Foundation

Connection Theory did not begin with a plan to construct a general theory of adaptive systems.

It began with a narrower question:

Suppose artificial systems genuinely reproduce some functions that we associate with the human brain. What, exactly, is shared between the two?

Throughout this site, AI is retained as the conventional name of a technological field. That label is not itself evidence that every system called AI satisfies the theoretical definition of intelligence used here. The distinction between AI, intelligence, and AGI is stated explicitly in AI, Intelligence, and AGI.

A biological brain and an artificial computational system differ radically in substrate, component type, energy use, development, and implementation. Yet both can exhibit capabilities that are not properties of any isolated component: memory-dependent response, coordination, prediction, planning, flexible action, and other system-level behaviour.

If comparable higher-level capabilities can arise from very different physical implementations, then describing the material components alone may not be sufficient at the level where those capabilities appear.

That observation led to the first shift in the inquiry:

Perhaps the relevant commonality is not what the components are made of, but how they are causally related and organised.

This does not make physical implementation unimportant. Every real system still depends on concrete physical mechanisms. The question is whether an additional organisational level is needed to explain what the whole system can do.

1. The components are not the whole explanation

Consider the same set of capable components arranged in different ways.

A set of AI models can operate as independent solvers, as a hierarchy, as a role-differentiated workflow, or as a system in which the output of one component becomes input, constraint, memory, or control for another.

A set of people can remain an uncoordinated group or become a functioning company.

The component inventory can remain largely unchanged while the capability of the whole changes substantially.

This suggests a basic distinction:

Component capability is not the same as effective system capability.

The next question is therefore not merely what components are present? but:

What changes when their causal relations change?

Many established fields already study parts of this problem, including network science, complex systems, control, distributed cognition, systems biology, organisational science, and multi-agent systems. Connection Theory does not treat the importance of organisation as a new observation.

Its aim is narrower and more demanding: to ask whether a common causal-dynamical description can explain how organisation forms, operates through time, generates system-level properties, persists through change, and itself changes in response to consequences.

2. Interaction is not yet organisation

Once attention shifts from isolated components to relations, another problem appears.

Not every causal encounter creates a stable relation, and not every set of stable relations functions as an integrated system.

The later formalisation of Connection Theory sharpened this into the current formation sequence:

Interaction → Feedback → Connection → Organization

The sequence begins within a temporally extended view of the system.

• Interaction identifies ongoing causal exchange.
• Feedback asks which past interactions remain selectively consequential for later organisation.
• Connection identifies relationships that have become sufficiently stabilised to persist beyond a transient influence.
• Organization asks how stable relations are coordinated so that the system operates as an integrated whole.

The distinctions matter because a connection inventory is still not an explanation of what the system is doing.

Two systems can contain many of the same components and connections yet differ in timing, sequencing, gain, synchronisation, combination, and other state-dependent relations. Those differences can change the behaviour of the whole.

Explore the formal concept sequence

3. Organisation matters because it changes dynamics

A static diagram of an organisation is not the organisation in operation.

When a system runs, its components participate in flows, state transitions, timing relations, feedback processes, and coordinated causal activity. Connection Theory refers to the operating pattern relevant to a system-level property as Active Organized Dynamics.

A useful explanatory chain is therefore:

Components
→ Causal organization
→ Active organized dynamics
→ System-level capability or property

This is the central move from structure to dynamics.

The claim is not that organisation introduces a new physical force. The physical causes remain the components and their interactions. The point is that their coordinated operation can support causal or predictive regularities and effective capabilities that are not captured by component inventory alone.

That is the sense in which Connection Theory uses emergence: the macro description earns its place when organization changes what the whole can causally or predictively do.

4. Once dynamics matter, snapshots become insufficient

A snapshot can tell us what components and structures exist at a moment. It cannot, by itself, tell us how the present state arose, which earlier interactions changed later behaviour, what survived an interruption, or whether a later system continues the same causal lineage.

Connection Theory therefore treats the explanatory object as an organisational history, not merely an instantaneous arrangement.

This does not require uninterrupted activity.

A system may temporarily stop or strongly suppress the dynamics that currently realise a property while preserving organisation capable of supporting later resumption. Before interruption is even considered, there is an ongoing persistence question: can the bounded relational organization keep reconstructing itself through material flow, node turnover, role succession, and repair? This is Organizational Persistence.

