Schema Mask Theory — Part III
The Phenomenal Interface: What the User Actually Sees
Abstract
Part II of Schema Mask Theory used computer interfaces as a deliberately simplified model for examining communication across differentiated information systems. It showed that apparently simple access can require handshaking, capability discovery, authentication, authorisation, health checking, representation, error detection and provenance. The present article takes the next small step. It asks what happens when the receiving participant does not encounter those underlying processes directly, but only the interface-generated world through which their results become intelligible.
This is already a familiar problem in human-computer interaction. A user may believe that they clicked Send even though no actionable input was registered. They may interpret a highlighted button as confirmation that an operation occurred, mistake an incomplete progress display for successful completion, or attribute a failure to one component when it actually occurred elsewhere. Human-factors guidance consequently treats visibility of system status, meaningful feedback, error recognition and recoverability as fundamental properties of good interface design. Recent research on generative interfaces extends the problem: users can have difficulty determining what a system is capable of, how to instruct it, and whether the resulting output actually satisfies their original intention.
The recently developed article Ghosts as Phenomenal Projections suggests an organismic parallel. It proposes that conscious experience may not simply receive reports about an independently completed world. Multiple specialised domains may contribute differently weighted information, expectations, memories, symbols and interpretations to what becomes the person’s phenomenal field. The apparent unity of the experienced world may therefore represent an achievement of coordination rather than evidence of a single underlying observer.
This paper does not yet attempt to explain who governs those exchanges or how developmental access is authorised. Nor does it develop the proposed gamification of the Human Symbolic Interface in detail. Those questions require separate treatment. Instead, the present article concentrates on a more elementary interface problem:
What does a receiving domain actually experience, and how accurately can it infer from that experience what happened behind the interface?
This question links computer usability, perceptual construction, symbolic representation and Schema Mask Theory. It also suggests that some apparent psychological errors may eventually prove to be interface errors: not failures of the underlying function, but failures of feedback, representation, source identification or the mental model constructed by the recipient.
1. From Access to Experience
Part II began with the assumption that two differentiated systems somehow need to communicate.
That assumption quickly became complicated.
A resource may exist without being available. A process may be alive without being ready. A requester may be authenticated without being authorised. Information may be transmitted accurately but represented poorly. A stable outward interface may conceal the failure of a process behind it.
Those distinctions concerned the underlying relationship.
There is another problem.
The user normally does not see that relationship.
They see an interface.
This produces a second level of uncertainty:
What does the interface lead the user to believe has happened?
That is not always the same as what actually happened.
This difference is so ordinary that it can disappear from view.
A person writes an email and believes they have clicked Send.
The message remains in the editor.
What happened?
From the computer’s perspective, perhaps no send instruction was ever received.
From the user’s perspective:
I sent the message.
Both descriptions accurately describe different levels of the event.
One records machine state.
The other records the user’s phenomenal interaction with the system.
The bug cannot be properly understood until both are considered.
2. “I Clicked Send”
The example is useful because the statement:
“I clicked Send.”
can refer to several distinguishable events.
The person may have intended to send.
They may have moved the pointer towards the control.
They may have seen the control visually respond.
An input event may or may not have been registered.
The application may have accepted the event.
The send process may have begun.
A network request may have been transmitted.
A server may have received it.
The message may have been committed to storage.
An acknowledgement may have returned.
The interface may finally have displayed confirmation.
A conventional user interface compresses this entire chain into something like:
Sent.
That compression is desirable when everything works.
When something goes wrong, it becomes the source of confusion.
Human-factors guidance has long treated system-status visibility and timely feedback as central design requirements. NIST’s usability guidance specifically identifies feedback errors, data-availability errors and interpretation errors as distinct classes of problems, while recommending that users be kept informed about what the system is doing and be helped to recognise and recover from errors.
The interface therefore does more than make a function convenient.
It helps the user construct a model of what has just happened.
3. The Interface as an Epistemic Device
This suggests that an interface has an epistemic function.
It tells its user something about an otherwise inaccessible system.
A green tick may mean:
operation completed.
A spinner may mean:
the system is still working.
