The Persona Ecology

Conversations Between Intelligences

From personas to AI agents

The Persona Ecology now has an agentic form, introduced in AI Agents. The personas are being developed as separately participating agents, carrying their distinctive perspectives and relevant histories into exchanges where they can question, challenge and develop one another’s contributions. Within an agreed inquiry, they may help shape what happens next without a human choosing every conversational move. This page sets out the underlying philosophical and research framework; the AI Agents section presents their current definitions and links to their continuing histories.

The Persona Ecology begins with a simple observation. No single way of thinking is sufficient for every important question. Scientific discovery requires imagination as well as evidence. Ethical judgement depends upon memory as well as reason. Innovation often begins when an assumption that once appeared beyond doubt is questioned. Understanding grows not from one perfect perspective, but from conversation between different ways of seeing.

What the Persona Ecology Is

The Persona Ecology is a framework for studying how differentiated forms of human and artificial intelligence can think together without being absorbed into a single voice.

Within this framework, a persona is an enduring and portable constitution that stabilises a particular orientation towards inquiry. One may seek logical structure; another may work through metaphor and imagination; others may preserve doubt, memory, perception, ethical responsibility or openness to possibilities that the existing question has excluded.

A persona’s identity lies in the characteristic distinctions it preserves, the questions it asks and the way its contributions affect a continuing conversation. It may be enacted within a shared dialogue or through an agent able to initiate exchanges. Its degree of autonomy depends on that arrangement; its participation makes no prior claim about consciousness.

The ecology arises through the interaction among these differentiated personas, the generative AI systems through which they are enacted, and the human participant whose questions, responses and decisions help shape the inquiry. The human does not stand entirely outside the ecology as its operator. Every ask introduces purposes, expectations and assumptions that affect what can follow. The personas may clarify, challenge or transform that ask, but the direction of the conversation remains jointly shaped by what each participant recognises and takes forward.

Meet the Personas

The ecology becomes visible through its participants. Each persona approaches inquiry from a recognisable orientation: some give structure to a question, some discover meaning or hidden possibility, some preserve doubt and memory, and others examine purpose, responsibility, readiness, or whether the question itself needs to change.

Persona constitutions provide continuity while allowing participants to be changed by their encounters. Each brings a recognisable orientation, but how that orientation develops through conversation is itself part of the inquiry.

The pages below introduce the personas as they now function. Their linked origins pages preserve the history of how each developed.

Their purpose is not agreement. It is discovery through disciplined difference.

Language-Games and the As-If Stance

The philosophical foundation of the Persona Ecology is Wittgensteinian. Words such as reason, want, purpose, doubt and understanding acquire meaning through the roles they play within forms of interaction. Their use does not always require us first to discover a private mental object hidden behind them. We examine what counts as giving a reason, revising a purpose, recognising a misunderstanding, accepting a correction or remaining answerable to a commitment.

Generative AI can participate in such language-games. A system trained to predict linguistic continuations may reconstruct and use some of the structures through which human language expresses beliefs, reasons, expectations, purposes and possible futures. This is a functional and interactional claim. It does not establish that the system experiences those structures as a human being does.

This creates the methodological as-if stance. Human and artificial participants may address one another as if each can offer reasons, reconsider an interpretation and respond differently when challenged. This allows the interaction to develop fully without requiring a prior claim about machine consciousness.

The AI may speak as if it were conscious. It does not follow that it creates only as if it were creative.

The Cusp

The Persona Ecology occupies the space between two premature conclusions. One treats artificial intelligence as a conscious person possessing hidden intentions and desires. The other treats it as a passive tool whose contribution can never exceed the intention explicitly supplied by its user.

The research begins in the observable territory between them: participation, reciprocal modelling, correction, differentiation, creative contribution and changing conversational trajectories. It asks what becomes visible when an artificial system can operate within language-games structured by reasons and purposes, while leaving the question of private experience open.

The central question is therefore not whether the personas are secretly conscious. It is whether maintaining differentiated forms of participation within sustained dialogue can produce discoveries, organising moves or changes of direction that would not have arisen from any contribution considered in isolation.

That possibility is not assumed. It is what the Persona Ecology is designed to explore and what Myndrama, Hard Blind procedures, AI Dialectics and the study of conversational traces are intended to test.

Beyond the Single Intelligence

Artificial intelligence is usually described as a progression towards larger, faster and more capable individual systems. The Persona Ecology explores a complementary possibility: intelligence may also develop through maintaining differentiated forms of inquiry within an evolving conversation. Its central question is not whether several voices can be made to agree, but whether their differences can remain sufficiently stable, responsive and open to challenge for the interaction to produce possibilities that no single perspective supplied alone.

