Beyond Dennett

The error of architectural scope

Daniel Dennett was among the most influential thinkers to insist that consciousness be explained without a Cartesian inner witness. He exposed the homunculus fallacy and replaced an all-seeing inner subject with parallel, competing processes. The previous chapter borrowed the title of his most famous book deliberately.¹ This chapter makes the lineage explicit by isolating the point where, on MST’s interpretation, his architecture leaves functional room for an additional subsystem. Calling that subsystem an experiencer is precisely what remains to be defended.

My claim is that his demolition was correct while a later inference was too quick: limitations of introspective report do not by themselves show that no other process supports experience. MST redescribes some reporting functions as the Controller and asks whether the architecture requires another process beneath it. This is a reconstruction of Dennett through MST’s vocabulary, not an identification Dennett accepted.

The Narrator He Found

Dennett’s multiple-drafts model is recast by MST as a functional analogue of the Controller — the role that, in the Modeler-Schema architecture, selects and executes actions, interfaces with language, constructs and maintains the autobiographical narrative, and generates the reports that scientists elicit in experiments. When a subject says “I saw X,” “I felt Y,” or “it seemed like Z,” MST assigns the report to this role. That mapping is the theory’s reconstruction, not Dennett’s own decomposition or an established neural localization.

What the Controller does not have is direct access to the full sensory field, the continuous world model, or the representational substrate in which perceptual comparisons occur. It receives only a thin, focal, target-bound feed annotated with descriptors and valence. Its view of the mind’s internal activity is narrow and heavily preprocessed. Its narration is a log file written over deeper non-verbal cognition, not a transcript of it — the point developed in the geometry of inner speech.

Dennett mapped severe limitations of introspective report. On MST’s reading, he then universalized them: the theory charges him with treating the reporting system’s inability to reveal an inner theater as evidence that no other process could support experience. This is MST’s proposed hinge in illusionism:

The narrator is blind → therefore the mind is blind.

If the reporting and experiencing roles are distinct, that inference is an error of architectural scope. A narrator, on the architecture MST proposes, is never the experiencer — but the limits on what a reporting process can access do not by themselves settle what the whole system undergoes. This establishes logical room for MST’s separation; it does not yet establish that the proposed Modeler-schema occupies the phenomenal role.

What the Experiencer Must Have

MST hypothesizes that, if phenomenality is realized by a specialized comparison process, its mechanism should possess informational privileges the assigned Controller lacks. In particular, its candidate experiencing subsystem is expected to have:

1. Access to the full Concrete World Model, including the non-focal periphery.
2. Direct access to pre- and post-saccadic sensory states, required for maintaining visual continuity across eye movements.
3. Access to internally generated content — recall, imagination, counterfactuals — without linguistic mediation or bottlenecking through symbolic tokens.
4. A dedicated comparison process capable of detecting mismatches and issuing bottom-up refinement signals to the rest of the system.
5. A stable, analog-like representational substrate in which these comparisons are realized as continuous patterns rather than discrete symbols.

The Controller is defined within MST as lacking these channels. That makes the contrast internally sharp, but it does not demonstrate that biological report is implemented by one component with exactly these limits. The more general point survives: a functional mind can contain processes with different informational privileges and representational formats, so evidence about report should not be universalized without an architectural argument.

The Modeler-schema is the subsystem MST assigns the relevant access and operations. In the theory, it receives the full state of the Concrete World Model, monitors sensory-level data used for pre- and post-saccadic consistency checking, and maintains a proposed Quale World Model — an internal, non-symbolic representation not directly available to the Controller. If that decomposition is right, inspecting report alone would be like inspecting a flight console to understand aerodynamics: informative about outputs, incomplete about the process producing them.

Explaining His Data

Dennett’s case for illusionism draws strength from an empirical fact: humans misreport aspects of their own experience. We misjudge color saturation, timing, vividness, agency, peripheral content, and the presence or absence of detail during saccades. MST offers one way to explain these failures without eliminating experience.

On this architecture, reports are unreliable because the Controller is not identical to the hypothesized experiencing process. It receives target-bound metadata and evaluative tags rather than direct access to the proposed Quale World Model, then retrofits narrative coherence where gaps appear. If those channel assignments are correct, unreliable introspection is what one should expect: the reporting role infers the structure of phenomenality through a narrow, lossy path.

Unreliable report does not entail absent experience unless further premises connect the two. MST proposes that they do not share one data channel and can therefore accommodate the reporting failures. This blocks a quick inference from introspective error to no phenomenality; it does not, by itself, falsify every illusionist explanation of the same data.