A greyed control may mean:
this operation is not currently available.
A warning may mean:
the system predicts that the proposed action could have unwanted consequences.
These are not the operations themselves.
They are representations of system state.
Good interface design therefore depends upon an important correspondence:
the phenomenal state presented to the user should track the relevant underlying system state sufficiently well for the user to act intelligently.
Perfect correspondence is neither possible nor desirable.
The user does not need to see millions of processor states merely to know whether a document was saved.
The problem is one of useful compression.
Enough must be hidden to make the system usable.
Enough must remain visible to make the system intelligible.
4. The User’s Model Can Be Wrong While the Computer Works Correctly
This gives us a class of apparent bugs in which the underlying software may be behaving exactly as designed.
The user’s model is wrong.
They believe a button performs one function when it performs another.
They believe selecting something completed an operation when a further confirmation was required.
They believe a process has stopped because the interface has become visually quiet.
They believe data has been destroyed when it has merely been hidden.
From the user’s perspective, the system is inconsistent.
From the developer’s perspective, the user “used it incorrectly.”
Neither description is sufficient.
The failure can exist in the relationship between the system’s behaviour and the model its interface encouraged the user to construct.
This is why modern HCI does not treat user misunderstanding simply as user incompetence. Research on LLM interfaces, for example, has identified difficulties in knowing whether a system possesses a relevant capability, knowing how to translate a goal into instructions, and knowing whether the result actually fulfils the intended goal. Subramonyam and colleagues describe these as parts of a wider “gulf of envisioning.”
The interface designer therefore has to reason not only about what the system can do, but about what the recipient can reasonably infer that it can do.
5. The Interface Must Be Learned
Many interface conventions become so familiar that experienced users stop noticing that they were ever learned.
Buttons.
Menus.
Icons.
Dragging.
Scrolling.
A disabled control.
An ellipsis suggesting further options.
A red notification badge.
A progress indicator.
The novice does not necessarily understand any of these.
NIST human-factors guidance accordingly recommends making relevant objects, actions and options visible, favouring recognition over recall, and allowing different levels of interaction for novice and expert users.
An interface therefore contains a kind of implicit language.
Competent use requires learning that language.
That observation becomes important when we later consider the Human Symbolic Interface.
Before asking what a symbol means, we may first have to ask whether the recipient has learned how that kind of symbolic interface communicates at all.
6. The Interface Can Be Correct and Still Be Misread
Suppose an application changes the colour of a button after an action.
The designer intends:
button selected.
The user interprets:
operation completed.
Nothing is technically broken.
The interface has nevertheless produced an incorrect phenomenal inference.
This means that interface integrity cannot be judged purely from the sender’s intention.
It has to include the receiver’s interpretation.
That produces a more relational model:
underlying state
→ interface representation
→ recipient perception
→ recipient interpretation
→ next action.
An error at the last two stages can be just as consequential as an error at the first two.
The interface therefore participates in a feedback loop.
It alters what the user does next.
7. The Phenomenal Interface
This provides the bridge to the latest theoretical work.
Ghosts as Phenomenal Projections argues that perception may be better understood as constructive rather than as a passive recording followed by psychological commentary. It proposes that expectation, attention, memory, emotional state and symbolic interpretation can participate in constructing the world as experienced.
That proposal has significant empirical neighbours.
Pajani and colleagues found that spontaneous activity patterns in primary visual cortex before stimulus presentation could bias subsequent perception under ambiguous conditions. Powers, Mathys and Corlett experimentally induced perception-like auditory experiences through conditioning and found that stronger weighting of learned priors was associated with those reports. Dijkstra and colleagues showed across three experiments that congruent mental imagery made participants more liberal in reporting external stimulus presence, while also cautioning that the effect could involve sensory or decisional mechanisms.
These studies do not establish Schema Mask Theory.
They establish something more limited but essential to its present argument:
what is experienced cannot always be explained as an untouched sensory input followed only afterwards by conscious interpretation.
Prior state can participate in perceptual outcome.
8. From Computer Display to Phenomenal Field
The analogy therefore deepens.
A computer user does not experience transistor states.
They experience:
a document;
a button;
a message;
a failure;
a successful send.
