Nested Agency and the Evolution of Intelligence: A VGF Analysis

The emerging scientific understanding of nested agency in biological systems offers a potentially important foundation for the VGF analysis of evolutionary intelligence.

In particular, it suggests that biological intelligence may be understood as developing through hierarchies of partially autonomous, goal-directed regulatory systems. These systems coordinate to generate higher-level agency, which in turn regulates the activity of the systems from which it emerges.

This is closely related to the VGF principle of the self-producing hierarchy of closures.

However, the relationship can be made more precise.

Central proposition: Nested agency provides an organisational architecture for the evolution of intelligence. Recursive modification of that architecture provides a possible mechanism for the evolution of increasingly sophisticated intelligence.

These are two distinct propositions. Their relationship may help explain the progression from cellular regulation through organic intelligence to symbolic intelligence, without requiring intelligence to be identified with consciousness.

1. Nested Agency in Biological Research

The concept of nested agency is closely associated with Michael Levin's research programme on the multiscale competency architecture of biological systems.

In this approach, biological organisation is understood not merely as a hierarchy of physical structures, but as a hierarchy of systems possessing different degrees of regulatory and problem-solving competence.

Different levels operate in different problem spaces.

Molecular systems regulate biochemical processes.
Cells regulate metabolic and physiological conditions.
Cellular collectives regulate anatomical development.
Organisms regulate their behaviour in an environment.
Social collectives may coordinate activities across multiple organisms.

The essential feature is that higher-level systems can influence the conditions under which their constituent systems operate, while depending upon the regulatory competencies of those constituents.

This introduces a significant perspective on biological evolution.

Evolution does not necessarily construct every new level of organisation from passive components.

It can construct new levels of agency by coordinating regulatory competencies that already exist at lower levels.

What is meant by agency?

In this context, agency does not necessarily mean consciousness, psychological intention, or subjective experience.

Instead, biological agency can be defined operationally.

Biological agency: The capacity of an organised system to regulate its activity in relation to an identifiable range of possible outcomes, using feedback and adaptive responses to maintain or achieve particular conditions.

This is a graded concept. A relatively simple physiological system may possess limited agency, while a complex organism can coordinate multiple goals across extensive spatial and temporal scales.

The distinguishing characteristic is not whether the system consciously experiences a goal, but whether its organisation enables flexible regulation in relation to that goal.

Consequently, agency can be investigated within the scientific register without presupposing consciousness.

Nested agency does not require nested consciousness.

2. Nested Agency and the Self-Producing Hierarchy

The self-producing hierarchy principle developed within VGF analysis describes a reciprocal relationship between levels of organisation.

A higher-level closure arises through the interactions of lower-level closures. Once established, the higher-level closure modifies the conditions under which those lower-level closures continue to operate.

This relationship is not simply bottom-up.

Consider the development of an organism.

Individual cells regulate their internal states. Groups of cells coordinate to produce tissues. Tissues participate in the formation of organs, and organ systems contribute to the organisation of the developing organism.

The resulting anatomical organisation then imposes constraints upon the cells and tissues that constitute it.

A cell integrated into an organised tissue may behave differently from the same cell placed in a different physiological environment.

Higher-level organisation therefore influences which developmental possibilities remain accessible to its constituent cells.

This corresponds to the VGF relationship between nested beta and gamma closures.

A schematic representation

Let beta represent a dynamically active domain of regulatory possibilities and gamma a relatively stabilised closure.

β₁ → γ₁ γ₁ + β₂ → γ₂ γ₂ + β₃ → γ₃

Each gamma closure represents a relatively stable organisation capable of participating in the formation of a further level of organisation.

These expressions are schematic rather than quantitative dynamical equations.

Importantly, the lower-level systems retain their own regulatory competencies.

They do not become inert components of the higher-level closure.

The higher-level organisation can therefore coordinate their activity without directly specifying every detail of their behaviour.

The virtual governor

This is closely related to the idea of a virtual governor.

A virtual governor need not be a physically separate controller issuing instructions to individual cells.

Instead, it may be a distributed regulatory pattern that coordinates numerous lower-level processes.

Examples of relevant biological mechanisms include bioelectric signalling, chemical gradients, mechanical interactions and other collective regulatory processes.

The higher-level goal is achieved through the coordinated competencies of the lower-level systems.

The VGF interpretation

A higher-level agent is a closure whose stability is generated through the coordinated competencies of lower-level agents, while its own regulatory organisation constrains, redirects and sometimes reopens the possibility spaces of those constituent agents.

We therefore obtain a hierarchy not merely of physical structures, but of regulatory possibility spaces.

This is the critical connection between nested agency and the evolution of intelligence.

3. How Nested Agency Can Generate New Levels of Intelligence

A useful distinction can now be made between the evolution of agents and the evolution of agency itself.