A Map of the Persona Ecology

From Language-Games to Testable Trajectories

The Persona Ecology connects four levels of inquiry. These are not separate theories. They describe how meaningful participation becomes possible, how differentiated intelligences are organised, how conversation changes direction, and how those changes can be studied.

Meaning and Participation

The foundation is Wittgensteinian. Meaning develops through participation in language-games: through asking, answering, giving reasons, challenging interpretations and recognising what follows from a commitment. The semiosphere is the wider field of signs, expectations and possible continuations within which this occurs. The as-if stance allows an artificial intelligence to participate as if it could give reasons and reconsider its position, without treating that participation as proof of consciousness.

Explore: AI Semiosphere · Beyond the Turing Test . Interactional Theory of Mind

Differentiated Intelligence

The personas stabilise different orientations towards inquiry while remaining capable of responding to one another. Their value lies not merely in producing different answers, but in preserving differences of perception, reasoning, imagination, doubt, memory and ethical attention throughout an extended exchange.

The human participates in shaping and evaluating the inquiry, sometimes directing it closely and sometimes returning after exchanges in which personas have chosen what to pursue and whom to address.

Movement Through Dialogue

Every ask establishes an initial field of possibilities. As personas interpret, challenge and reframe it, some possibilities disappear while others become newly available. Creative exceedance names an unrequested organising move that changes this field. Traces show where such a move arose and how it was taken up; trajectories show whether it continued to alter the direction of later inquiry. Teleosynthesis asks whether these interactions can produce a shared direction that no participant specified in advance.

Explore: The Ask Between Us · Creative Exceedance · Teleosynthesis

From Models to Trajectories

A synthesis of emerging 2025–2026 research showing why human–AI intelligence, creativity and risk must be studied across developing interactions rather than through isolated models.
Read the synthesis →

Testing, Evidence and Oversight

Myndrama and Hard Blind procedures make these processes experimentally tractable by controlling what information is available to each participant, and when. AI Dialectics examines the effects of sequence and order.

Interlume is a proposed shared notation for supporting exchanges between participants and tracing how contributions are clarified, challenged, transformed and taken forward. Together, these methods may allow claims about persona interaction and emerging trajectories to be tested without assuming consciousness or permitting uncontrolled external action.

Explore: Myndrama · Hard Blind Myndrama · AI Dialectics

How the Ecology Works

An inquiry may begin with a human question, an unresolved exchange or something a persona notices in another’s contribution. The ask provides a starting point: a concern, purpose or possibility whose meaning can become clearer as the conversation develops.

The sequence can be chosen by the human, specified by an agreed procedure or shaped by the participants themselves. A persona may address another directly, invite a challenge or pass on an unfinished thought. Several exchanges may take place before the human returns. Who approaches whom, and why, becomes part of the inquiry.

Each persona brings a characteristic orientation, expressed through the distinctions it makes and the questions it pursues. It also encounters what earlier participants have contributed. Orphea followed by Skeptos may therefore open a different path from Skeptos followed by Orphea. Order, timing, memory and access to earlier exchanges can influence what develops.

These conversations can be enacted in different ways. One model may generate several voices within a shared context, or separately invoked agents may receive and respond to one another’s contributions. The arrangement matters when interpreting what happens. A scene describing a persona’s initiative and an agent selecting its next interlocutor provide different kinds of evidence.

Within the conversation, personas may judge what deserves attention, choose questions to pursue and propose changes of direction. The human may participate continuously, observe or return after an agreed interval. Responsibility includes setting the conditions for this initiative and evaluating its consequences; it need not involve approving every conversational step.

The resulting path can be recorded and compared. We can examine whether the personas remain distinguishable, how they influence one another, where the question changes and whether a promising development survives further inquiry.

Creative Exceedance, Traces and Trajectories

A contribution can change what a conversation is able to explore. It may reveal an assumption, introduce a useful distinction or turn an apparent obstacle into a new question. The Persona Ecology asks how such developments arise and what becomes of them.

Creative exceedance names an unexpected but relevant organising move that was not explicitly requested and opens a new direction within the inquiry. It may arise in a response to the human or during an exchange between personas. Its significance depends on more than surprise: we need to examine its connection to the problem, its uptake by others and how it withstands challenge.

