The Aperture Framework

In one sentence #

Before asking “Is this conscious?”, the framework asks:

What can this system represent about itself, how does that affect its activity, and what limits it?

Why this framework exists #

A yes/no question about consciousness can hide differences between humans, animals, AI models, organisations, states of meditation, and social interactions. We first need to describe the system and the situation we are asking about.

This framework starts with observable or otherwise assessable features:

• what it can sense or receive;
• how it integrates information;
• whether it models itself;
• whether that self-model persists over time;
• whether it can model counterfactual futures;
• whether anything matters to it;
• how it maintains a boundary;
• how it changes through relation.

These comparisons leave the hard problem open: why would any of these features involve experience? They provide a way to describe systems while that question remains unresolved.

Core terms #

Reflexive system

A reflexive system uses information about its own states or operations to guide what it does next. Mere feedback is insufficient. A thermostat responds to temperature without modelling its own activity of measuring temperature.

Aperture

An aperture means an opening. Here it is a metaphor for the ways a system encounters and responds to its environment.

An aperture describes a system’s inputs, memory, self-model, priorities, boundaries, and capacity to change through interaction.

The word describes an arrangement of capacities. It does not refer to a soul or a separate observer inside the system.

Fold

A fold is the primitive structure of self-reference: a system takes relations, including its own operations, as content.

The distinction from Space Immanence is:

“Fold” names the proposed structure of self-reference. “Aperture” describes how a particular system senses, represents, and acts.

What the framework does not claim #

The Aperture Framework does not claim:

1. that all apertures are conscious;
2. that current AI systems are conscious;
3. that consciousness is everywhere;
4. that the material a system is made of does not matter;
5. that self-reference alone proves experience;
6. that contemplative insight settles metaphysics;
7. that a scoring rubric can replace lived experience.

Describe the system first. Then ask what evidence, if any, gives reason to think it has experience.

The eight dimensions of an aperture #

These eight dimensions are proposed categories for comparison. Their usefulness and measurement remain to be assessed.

1. Modelling self and environment

Question: Does the system model itself in relation to an environment?

A reactive system may track inputs without distinguishing itself from its environment. Animals, humans, and potentially artificial agents can represent their own states, capacities, and vulnerabilities in relation to that environment.

Why it matters: The framework requires a self-world distinction for its stronger sense of aperture.

2. Combining information across the system

Question: Are signals integrated across the system in a way that constrains whole-system behavior?

This asks how information from different parts of a system affects its overall activity.

Why it matters: Separate parts can process information well without sharing information that guides the system as a whole.

3. Modelling one’s own modelling (recursive depth)

Question: Can the system model its own modelling?

The proposed levels are modelling the world, modelling one’s own modelling, and revising how that modelling works.

Why it matters: Representing a process can give a system information with which to assess or revise that process.

4. Memory and continuity (temporal thickness)

Question: How do memory and anticipation affect its current activity?

Temporal thickness describes how past states, possible futures, and present activity form a continuing pattern. Storing memories alone does not establish this continuity.

Why it matters: A single moment of feedback is not a self. Temporal continuity increases as memory becomes part of what the system is doing now.

5. Representing alternatives (counterfactual range)

Question: Can the system represent alternatives to what is happening?

Counterfactual range covers possible futures, alternative actions, and possible changes to the system itself.

Why it matters: Consciousness-relevant systems represent possibilities as well as what is happening now.

6. What matters to the system (valence and stakes)

Question: Does anything matter to the system?

Valence concerns positive or negative significance. In organisms it can involve hunger, pain, attraction, fatigue, fear, care, or the need to maintain bodily conditions. Proposed analogues in artificial systems include cost, risk, reward, conflicting goals, reputation, energy budgets, and self-maintenance.

Why it matters: Without stakes, information remains inert. Positive and negative significance connect perception with meaning.

7. Maintaining a boundary

Question: Does the system maintain a distinction between itself and what is not itself?

A boundary can depend on metabolism, cognition, computation, social membership, or a system’s organisation; it need not be a skin.