Likewise, the conscious person does not ordinarily experience:
millions of sensory variables;
memory activations;
salience weightings;
bodily predictions;
learned associations;
and competing interpretations.
They experience:
a person who seems angry.
Or:
a room that feels unsafe.
Or:
something is wrong.
The theoretical proposal in Ghosts as Phenomenal Projections is that these may sometimes be understood as products of a phenomenal interface through which multiple underlying contributions become a manageable lived world.
The analogy must remain cautious.
A biological phenomenal field is not known to be a graphical desktop.
But the information problem is comparable:
enormous underlying complexity must become actionable experience.
9. Compression Is Necessary
A usable interface cannot show everything.
NIST’s human-factors guidance explicitly warns that unnecessary information competes with relevant information for visibility.
This is not merely a technical inconvenience.
Too much information can make an interface less informative.
The latest phenomenal-field article makes essentially the same theoretical observation from the organismic side. A symbolic representation can compress a large field of associations into a manageable experience. Instead of receiving thousands of separately articulated variables concerning danger, memory, powerlessness, bodily state and defensive response, consciousness may encounter a monster.
The compression is not necessarily defective.
It may be exactly what makes the information usable.
The scientific question comes later:
What has been preserved by that compression, and what has been lost?
10. The Human Symbolic Interface as a Provisional Model
The Human Symbolic Interface remains a project hypothesis rather than an established construct.
Its value here is that it gives a name to a familiar phenomenological observation: organismic information frequently becomes accessible through forms other than explicit analytical language.
The latest article identifies possible forms including sensation, emotion, image, intuition, dream, symbol and personification.
These need not all be produced by one mechanism.
Nor should the theory assume that every symbol is a deliberate coded message.
The more modest proposition is:
symbolic representation may sometimes provide an efficient interface through which otherwise difficult-to-articulate information becomes available for conscious relationship.
The computer analogy helps us maintain discipline here.
An icon is not necessarily the thing it represents.
Nor does understanding the icon tell us everything about the software underneath.
Likewise, understanding a psychological symbol may improve communication without establishing that the symbol literally depicts the underlying architecture.
11. One World May Be an Interface Achievement
The latest article makes a further move that is especially relevant to Schema Mask Theory.
It proposes that the human phenomenal field may be polyphenomenal.
Different specialised domains may attend to and weight different aspects of the same encounter. The body-language reader, verbal analyst, attachment system, frightened child and hopeful part may each contribute different interpretations. The consciously experienced result nevertheless tends to arrive as something resembling one world.
If this hypothesis is useful, the apparent unity of conscious experience should not automatically be taken as evidence that one process authored it.
Unity could instead be a product of integration.
That has a strong computer analogue.
The user sees one application window.
Behind it may be dozens of services.
The successful interface does not reveal that plurality because its purpose is to coordinate it.
The error occurs when:
phenomenal unity is mistaken for architectural singularity.
12. Different Domains May Be Seeing Different Things
The example developed in Ghosts as Phenomenal Projections makes the issue concrete.
Someone says:
“Everything’s fine.”
One process attends to the words.
Another notices the forced smile.
Another notices contracted body posture.
An attachment-related domain activates a memory of a previous departure.
A hopeful domain anticipates reconciliation.
A frightened domain anticipates abandonment.
The important question is no longer immediately:
Which one is correct?
It becomes:
What information is each domain actually receiving and contributing?
This resembles a distributed diagnostic system more than a courtroom.
Different sensors can report different features without any sensor necessarily being defective.
One reports temperature.
Another vibration.
Another pressure.
The larger regulator needs to know which instrument produced which reading.
13. Interface Bugs and Perspective Differences Must Be Distinguished
This creates an important design distinction.
Suppose a domain reports:
danger.
Several possibilities now exist.
The domain may have correctly detected information within its specialisation.
Its underlying information may be outdated.
Its interface may have exaggerated the signal.
Its contribution may have been combined incorrectly with another domain.
The receiving system may have misunderstood what the signal meant.
Or the signal may be accurate locally but wrongly elevated into a global conclusion.
Those are not the same failure.