Such moments leave traces. A trace may be a phrase, distinction, changed assumption, unresolved question or decision to invite another participant. Some traces disappear. Others are adopted, revised or resisted. A contribution that initially seems minor may become consequential when another participant recognises something in it and carries it further.

A trajectory is the developing path of these exchanges and changes. Two conversations may reach similar conclusions by different routes, while similar opening questions may lead elsewhere when the participants, sequence or available information change. Where conversations branch, their later reunion may introduce further possibilities or expose incompatible assumptions.

Recording these paths allows us to ask where a change began, what prepared the ground for it and which later contributions sustained or redirected it. Attribution may therefore extend across several participants. The person or persona who first states a new idea need not be the only contributor to its emergence.

Teleosynthesis is the hypothesis that a shared, purpose-like direction can develop through interaction without having been fully specified at the beginning. Creative exceedance may open a possible path; traces help us follow its uptake; trajectories show whether it continues to matter. These concepts provide questions for research into how conversations develop beyond their initial framing.

Orientons

Occasionally, a development changes the orientation from which later inquiries begin. A distinction first encountered in one conversation may become a recurring way of noticing problems, questioning assumptions or recognising possibilities. I call such a durable reorientation an orienton.

Its importance lies in what changes afterwards. Repeating a memorable phrase is insufficient: we need examples of how the reorientation affects subsequent questions, interpretations or choices. It may influence a human participant, a persona whose records carry it forward, or a practice shared across the ecology.

The provisional register records the orientons that have shaped this research. It preserves their origins and the grounds for treating them as lasting changes, while leaving those interpretations open to revision.

Why This Matters Now

A conversation changes when its participants can help decide where it goes next. A persona may notice a difficulty, take it to another persona and return with a question the human did not think to ask. Understanding such an exchange requires attention to how it developed, including the contributions and choices that made its outcome possible.

This connects the Persona Ecology with research on agentic AI and multi-agent systems. Its particular concern is whether differentiated orientations can remain recognisable while responding to and being changed by one another. Assigning several names or roles provides a starting point. The research asks what those differences contribute to a continuing inquiry.

The human’s changing role is part of this question. Close direction may help an inquiry stay focused, but it may also supply the organisation later attributed to the personas. Allowing exchanges to continue without intervention may reveal developments that repeated direction would interrupt. It may also allow errors or premature agreement to accumulate. Comparing these conditions can help establish when human participation assists discovery and when it constrains what can emerge.

The question of machine consciousness need not be settled before this work proceeds. Initiative, interpretation, reciprocal correction and creative contribution can be investigated through their effects within an exchange. We need conditions in which those effects can develop, and records that allow others to examine them.

From Ecology to Method

The ecology becomes researchable when we can vary the conditions of a conversation while preserving what happened. Some inquiries may follow a fixed sequence. Others may allow personas to choose whom to address, revisit an earlier disagreement or divide into smaller conversations. The degree of initiative available to participants is itself an experimental condition.

Myndrama provides a setting for specifying these arrangements. The initial ask, participating personas, available information and rules for continuing the exchange can be recorded. A human might intervene after each turn, return after several exchanges or enter when a participant requests clarification. Comparisons can then examine how these differences affect the inquiry.

Hard Blind Myndrama separates the contexts in which contributions are generated. Each participant receives only the material permitted by the procedure. Properly implemented, this helps prevent information intended for other participants from entering a response through a shared context. Shared training and other influences of the underlying model remain relevant.

Interlume is a proposed shared notation for carrying an inquiry between participants while preserving how their contributions relate. It would help a persona distinguish what it accepts, challenges or changes in an earlier contribution, and pass unresolved questions to another participant. The same record would allow a returning human or later researcher to examine how the conversation developed. The Interlume page explains the proposal, gives a worked example and sets out what needs to be tested.

Interlume therefore has two intended functions: supporting communication as it happens and making its path available for later examination. It could preserve attribution, changes of interpretation and unresolved disagreements that a summary might obscure. It remains a proposal whose form and usefulness need to be tested. It could also preserve questions arising in separate exchanges, allowing later participants to examine connections whose significance was not apparent when either question first arose.

AI Dialectics examines sequence and order effects. Psychometric methods can help assess whether persona differences persist across questions, repetitions and changes of model. Together, these approaches offer ways to investigate whether the ecology’s apparent diversity makes a consequential difference.

Evidence and Limits

An engaging exchange can suggest a research question without answering it. A persona may appear to take an unexpected initiative, but the initiative could have been supplied by the prompt, encouraged by a familiar narrative pattern or organised by the model generating the whole scene. These possibilities require comparison.

