Frame

In plain language: The word is plain. Frame is what the Agent thinks its Boundaries are.

Definition

An Agent’s Frame is the Agent’s self-model of its own Boundaries — what the Agent thinks it can perceive, what it thinks it can model, and what it thinks it can affect. Frame is a single object per Agent: one self-model that covers all three Boundaries. Its accuracy varies — an Agent can have an accurate self-model of his Causal reach and a wrong self-model of his Cognitive depth, or an accurate self-model of his Intrinsic capabilities and a wrong self-model of his Positional ones. These are not separate Frames; they are regions of one self-model with varying accuracy.

Frame is itself a Cognitive function. The Cognitive Boundary models the world; when it models itself, the output is Frame. There is no separate self-modeling process — Frame is what the Cognitive Boundary produces when its target is itself rather than the world. This means the Cognitive Boundary is the only Boundary that models itself, which is one reason Cognitive blind spots are particularly hard to detect from the inside: the instrument that would detect the blind spot is the same instrument that has the blind spot.

Frame can be wrong in either direction:

• Frame > Boundary (the hubris case) — the Agent’s self-model overstates his reach. He thinks he understands the regulatory landscape when he doesn’t. He thinks he can manage the political dynamics of a board when his position carries no mechanism to do so. This is the structural source of the Franz Ferdinand Effect: the Agent acts on a self-model that exceeds his actual Boundary, producing consequences he cannot predict because his Cognitive Boundary doesn’t extend to where his Frame told him it did. The intervention depends on the source of the mismatch. If the Agent has the Cognitive capacity to model the domain but doesn’t realize his self-model is wrong, evidence can close the gap — a failure, a pushback, an audit that exposes the overestimate. If the Agent lacks the Cognitive capacity itself, evidence alone won’t help; the intervention must address the Cognitive Boundary first (training, education, new Cognitive Tools) before the Frame can update.

• Frame < Boundary (the ignorance case) — the Agent’s self-model understates his reach. He doesn’t know the CRM has a predictive-scoring feature. He doesn’t realize his budget decisions structurally shape the company’s talent pipeline for the next two years. This is the structural source of under-realization: Capacity at hand goes unused because the Agent’s Frame doesn’t include it. The intervention for Frame < Boundary is awareness — making the Agent aware of Capacity already provisioned.

Frame is not a Boundary. Boundaries are the three (Computational, Cognitive, Causal). Frame is the self-model of those Boundaries. It lives inside the Cognitive Boundary but is about all three. It can match the true Boundary, fall short of it, or exceed it — but it is not itself a Boundary in the framework’s defined sense.

At any analytical moment, Frame is fixed alongside the Agent’s Boundaries, Capacity, and Realization. Over time, Frame can change: Computational input combined with Cognitive modeling may produce a Frame revision, updating the Agent’s self-model without any change to the underlying Boundaries or the world. The productivity of this feedback loop is Agent-dependent — some Agents update Frame readily; some are calcified. The loop can run in either direction: an upward spiral where the Agent discovers underused Capacity, uses it, finds more, and expands Frame further; or a downward spiral where bad evidence contracts Frame, which contracts Realization, which produces less evidence and further contracts Frame.

The malleability of Frame makes it the intervention that requires changing the least: nothing about the Agent’s Boundaries needs to change, nothing about the world needs to change — only the Agent’s self-model. Change Frame alone and Realization shifts. This is, as a structural matter, what coaching, the alter-ego effect, onboarding interventions, and mindset-reset techniques exploit — independent of whether any given practitioner is effective. The structural claim is that the intervention point exists, not that any specific technique works.

What Frame is not:

• Not divided into sub-Frames. An Agent has one Frame. When a Composition contains sub-Agents (a person’s liver is itself an Agent at a different scale), each sub-Agent has its own Frame — but those are separate Agents’ Frames, not sub-Frames of the containing Agent. Per Turtles All The Way Down, every Agent at every scale has one Frame; the principle applies recursively.
• Not the Boundary itself. Frame can exceed the true Boundary (the hubris case — Frame > Boundary) or fall short of it (the ignorance case — Frame < Boundary). Both are real; neither is pathological. The gap between Frame and Boundary is where the framework’s diagnostic value lives.
• Not what the Agent is currently attending to. The same Agent’s Frame covers both the work region and the home region of his Boundary (the pinched ellipse — the framework’s image of how a single Boundary has distinct regions, like two lobes of a peanut, that coexist within one Agent). Attention shifts between regions; Frame includes both. An a-ha moment about a work problem while playing with his child at home is Frame updating in the work region while attention is on the home region — both regions always loaded in the same Frame.

Relations

Frame gates Realization. Tools on hand extend the relevant Boundary at Capacity whether or not the Agent is aware of them; Frame determines which Tools the Agent actually uses. A Tool outside the Agent’s Frame extends Capacity but not Realization — the Tool is there, but the Agent cannot reach it until the Tool enters the Frame (through training, disclosure, a colleague’s mention). This is what makes the gap between Capacity and Realization diagnosable: when Realization falls short of Capacity, the first question is whether the gap is a Frame problem (the Agent doesn’t know) or a Cognitive problem (the Agent knows but can’t model how to use what he has).

Frame is itself subject to the constraints of the Cognitive Boundary. The thing that gates Realization (Frame) is itself a Cognitive function, which means the shape of an Agent’s Cognitive Boundary constrains the shape of his Frame. An Agent whose Cognitive Boundary is deep in one domain and shallow in another will have a Frame that reflects those contours — confident and accurate where his modeling is strong, overconfident or unaware where his modeling is weak. The constraint is not uniform; it follows the Cognitive Boundary’s own shape.