Why it matters: A system that cannot maintain itself as a system cannot sustain an aperture.

8. Changing through interaction (relational permeability)

Question: How can interaction change the system?

Language, touch, teaching, trauma, love, design, conversation, and environment can change how a system responds and understands.

Why it matters: Many major changes in consciousness arise through relationships, which can expand, limit, clarify, or disrupt a person’s experience.

A simple scoring rubric #

The scores describe proposed levels on each dimension. They are a comparison aid, with no validated interpretation as a measure of consciousness.

A profile can help compare systems. No score proves that a system has experience.

Examples #

The examples illustrate proposed profiles and their limits.

Questions about AI consciousness #

The question about consciousness remains:

Is the AI conscious?

To assess it, first describe:

What can this system represent, remember, and change about its own activity?

Current AI systems can influence what a user notices, understands, or finds meaningful. In this framework, that gives them epistemic and relational roles. Those roles do not imply that the systems have experience.

An artificial system is a candidate phenomenal aperture only if it meets all four of the working paper’s fold criteria in operationally robust ways:

• integrated self-world modelling;
• temporal continuity across perturbation;
• counterfactual self-modelling;
• self-maintaining boundary or operational integrity;

This companion framework adds three dimensions for further comparison once those criteria are met:

• access to and assessment of its own modelling;
• needs or priorities that belong to the system rather than being attributed by an observer;
• interaction that changes how it organises itself in future.

The four criteria are the canonical filter in the working paper. The additional aperture dimensions elaborate them; they do not replace them. Meeting these conditions would not prove consciousness. It would justify serious, cautious scrutiny.

Possible effects of contemplative practice #

The framework proposes that different contemplative practices may change different features of experience:

• attention practices may help coordinate information across the system;
• self-inquiry may increase awareness of one’s own thinking;
• ethical practice may change what feels significant or valuable;
• embodiment practice may refine the perception of internal bodily signals;
• devotion may change how relationships affect the practitioner;
• nondual practice may change how people distinguish what they experience from the activity of experiencing.

Research question 1 below would test these proposed relationships; none is established. The general suggestion is that practice changes how people recognise the activity shaping their experience.

Implications for design #

The AI design principles asks how technology can help users understand their needs and choices. It contrasts that goal with optimising a product’s ability to predict and influence behaviour.

Relationship to Space Immanence #

Space Immanence proposes that consciousness and spacetime may be two aspects of self-reference. These descriptive comparisons do not require that metaphysical claim to be true. The two projects ask related questions:

Their different tasks are:

Space Immanence proposes an account of self-reference. The Aperture Framework compares particular systems and their capacities.

Limits of the comparison #

The epistemic and relational categories concern knowledge and interaction. Only the phenomenal category makes a claim about experience.

Keeping those questions separate allows the following possibilities:

• AI can support knowledge or interaction without being conscious.
• Human–AI interaction can produce meaning for a person without requiring AI experience.
• Animals can have complex capacities without human-like reflection.
• Contemplative practice can change experience without establishing a metaphysical account.
• Technology can support users’ understanding and choices while the science of consciousness remains unsettled.

A system’s capacity to help someone learn, change, or understand does not by itself establish that the system has experience.

Research questions #

1. Do contemplative traditions alter aperture dimensions in systematically different ways?
2. Can greater awareness of one’s own thinking be measured through self-modelling, changes in perceived boundaries, or the weight assigned to predictions and errors?
3. Which aperture dimensions distinguish ordinary cognition, flow, meditation, trauma, depersonalization, and psychedelic states?
4. Which AI architectures support meaningful interaction, and which might eventually warrant examination for possible experience?
5. Do technologies designed to support users’ understanding and choices improve those outcomes compared with systems designed to predict and influence behaviour?

Closing #

The framework offers descriptions to test and improve. It leaves the existence and explanation of experience open.

A useful comparison specifies the system, its capacities, and the conditions under which they appear.

A claim about consciousness requires an additional argument and evidence.