The evolution of agents concerns the appearance of organisms with new or increasingly sophisticated capacities.

The evolution of agency concerns changes in the architecture through which regulatory systems coordinate their activity to produce higher-level goal-directed organisation.

For VGF analysis, the second process may be especially fundamental.

The transition to multicellularity

An independently living cell possesses regulatory capacities directed toward maintaining its physiological conditions and supporting its continued existence and reproduction.

When cells become integrated into a multicellular organism, some of these capacities must become coordinated with the requirements of the collective.

Individual cellular regulation becomes coupled to a larger regulatory organisation.

The important innovation is not merely the presence of multiple cells.

It is the emergence of collective regulatory capacities unavailable to the individual cells acting independently.

The multicellular system can regulate anatomical structures, coordinate regeneration and maintain organism-level organisation despite substantial changes in its components.

This represents the formation of a new level of agency.

That new level can subsequently become the foundation for further evolutionary developments.

The expanding scope of agency

Levin has developed the concept of a cognitive light cone to describe the scope of the goals that a biological system can pursue.

A cell may regulate its immediate physiological environment.

A multicellular organism can coordinate regulatory activity across greater distances and timescales.

An animal possessing a nervous system can act in relation to distant environmental conditions, remembered events and anticipated future outcomes.

In VGF analysis, this can be interpreted as the expansion of the possibility spaces over which a closure exercises adaptive regulation.

But there is an additional feature.

As higher-level agency evolves, the competencies of the lower-level agents can themselves change.

The hierarchy therefore evolves at multiple levels through reciprocal modification of its constituent closures.

Evolution can generate new levels of agency, while those levels change the developmental and selective conditions under which further agency evolves.

This does not imply that evolution itself possesses an overarching goal.

Rather, evolutionary processes can produce systems with increasingly extensive capacities for goal-directed regulation.

4. Nested Agency and Recursively Evolving Agency

A further distinction is necessary.

Suppose a cellular collective possesses a higher-level anatomical goal, such as regenerating a particular structure.

Numerous cellular processes may be coordinated toward that outcome.

Different combinations of cellular activity may achieve the same anatomical result.

This illustrates nested agency.

Now suppose that the collective regulatory organisation undergoes a persistent change, so that subsequent regeneration produces a different anatomical configuration.

We have moved beyond the regulation of behaviour toward an existing attractor.

We are now considering modification of the attractor governing subsequent development.

Three degrees of agency

First-order agency: The capacity to regulate a system's trajectory toward an established attractor.

Second-order agency: The capacity to modify which attractor governs the system's subsequent regulation.

Recursive agency: The capacity to modify the regulatory organisation through which the system establishes, changes or selects its future attractors.

These distinctions are proposed analytical categories within VGF analysis, rather than established classifications in biology.

It is also important not to confuse externally imposed developmental reprogramming with autonomous recursive agency.

A regulatory pattern modified by an experimenter does not necessarily demonstrate that the organism can independently modify the rules governing its own goal formation.

Nevertheless, experimental reprogramming demonstrates that some developmental attractors are dynamically alterable without changes to the underlying genome.

This establishes a concrete biological basis for investigating higher-order regulatory organisation.

A hierarchy of regulatory capabilities

Organisation	VGF interpretation
Self-maintaining regulation	A system regulates its own persistence within an established attractor.
Nested agency	Interacting regulatory closures generate a higher-level regulatory closure.
Adaptive collective agency	The higher-level closure coordinates flexible responses among its constituents.
Self-modifying agency	Regulatory processes alter their own future organisation.
Recursive agency	The conditions governing regulatory modification themselves become modifiable.
Symbolic intelligence	Representations expand the range of goals and regulatory structures that can be reconsidered and modified.

These are distinctions between organisational capabilities, not necessarily a chronological sequence of evolutionary stages.

Some capacities overlap, and sophisticated adaptive regulation can exist without symbolic intelligence.

Key distinction: Nested agency establishes increasingly extensive domains of coordinated regulation.

Recursive agency makes the organisation of those domains itself subject to modification.

5. The Beta Domain and the Maintenance of Nested Agency

The relationship between nested agency and the VGF beta domain is especially significant.

A higher-level agent can be regarded as a relatively stable gamma closure in relation to its constituent agents.

However, those constituents retain their individual regulatory dynamics.

Their continuing activities form part of the beta domain through which the higher-level closure is maintained.

Because this domain remains dynamically active, it may also support changes to the higher-level organisation.

Closure through coordinated agency

Nested agency ↓ Higher-level closure ↓ Regulation of constituent agency

But the reciprocal process is equally important.