Capacity gates Frame as a ceiling: an Agent whose Frame exceeds his actual Capacity has a Frame > Boundary mismatch. Frame gates Realization downward: an Agent whose Frame is narrower than his actual Capacity has under-realization. The relationship is bidirectional — Capacity sets the upper bound of what is real; Frame determines how much of that real Capacity gets used.

The Finite Agent Limit and the Infinite Agent relate to Frame through a bell curve across the complexity spectrum. At the simple end (a thermostat, a very basic agent), Frame approaches Boundary because the system is simple enough to fully model itself — the gap is near zero, and the framework has little to diagnose. At the Infinite Agent end, Frame = Boundary is guaranteed because modeling capacity is unbounded — the unique case where the gap is structurally impossible. In the wide middle — where every Agent the framework is designed for lives — the gap is real and significant, and this is where the framework’s diagnostic tools do their work.

The Franz Ferdinand Effect has its structural source in Frame > Boundary. Trapped Intelligence at the Realization layer has its structural source in Frame < Boundary — Capacity at hand that the Agent doesn’t know about.

Example — CEO

A VP of operations has been in his role for three years. His Frame — his self-model of his own Boundaries — is wrong in both directions simultaneously.

On the Cognitive side, his Frame overstates his reach. He believes he understands the regulatory environment well enough to evaluate compliance risk in his department’s processes. He does not. His Cognitive Boundary doesn’t extend to regulatory analysis — he lacks the training, the pattern recognition, and the domain vocabulary. But his Frame says he has it, so he makes compliance-related decisions without consulting the legal team, and those decisions produce consequences he cannot model. This is Frame > Boundary: the Franz Ferdinand Effect, powered by a self-model that claims territory the Cognitive Boundary doesn’t actually cover. The intervention is evidence — a compliance audit that exposes the gap, or a mentor who pushes back on his confidence in the domain.

On the Causal side, his Frame understates his reach. His Positional Causal Boundary includes the structural ability to reshape the company’s talent pipeline for the next two years — his budget and headcount decisions determine which roles get filled, which teams grow, and which capabilities the company develops. He doesn’t realize this. His Frame about his Causal reach is narrower than the actual Positional Boundary: he thinks his decisions affect his department’s quarterly performance, not the company’s multi-year trajectory. This is Frame < Boundary: under-realization at the Positional Causal layer, with the gap diagnosable as a Frame problem. The intervention is awareness — a board member or a coach who helps him see the structural consequences of decisions he’s already making.

One Frame. Two regions wrong in opposite directions. Neither mismatch is a character flaw — both are structural consequences of a finite Agent whose Cognitive Boundary cannot fully model itself.

On a Saturday afternoon, the VP is building a treehouse with his daughter. He is not thinking about work. But his Frame about work hasn’t disappeared — both the work lobe and the home lobe of one Boundary (the pinched ellipse) are always loaded. Halfway through cutting a board, he realizes that the compliance issue he dismissed last Tuesday is actually a version of a problem he saw at his previous company. His Frame about his Cognitive reach on regulatory risk just updated — not because he was at work, not because someone pushed back, but because his Cognitive Boundary’s self-modeling produced a revision while his attention was elsewhere. The feedback loop runs regardless of which region attention engages.

Example — Research

A regulatory T cell in the immune system has a Frame — the cell’s regulatory programs operate on an implicit model of what the cell’s inputs mean and what the cell should do in response. The regulatory programs are not reflective; the cell does not “think about” its own Boundaries. But the structural pattern is the same: the cell has a self-model (regulatory state), the self-model can be wrong, and the mismatch between self-model and actual Boundary is where the diagnostic value lives.

Frame > Boundary: autoimmunity. The cell’s regulatory programs classify a self-marker as a threat. The cell’s Cognitive Boundary — its signaling cascades and pattern-recognition machinery — hasn’t changed. What’s wrong is the cell’s implicit model of what its inputs mean: its regulatory state says “attack” about a molecule that is part of the cell’s own tissue. The cell acts on a self-model that overstates what it should respond to, and the Causal output (attacking self-tissue) produces damage the cell’s modeling cannot predict or prevent. Same structure as the VP who thinks he understands regulatory risk: the modeling apparatus is intact; the self-model is wrong about what the apparatus should be doing.

Frame < Boundary: pathogen tolerance. The same cell type, after chronic exposure to a genuine pathogen, stops responding to it. The capacity to attack is still there — the receptor machinery is unchanged, the effector pathways are intact. But the cell’s regulatory state has reclassified this signal as “not worth responding to.” The cell’s Frame has contracted below the actual Boundary: Capacity at hand, unused, because the self-model no longer includes it. The pathogen persists because the cell has the reach to eliminate it and doesn’t use it.

The diagnostic leverage point lands at the biological scale: tolerance can be broken by an adjuvant or a new inflammatory signal — an external intervention that resets the cell’s regulatory state without changing any of its underlying machinery. The cell starts attacking the pathogen it was ignoring, using capacity that was there the entire time. This is the biological equivalent of the coaching intervention: change Frame, nothing else changes, Realization shifts. The same structural pattern operates without reflective cognition — the feedback loop between regulatory state and Causal output runs in the cell exactly as it runs in the executive, and the intervention has the same structural shape.