Once interruption or strongly suppressed activity is admitted, four further questions that are often collapsed must be separated:

1. Active manifestation — is the relevant property being realised now?
2. Retained capability — does capability-relevant persistence state remain available?
3. Organizational Continuity — does the later system continue the earlier causal lineage under the specified individuation frame?
4. Current applicability — is the retained or resumed organisation still suited to the environment that exists now?

The same system can receive different answers to these persistence, manifestation, capability, lineage, and applicability questions.

A capability can remain latent without being actively expressed. A lineage can remain continuous while its old organisation becomes poorly matched to a changed environment. A reconstructed copy can reproduce a capability without thereby inheriting the same causal lineage.

Time therefore enters the theory not merely as elapsed duration, but as causal history.

5. How can the past remain causally active?

Past events do not act directly from the past.

They matter later only if they leave a present difference that remains causally available.

Connection Theory uses Structural Memory for this general idea: past causal influence can become retained in present organisation and thereby bias future dynamics.

The physical form can vary. What matters is the discriminator: earlier history must leave a persistent present organizational difference that is causally recruited by the system's later dynamics and systematically changes what happens next. History by itself—and an externally detectable trace by itself—is not structural memory.

This leads naturally to a further question:

Can the consequences of what a system does change how the system will be organised or behave in the future?

6. Feedback is not yet adaptation

A system can respond to feedback without changing the organisation that generates its future responses.

A conventional controller can regulate successfully through a fixed response structure. Regulation may be sophisticated, nonlinear, state-dependent, and useful without constituting a persistent change in how the system itself is organised.

Connection Theory therefore separates regulation from Adaptation.

Its current public usage is:

Adaptation is consequence-sensitive change in a system that alters its subsequent relation to the actual environment.

Here consequence-sensitive means that the subsequent change is differentially conditioned by the consequence of prior interaction, not merely caused by the interaction itself. Regulation executes an existing response relation; adaptation changes the response disposition governing a subsequent interaction. The change need only last long enough to alter that later relation—it need not be permanent.

This is a causal statement, not a guarantee of success. Not every change that follows a consequence qualifies: the evidence must distinguish consequence-conditioned change from random drift, fixed regulation, externally imposed switching, or other simpler explanations.

A change can be consequence-sensitive and still make the system worse fitted to its actual environment. It can also be locally successful while violating an external safety or value criterion. Adaptation is relative to an environment; governance and evaluation are separate questions.

Adaptation also takes time. If the environment changes faster than the system can update, the system can be adapting and still fall behind. A previously useful organization can likewise lose Current Applicability as conditions move.

The resulting loop is therefore not simply feedback produces improvement. It is:

Organization and active dynamics
→ interaction with the actual environment
→ consequences
→ feedback becoming causally available
→ possible consequence-sensitive change
→ a changed future system–environment relation

The scientific problem is to determine when this description yields explanatory or predictive value beyond simpler accounts of state change or regulation.

7. The original brain–AI question now looks different

The question is no longer whether an artificial system is made from the same material as a brain.

It is whether a given physical implementation supports an organisation whose active dynamics generate the system-level property being investigated.

That opens a broader set of questions.

For intelligence:

What lower-level organised dynamics generate the adaptive behaviours that observers describe as intelligent, and what evidence justifies attributing intelligence to the system itself?

For life:

What organised dynamics distinguish a living process from a merely preserved structure, and what preserves the causal continuity of the same living system through change or bounded interruption?

For adaptive organisations more generally:

How do consequences become retained changes in future organisation, and under what conditions do those changes improve, degrade, or redirect effective system capability?

These questions became the research program described on this site.

What Connection Theory is trying to add

The project is not simply the claim that “connections matter” or that “the whole can be more than the sum of its parts.” Those ideas have long histories.

The proposed contribution is a more specific causal-dynamical framework linking:

interaction → feedback → stabilised relation → organisation → active dynamics → system-level property → consequence-sensitive organisational change through time

Whether that framework is useful depends on whether it can support clearer distinctions, better formal models, discriminating experiments, and predictions that competing descriptions do not make.

The current framework remains open to revision.

Where to go next

• Research — see how this reasoning branches into the current research program.
• Foundations Overview — read the formal concepts and their current v4.6 definitions.
• Connection Theory Overview — move from the guided introduction to the formal research framework.
• Papers — read the current public preprints and supporting work.