Reopening through constituent agency

Reopening of constituent regulation ↓ Modification of higher-level agency

Together, these processes describe the conditions under which a self-producing hierarchy may also become a self-modifying hierarchy.

The beta and gamma classifications remain relative to the level of organisation under consideration.

An individual cell may be a stable gamma closure relative to its molecular processes.

Yet the regulatory activity of that same cell may participate in the beta domain of a developing tissue.

The tissue can then become part of the dynamic organisation supporting organism-level stability.

Agency is therefore nested within the VGF in a manner analogous to the nesting of closure itself.

However, agency introduces an additional feature:

The constituent closures possess regulatory competencies that can participate in changing the organisation of the hierarchy.

Conflicting levels of agency

Nested agency does not automatically produce higher-level stability.

The regulatory competencies of constituent agents must remain sufficiently coordinated.

Cancer provides an instructive example.

The proliferation and survival of individual cellular lineages can become incompatible with the requirements of the organism.

Regulatory behaviour that favours the expansion of particular cells may become destructive to organism-level organisation.

Michael Levin has proposed interpreting aspects of cancer as a breakdown in the integration of cellular regulatory activity into higher-level anatomical goals.

This is an interpretative approach that complements, rather than replaces, established cancer biology.

In VGF terms, the coordination between cell-level and organism-level regulatory closures has become disrupted.

Nested agency alone does not guarantee higher-level intelligence or stability.

The relationships between agents must themselves constitute a viable regulatory closure.

The capacity to maintain, repair and modify those relationships may therefore be an important dimension of biological intelligence.

6. Nested Agency and the Evolution of Symbolic Intelligence

The nested-agency principle also provides a structural connection with the distinction previously developed between symbol-producing organic intelligence (SPOI) and recursively self-modifying symbolic intelligence (RSMSI).

In developmental morphogenesis, a biological collective coordinates cellular regulatory competencies to produce anatomical organisation.

In nervous-system intelligence, neural and physiological processes are coordinated to produce adaptive behaviour.

In symbolic intelligence, neural, conceptual, linguistic and behavioural processes become coordinated within more extensive forms of goal-directed organisation.

The underlying organisational principle may remain continuous across these domains.

What changes is the range and character of the possibility spaces over which coordinated regulation can operate.

Conceptual possibility spaces

Symbolic intelligence can construct representations of possible actions, anticipated outcomes, social relationships, goals and alternative futures.

Such representations can influence the organisation that generates them.

An organism can therefore regulate its behaviour not merely through immediate physiological feedback, but through the manipulation of symbolically represented possibilities.

The organisation of intelligence itself may become an object of symbolic investigation.

This extends nested agency into a particularly powerful form of recursive regulation.

The transition to reflexive intelligence

Within VGF analysis, a possible further development involves symbolic intelligence becoming capable of investigating and reopening its own stabilised symbolic organisation.

Consider the distinction between three kinds of question.

First: What goal should I pursue?

Second: Why does this particular goal arise within my intelligence?

Third: What regulatory organisation gives rise to my goals, and can that organisation itself be changed?

The third question introduces explicit investigation of the conditions governing goal formation.

In VGF terms, the symbolic closure becomes capable of influencing aspects of the beta domain through which its own organisation is maintained.

This does not establish a distinct post-symbolic evolutionary stage.

It does, however, provide a possible structural interpretation of increasingly reflexive intelligence.

Symbolic intelligence may extend nested agency by making aspects of its own regulatory organisation available for conceptual investigation and deliberate modification.

7. Intelligence and Consciousness: Keeping the Registers Distinct

One important advantage of the nested-agency approach is that it permits the development of a scientific account of evolutionary intelligence without presupposing consciousness.

The following questions can be investigated within the scientific register:

How do individual regulatory systems become integrated into collective agents?
How do collective agents acquire higher-level regulatory goals?
How do these goals constrain the activities of constituent systems?
How does regulatory organisation become adaptive and modifiable?
Under what conditions does such modification become recursive?

These questions concern biological organisation, adaptive regulation and evolutionary history.

They do not require us to assume that the constituent systems possess subjective experience.

Nor do they establish that consciousness emerges automatically from the integration of increasingly complex forms of agency.

Agency and consciousness remain conceptually distinguishable.

This is consistent with the methodological separation of registers developed in VGF analysis.

The scientific register

Nested agency belongs to the investigation of the evolution of biological intelligence as an objective organisation of regulatory competencies.

In the context of the VGF research programme, it concerns the evolution of biological organisation within the scientifically accessible domain.

The phenomenological register

The evolution of structured phenomenology remains a distinct trajectory of investigation.

Nothing in the concept of nested agency requires subjective experience to be reduced to biological regulation.

Nor does it require that every biological agent be regarded as phenomenologically conscious.