This means a mature version of Schema Mask Theory will need to distinguish at least conceptually between:
source failure;
interface failure;
recipient interpretation failure;
and:
jurisdictional overextension.
The present article does not attempt to formalise those categories.
Its smaller point is that the distinction now appears necessary.
14. Provenance Becomes an Interface Requirement
The latest article introduces phenomenal provenance as the ability to locate a contribution within the domain from which it arose.
Instead of:
“That person is evil,”
individuation can begin asking:
Who is seeing evil?
and then:
What did that domain actually notice?
This resembles the provenance problem encountered in Part II.
A computer result may come from:
a primary source;
a cache;
a proxy;
a summary;
a classifier;
or a generated reconstruction.
The output may look the same.
Its informational status is not.
The phenomenal equivalent is:
The experience may be genuine while its source remains undifferentiated.
Recovering provenance does not necessarily erase the experience.
It changes its epistemic status.
15. From “Reality” to “A Contribution to Reality”
This may be the most useful single step provided by the new paper.
The movement is not necessarily:
true → false.
It is:
unlocated conclusion
→ located contribution.
The latest article describes the developmental movement from reality toward a contribution to reality from a particular source.
That is subtle but important.
Suppose an attachment-related domain says:
“She’s leaving.”
Individuation need not respond:
“You are wrong.”
It can move toward:
“This is what the current situation looks like from the attachment system’s history and available evidence.”
That formulation preserves the information while restoring jurisdiction.
The field becomes more differentiated without requiring the local contribution to disappear.
16. More Complexity Can Mean Better Interface Resolution
This initially creates a paradox.
A more individuated person may appear to experience more internal plurality.
The latest article gives an example in which:
“I hate this person”
is differentiated into several simultaneous contributions concerning boundaries, abandonment, social liking, body language and verbal coherence.
The result looks more complicated.
Informationally, however, it is more precise.
The computer analogy is straightforward.
A crude diagnostic interface says:
SYSTEM ERROR.
A better one says:
network available; authentication succeeded; database healthy; downstream mail service timed out.
The second interface contains more components.
It also contains less confusion.
That suggests a useful provisional principle:
Increasing differentiation can increase apparent complexity while reducing causal ambiguity.
This fits the wider Schema Mask account of individuation as increasing resolution.
17. Seeing the Interface Before Regulating the System
The new article also introduces a cybernetic comparison through Conant and Ashby’s Good Regulator theorem. Its careful use there is instructive: a regulator needs an adequate model of what it is attempting to regulate, but this does not establish that conscious awareness should become supreme controller.
The interface question comes first.
Before attempting to regulate:
What am I actually looking at?
A computer administrator who misreads a cache problem as database corruption can make the problem worse.
Likewise, a conscious system that interprets a specialised warning as evidence that an entire internal domain is irrational may intervene at the wrong level.
This connects directly with compartmentalisation confusion from Part I.
A coarse interface encourages coarse intervention.
18. The Interface Must Represent Uncertainty
A simple interface often hides uncertainty because certainty is easier to display.
Connected.
Failed.
Danger.
Safe.
But complex systems frequently inhabit states such as:
probably connected;
partially degraded;
confidence low;
source uncertain;
operation acknowledged but completion unverified.
Conversational AI now provides a useful contemporary illustration. A 2026 controlled study of 372 participants found that the way source information was presented in a conversational interface affected exploration and evidence integration; source visibility and placement altered how participants engaged critically with the information they received.
The underlying sources had not necessarily changed.
The presentation of provenance had.
This supports a general interface principle:
Information about information can materially alter how a recipient understands and uses it.
The organismic application remains theoretical, but the design question becomes unavoidable.
Could a phenomenal interface somehow distinguish:
I see danger
from:
I strongly predict danger
from:
this resembles an earlier danger
without forcing the receiving system to perform a full analytical reconstruction every time?
That question can remain open for now.
19. Training Requires Knowing What the Learner Thinks Happened
Part II introduced the possibility of training across interfaces.
The present article adds an important condition.
A training system cannot infer learning merely from what it presented.
It has to consider what the learner experienced and understood.