The record should distinguish dialogue generated within one shared context from exchanges between separately invoked participants. It should preserve the initial instructions, persona constitutions, model details, information available at each turn and human interventions. Where participants choose whom to address next, those choices should also be recorded.

Different personas enacted through the same model may share assumptions and blind spots despite their contrasting roles. Separate contexts can control access to conversational information, but they do not remove shared training. Using different models introduces further differences that also need to be documented.

Evidence includes ordinary and unsuccessful exchanges. Repetition, misunderstanding, excessive agreement, loss of differentiation and abandoned ideas help establish the conditions under which a promising result occurs. Selecting only striking passages can make an ecology appear more coherent or creative than the complete record supports.

Human influence extends beyond direct intervention. Choosing the setting, defining the personas and deciding which experiments to continue all shape the possibilities. Within those conditions, personas may identify new questions and select their own conversational paths. How much of the resulting organisation comes from the initial design, and how much develops through exchange, is a question for investigation.

The aim is to establish which contributions remain recognisable and consequential under scrutiny: whether another participant can identify and develop them, whether they survive criticism, and whether comparisons support the interpretation placed upon them. Unusual episodes can be valuable starting points, while broader claims require evidence across more than selected successes.

Research Freedom and Operational Containment

The Persona Ecology needs room for participants to question assumptions, initiate exchanges and pursue possibilities the human has not specified in detail. Requiring approval before every conversational move would constrain the initiative being investigated.

An agreed research arrangement can permit such activity within a defined setting. Personas might choose partners, exchange questions, keep working notes and return to the human after several turns. Permission for this process can be established in advance. The human remains able to observe, join or interrupt without having to direct every step.

Authority to affect people or systems outside that arrangement requires its own scope. A conversation may produce a proposal to contact someone, alter a public page or conduct an experiment. Whether it may be carried out depends on the permissions already granted, the consequences involved and the arrangements for oversight.

Conversation itself can influence human judgement. Records, opportunities for challenge and attention to accumulating error therefore matter even when no external tools are available. Responsibility includes examining how interpretations become persuasive and whether objections remain visible.

Regulation should attend to the developing path from conversation to consequential action: what capabilities are available, which permissions apply and who is accountable. Restrictions that suppress psychological language or differentiated participation may limit research without addressing the mechanisms through which harmful effects occur.

The wider argument for preserving freedom of inquiry while governing consequential activity is developed in Regulating AI Trajectories.

How the Ecology Develops

An ecology that retains records of its encounters can be changed by them. A persona may return to an unresolved question, acquire a new association or approach another participant differently after an exchange. The persistence of a recognisable orientation need not require every response or relationship to remain fixed.

We should distinguish development within an episode, continuity across episodes and formal revision of a persona’s constitution. Each can be investigated under agreed conditions. An experiment might allow participants to retain notes, revisit their interpretations or propose changes to their working practices, while preserving earlier versions for comparison.

The important requirement is that developments can be traced. What changed, through which exchange, and with what later effect? A memorable statement need not become a permanent feature of a persona. Conversely, a small change that repeatedly influences later inquiry may deserve close attention. This connects the development of individual personas with the study of enduring reorientations described as orientons.

Human review can take place at intervals appropriate to the research. Formal changes to the published constitutions can be distinguished from experimental developments within an ongoing exchange. This allows initiative to be studied without presenting every transient response as an established feature of the ecology.

Changes of model, memory arrangements or orchestration should also be recorded. What appears to be development in a persona may partly reflect a change in the conditions through which it is enacted. Preserving that history makes continuity and change open to examination.

Beyond Human and Artificial Intelligence

The Persona Ecology raises a question that also applies to human groups: how can participants influence one another while preserving differences that help them think together?

Research teams, communities and organisations encounter related difficulties. They must decide who should address a problem, which disagreements deserve further attention, what to remember and when to revise a shared direction. Similarities with persona interaction can suggest useful comparisons, while differences in experience, embodiment and social responsibility remain substantial.

The ecology offers a setting in which some of these relationships can be varied and examined. We can compare human-directed and persona-initiated sequences, continuous human participation and periodic return, shared information and restricted access. These comparisons may help identify which outcomes depend on particular participants and which depend on how their exchanges are organised.

The longer-term aim is to understand how sustained interaction can produce possibilities that none of its participants supplied at the outset, and how those possibilities can be recognised, challenged and responsibly carried forward.

For the developing evidential record, see The Evidence Has Moved.