The correlation register

A separate investigation concerns the relationship between the evolution of biological intelligence and the evolution of structured phenomenological experience.

The trajectories may be deeply coupled without being collapsed into a single explanatory register.

Nested agency can be investigated entirely within the scientific account of evolutionary intelligence, while the relationship between agency and consciousness remains a separate question.

8. The Principle of Evolutionary Nested Agency

The preceding considerations allow us to formulate a proposed principle extending the self-producing hierarchy model.

The Principle of Evolutionary Nested Agency

Biological intelligence evolves through the formation of nested regulatory closures in which relatively autonomous agents become coordinated into higher-level agents capable of pursuing goals across more extensive possibility spaces.

These higher-level agents regulate the activities of their constituents while remaining dependent upon their continuing competencies.

Because the relationships between levels are dynamically maintained rather than permanently fixed, they can undergo reopening and reorganisation.

The evolution of intelligence can therefore be investigated as the progressive development of nested agency, culminating in systems capable of modifying not only their responses and goals, but also the regulatory organisation through which their future goals and possibilities arise.

This is proposed as an organising principle within VGF analysis, rather than as an independently established biological law.

9. The General VGF Interpretation

The development of the argument can now be summarised.

We began with the idea that the Vast Generative Field produces stable closures.

We then recognised that closures may preserve their organisation while retaining the capacity for reopening.

Organisms were subsequently characterised as closures that have evolved to regulate their own reopening.

Nested agency introduces a further level of understanding.

The evolution of biological intelligence involves the development of hierarchies of closures in which the capacity for reopening is itself distributed, coordinated and regulated across multiple levels of agency.

This gives intelligence a more specific interpretation than adaptation or problem-solving alone.

A nervous system is not necessarily the beginning of biological agency.

It is one evolutionary development in the architecture through which biological regulatory competencies are integrated into increasingly extensive forms of agency.

Symbolic intelligence represents a further development in which systems acquire new capacities to represent, investigate and potentially modify their own regulatory organisation.

At the most general level, the evolution of intelligence can therefore be investigated as the evolution of systems capable of changing the conditions under which their subsequent organisation becomes possible.

10. Three Related Principles

It is useful to distinguish three levels of the VGF interpretation.

A. Self-Producing Hierarchy

The general structural principle by which interacting closures produce higher-level closures that constrain and sustain their constituent organisation.

This principle may apply broadly to self-organising systems, including systems whose components do not possess identifiable goal-directed competencies.

B. Nested Agency

A special form of self-producing hierarchy in which constituent closures possess regulatory competencies that contribute to the goal-directed behaviour of higher-level closures.

Nested agency therefore introduces a specifically regulatory relationship between levels.

C. Evolutionary Intelligence

The evolving capacity of nested regulatory systems to adapt, reorganise and increasingly modify the possibility spaces governing their own subsequent activity.

Evolutionary intelligence in this sense is not identified with consciousness.

It concerns the changing organisation and competency of biological regulatory systems.

Conclusion

The distinction between self-producing hierarchy, nested agency and evolutionary intelligence preserves the generality of the IIP–VGF framework while giving biological intelligence a more specific character.

A self-producing hierarchy need not necessarily be a hierarchy of nested agency.

But when its constituent closures possess identifiable regulatory competencies, a new organisational possibility arises.

Higher-level organisation can coordinate those competencies to establish and maintain goals extending beyond those of its individual constituents.

Because the coordinating relationships remain dynamically maintained, they may themselves become subject to reopening and modification.

The hierarchy can thereby become not only self-producing, but self-modifying.

When such modification extends to the regulatory organisation governing future modification, the possibility of recursively self-modifying intelligence arises.

Final proposition:

Nested agency provides a general organisational architecture for biological intelligence.

The evolutionary development of intelligence can be investigated as the increasing capacity of this architecture to regulate, reopen and recursively modify the possibility spaces through which its own future organisation arises.

This provides a possible bridge between the VGF analysis of self-organisation in complex systems and the evolutionary development of intelligence in biological species.

It also identifies a fundamental research question:

Under what conditions does a self-producing hierarchy of nested agency become capable of modifying the regulatory relationships through which the hierarchy itself is generated?

That question may be central to understanding the emergence of recursively self-modifying intelligence within the VGF.

Selected Scientific References

McMillen, P. and Levin, M. (2024). Collective intelligence: A unifying concept for integrating biology across scales and substrates. Communications Biology.
https://www.nature.com/articles/s42003-024-06037-4

Levin, M. (2023). Darwin's agential materials. Cellular and Molecular Life Sciences.
https://doi.org/10.1007/s00018-023-04790-z

Levin, M. (2023). Bioelectric networks: The cognitive glue enabling evolutionary scaling from physiology to mind.
Find publication on Google Scholar

Brian Capleton