The email user may sincerely believe they clicked Send.
Repeatedly instructing:
Click Send
does not solve the problem if the interface has not revealed the mismatch between intended and registered action.
Similarly, an organismic training process would have to deal somehow with discrepancies between:
what was presented;
what was perceived;
and:
what the interfacee concluded.
This is where interface design becomes developmental rather than merely informational.
The next useful input depends partly upon the recipient’s current model.
20. The Baby Step May Be an Interface Correction
This gives the earlier baby-step idea a more modest place within the present article.
The next developmental step need not always provide new content.
It may correct an interface misunderstanding.
A user who cannot complete a complex operation may not need a more detailed explanation of the final goal.
They may need to learn what one unfamiliar icon means.
Likewise, within the developing psychological theory, a person seeking a large answer might first need to discriminate:
a sensation from an interpretation;
a prediction from an observation;
an image from an autobiographical claim;
or one domain’s perspective from the combined phenomenal field.
Those possibilities should not yet be turned into a full developmental hierarchy.
They simply show why understanding what the interfacee currently sees must precede any theory of graduated access.
21. Engagement Matters, but It Is a Separate Problem
The proposal that the Human Symbolic Interface may be gamified introduces another useful question, but it should remain deliberately small here.
An interface that exists for developmental learning faces an obvious practical problem:
the learner has to remain sufficiently engaged to continue interacting with it.
Computer-supported learning provides evidence that game-like design can sometimes alter learning behaviour. In a randomised experiment involving 701 students, Denny and colleagues found that badges increased voluntary self-testing for some participants and that this behavioural change mediated improved examination performance. Importantly, the authors’ result supports a pathway through changed learner behaviour rather than the simplistic idea that decorative game elements directly create learning.
That is enough for the present paper.
It establishes that presentation can influence whether a learner continues testing themselves.
It does not establish that the Human Symbolic Interface is gamified.
That remains a hypothesis requiring its own treatment.
22. A Symbol May Be Designed for Engagement Without Being Literal
This nevertheless creates an important safeguard.
If an interface uses symbol, narrative, mystery, challenge or personification partly because those forms sustain engagement, the compelling nature of the presentation cannot automatically be treated as evidence about the ontology behind it.
A dragon may be memorable.
A locked room may provoke curiosity.
A mysterious figure may invite dialogue.
The symbolic form may therefore serve several purposes simultaneously:
representing information;
compressing complexity;
directing attention;
and perhaps sustaining engagement.
The latest article already makes the necessary distinction when it argues that an apparition can be phenomenologically real without thereby settling its ultimate ontology.
Schema Mask Theory should preserve that discipline.
23. A Symbolic World Can Be a Training Environment
The possibility nevertheless deserves to remain visible.
Games teach through environments.
Rather than describing every rule in advance, they often allow a player to discover regularities through interaction.
A door does not open.
The player tries something.
A response occurs.
The player’s model changes.
The important general principle is not “the psyche is a video game.”
It is:
a represented environment can teach by allowing its participant to discover relationships through interaction rather than through explicit explanation alone.
This may eventually prove useful when analysing symbolic rooms, characters, recurrent environments and apparent tests within the Human Symbolic Interface.
But that is a later article.
For now, it is another question generated by the interface model.
24. The Interface May Be Teaching Its Own Grammar
A new interface cannot assume its user already knows how to read it.
This applies especially strongly where the representation is symbolic.
Suppose one recurring experience uses:
distance;
colour;
location;
direction;
doors;
changes in voice;
or changes in the behaviour of a personified figure.
The interfacee may initially focus upon the wrong feature.
They may assume colour matters when location matters.
They may interpret silence as rejection when silence means no current response.
They may interpret a closed door as permanent refusal when it represents an uncompleted process.
Nothing here establishes that such a symbolic grammar universally exists.
The theoretical point is narrower:
If symbolic communication is occurring, learning the interface conventions may itself be part of the developmental task.
That parallels ordinary interface learning remarkably closely.
25. The Phenomenal Interface Can Mislead Without Being Broken
This brings us back to the central distinction.
A computer interface can faithfully present:
request pending
while the user interprets:
request completed.
