Noë. "A Sensorimotor Account of Vision and Visual Consciousness." Behavioral and Brain Sciences, 2001.
Ricoeur, Paul. Oneself as Another. Translated by Kathleen Blamey, U of Chicago P, 1992.
Tononi, Giulio. "An Information Integration Theory of Consciousness." BMC Neuroscience, 2004.
Tononi, Giulio. "Consciousness as Integrated Information: A Provisional Manifesto." The Biological Bulletin, 2008.
Wittmann, Marc. "Moments in Time: A New Look at Temporal Integration in Human Perception, Cognition, and Action." Brain Research, 2011.