The Function He Couldn’t Find

Dennett famously challenged proponents of qualia to explain why consciousness exists at all. Everything we attribute to subjective experience — attention, learning, decision-making, memory consolidation — can, he claimed, in principle be handled by unconscious machinery. He looked for the function of consciousness in the Controller and, unsurprisingly, found none.

MST directs the search to the Modeler-schema. The visual world is sampled across discontinuities, while we experience a comparatively stable scene. Maintaining perceptual constancy recruits multiple mechanisms. MST’s distinctive proposal is that a retained comparison format helps detect mismatches between pre- and post-saccadic content and drive bottom-up refinement of the Modeler’s assumptions about the world.

The theory identifies that substrate with qualia: an analog-like, high-dimensional field on which the Modeler-schema performs coherence operations. It contrasts this proposed format with the more discrete, symbolic tokens assigned to the Controller. On this view, qualia are not decorative; they are the medium of representational stabilization and error correction. The functional need for comparison is plausible. The further claim that the comparison format is experiential, and that no nonphenomenal architecture could perform equivalent work, remains the identity claim to defend.

MST’s diagnosis is therefore that Dennett searched for a function in the reporting role while the candidate function belongs to the role that builds and repairs the world model. Whether that candidate function is consciousness rather than a correlate of it remains open.

The Self He Got Right

If the first half of this chapter takes something away from Dennett, the second half gives something back. His account of the self — a “center of narrative gravity,” a useful abstraction rather than a thing — was correct, and it deserves a stronger defense than he gave it.

The intuitive picture is that each of us is a perfectly unified agent: an independent self that stands behind decisions and steers a life. Neuroscience, psychology, and philosophy complicate that picture. The narrated self integrates perception, impulse, deliberation, and habit into a relatively coherent story, but no discrete entity stands outside causality pulling levers. Brain, body, development, environment, and past experience collectively shape choice. That does not make agency unreal; it locates agency at the organized, higher-level process rather than in an uncaused inner essence.

So far, this is Dennett’s position, and the Modeler-Schema architecture vindicates it: the narrated self is the Controller’s product, and the Controller confabulates. But here is the essential twist that “illusion” talk obscures: the narrator may be constructed without agency being illusory. Agency belongs to the organized model–evaluation–control process, not to a homunculus standing behind it.

First, agency shapes real outcomes: an embedded system models alternatives, evaluates them, and makes action depend on that evaluation. The narrative of self is one higher-level model within that loop, and it can influence thoughts, emotions, decisions, and actions without being a separate cause outside the machinery. Second, agency structures social interaction. Human civilization is built on recognizing each other as agents capable of choice — accountability, justice, and cooperation all rely on it, while social recognition feeds back into self-interpretation. Third, reflective authorship grounds this book’s ethics and much of its account of meaning. The normative force comes from stated commitments to authorship, not from the mere existence of any control loop.

The confusion arises from conflating two claims that must be kept apart. Agency is not an independent force added to physics; decisions are implemented by causal chains. Yet the organized control loop is itself causally efficacious, the way a program can be real without existing apart from the hardware that runs it. The narrated self is a model the loop constructs and revises. It can be inaccurate, fragmented, or strategically useful without exhausting the agency beneath it. Denying an uncaused inner essence therefore leaves both basal control and the stronger question of reflective authorship intact.

Holding the distinction clearly yields real dividends. Responsibility can attend to capacity, reasons, control, and future behavior rather than judging an intrinsic moral essence. Personal identity becomes dynamic, a continually revised interpretation rather than a static substance. And the free-will debate loses its metaphysical urgency, shifting to practical questions of effective choice, accountability, and psychological autonomy. Far from undermining human dignity, understanding the self’s interpretive basis clarifies what selfhood is: a real model within a real agent, not a ghost at the controls.

Completing the Picture

Dennett was right to demolish the false picture of an inner spectator. There is no little person in the head watching an inner movie. MST proposes, however, a mechanistic experiencer: a cybernetic regulator that maintains representational coherence through comparison states identified by the theory with qualia. It is not a spectator but a candidate comparator. Its regulatory function can be investigated separately from the contested phenomenal identity.

Dennett’s multiple drafts, his narratives, his center of narrative gravity — these remain intact. In MST, the Controller stitches stories from partial data; the autobiographical self remains a pragmatic construction worth keeping. What changes is the proposed layer beneath: those drafts may emerge from a process whose comparison states the theory identifies with experience. The Controller narrates what happened in the model; MST assigns the candidate implementation of experience elsewhere.

The architectural claim is now clear: the subsystem that reports need not be identical to the subsystem MST assigns experience. The proposed experiencer is mechanistic and functionally motivated; whether it is phenomenally sufficient remains open. The claim that a self-model can genuinely be a subject, along with the objections it attracts, is where the argument goes next, in Mirrors of the Mind.