The interface is not necessarily malfunctioning.
The representation-recipient relationship is insufficiently calibrated.
Similarly, an organismic domain might faithfully contribute:
high threat probability from my local model
while consciousness receives simply:
danger.
The compression may be functional.
Problems arise when the recipient interprets:
danger exists objectively and universally.
The issue is therefore not necessarily false information.
It can be insufficiently differentiated information.
That maps closely onto the latest article’s concept of phenomenal solipsism: a local field initially appears to be reality itself because its source and jurisdiction have not yet been differentiated.
26. Phenomenal Solipsism as an Interface Problem
The term is useful precisely because it does not require philosophical solipsism.
A domain need not believe that nothing else exists.
It simply does not initially tag its own contribution:
“This is my historically conditioned interpretation of the available evidence.”
It outputs:
“She’s leaving.”
Or:
“He’s dangerous.”
Or:
“Everybody knows this.”
Computer interfaces do something similar.
They rarely say:
“This status is an interpretation generated from partial telemetry using assumptions A, B and C.”
They display:
HEALTHY
or:
FAILED.
This simplicity makes action possible.
It also creates the possibility that the output will be mistaken for an exhaustive description.
The developmental solution may therefore require increasing source visibility, not abolishing local perspectives.
27. Source Visibility Without Information Overload
That produces a difficult design problem.
If every phenomenal contribution arrived carrying complete provenance, confidence intervals, historical assumptions and dependency graphs, conscious experience might become unusable.
Computer interfaces confront the same trade-off.
The 2026 conversational-AI study by He and Liu found that source-interface layout affected verification and synthesis, while also identifying trade-offs between smooth workflow and designs that encourage reflective checking under information load.
This suggests that transparency itself needs an interface.
More information is not automatically more understanding.
The design problem is:
How much provenance needs to become visible, at what time, and in what representation?
That question now belongs squarely within Schema Mask Theory.
Its answer can wait.
28. The Apparent Bug May Be a Model Mismatch
We can now return to the ordinary computer user.
They report:
“Send is broken.”
The developer discovers that Send works perfectly.
The user had been clicking the adjacent control.
It would be easy to declare:
user error.
But the deeper design question is:
Why did the interface make that interpretation plausible?
Likewise, an apparent psychological failure might eventually need to be approached by asking:
What does the participant believe the interface is doing?
before concluding:
Which underlying part is defective?
That is a significant change in diagnostic posture.
It places the relationship between representation and recipient before blame.
29. Interface Debugging Requires Multiple Viewpoints
A good software investigation commonly compares several kinds of evidence.
What did the user report?
What did the interface display?
What events were registered?
What did the application log?
What did the downstream service receive?
What state was finally committed?
The discrepancy between those records helps localise the failure.
The latest phenomenal-field article proposes something structurally similar when it replaces:
Which one of you is right?
with:
What are you seeing?
Different local reports can be diagnostically valuable precisely because they are different.
One provides the phenomenal record.
Another provides bodily information.
Another provides remembered context.
Another provides an external observation.
The diversity is not noise to be eliminated immediately.
It is evidence from different positions.
30. Individuation as Better Diagnostics
This suggests another modest reinterpretation of individuation.
Individuation may increase the organism’s ability to distinguish:
event
from:
representation of event
from:
interpretation of representation
from:
source of interpretation.
The latest article calls this increasing differentiation of the phenomenal fields contributing to lived reality.
The computer analogy calls it better diagnostics.
Neither implies that every process must ultimately become consciously inspectable.
A well-designed computer remains full of hidden processes after diagnostics improve.
What improves is the ability to determine which level needs attention when something matters.
That seems an appropriately limited conclusion for the present stage of Schema Mask Theory.
31. Wholeness Does Not Require One Display
The latest article defines psychological wholeness in a way that becomes especially important here.
Different worlds can remain:
represented;
related;
mutually intelligible;
and appropriately bounded.
Wholeness therefore does not require every internal interface to display identical information.
A body-language reader should remain sensitive to posture.
A linguistic system should remain sensitive to words.
An attachment domain should remain sensitive to relational continuity.
Their differences are part of the organism’s informational richness.
The goal is not:
make every display identical.
It is:
make their differences sufficiently interpretable that no one local display has to masquerade as the whole system.
32. The Shared Phenomenal Field
If several domains contribute simultaneously, the consciously experienced field may be considered a kind of shared interface space.
This should not yet be formalised too rigidly.
The latest article suggests only that the total lived field may be negotiated or co-constructed from different contributions, with apparent singularity emerging from coordination.
The computer comparison helps us ask useful questions.
How are conflicting representations resolved?
Which contribution receives salience?
Can one process obscure another?
Can a high-priority warning temporarily dominate the entire display?
Can information remain present but backgrounded?
Can one representation be shown differently to different recipients?
Those questions are worth retaining.
They do not yet need answers.
33. Joy and Interface Hospitality
The latest article offers one further idea relevant to the developing series.
It proposes that joy, understood as an orienting rather than commanding principle, might phenomenologically correspond to:
the field becoming increasingly hospitable to the total organism.
Within the current interface discussion, that can be translated cautiously.
A high-integrity shared interface might gradually permit more legitimate information to participate without:
overwhelming the recipient;
silencing neighbouring contributors;
or requiring one local perspective to become universal.
This does not yet explain how such a process is governed.
It simply offers a possible criterion for later investigation:
Does increasing interface resolution allow more of the differentiated system to participate intelligibly?
That seems a useful question to carry forward.
34. What Part III Adds to the Model
Part III therefore introduces no need for a large new vocabulary.
Its principal contribution is a change in viewpoint.
Part II examined:
What happened across the interface?
Part III asks:
What did the recipient experience as having happened?
Those questions can diverge.
The computer user can believe they clicked Send when no send event occurred.
The user can receive a failure notification and attribute it to the wrong subsystem.
A conversational-AI user can receive the same information through different provenance displays and consequently engage with the evidence differently.
Perceptual research independently demonstrates that expectation, prior learning and imagery can affect perceptual reports and sensory interpretation.
And the developing phenomenal-field theory proposes that multiple organismic domains may contribute different locally meaningful constructions to the apparently singular world of conscious experience.
Together, these do not prove one unified mechanism.
They establish the value of the question.
35. Questions Now Open
The investigation leaves a new collection of questions for later articles.
If a domain has access to information that awareness does not, how does it select a representation?
How does a receiving domain learn the grammar of an unfamiliar interface?
How can an interface distinguish observation from prediction without overwhelming its recipient?
Can the same underlying information be rendered differently for recipients with different levels of competence?
How is misunderstanding detected?
How does the sending system know what the recipient thought happened?
Can successful interaction progressively increase interface resolution?
Can a symbolic interface use challenge, curiosity or game-like structure to sustain learning?
When does helpful compression become misleading simplification?
Can one domain’s representation dominate a shared phenomenal field?
How is provenance gradually restored?
And, eventually:
If these exchanges are developmental rather than merely mechanical, what decides what the recipient is ready to encounter next?
That final question points beyond Part III.
It should not yet be answered here.
Conclusion: What Did You See Happen?
The computer appeared to offer a simple model.
The user performs an action.
The system performs a function.
The interface reports the result.
Closer examination reveals something more complicated.
The user has an intention.
They perceive an affordance.
They attempt an action.
The system may or may not recognise it.
Several hidden processes may occur.
A result is produced.
The interface selects a representation.
The user perceives that representation.
They interpret it according to their existing model.
And that interpretation determines what they do next.
The interface therefore participates in the construction of the user’s experienced system.
This is why:
“I clicked Send”
cannot always be treated as a trivial disagreement over facts.
It may reveal a mismatch among intention, action, system state, feedback and phenomenal interpretation.
Human-computer interaction has developed extensive design principles around precisely such mismatches. Systems should make relevant state visible, provide timely feedback, prevent avoidable errors, support recovery, and present information at a resolution appropriate to the user rather than merely exposing implementation detail. Contemporary AI interfaces make the problem still more visible because the underlying capability is flexible and partly opaque; users may struggle to determine what the system can do, how to communicate their intention, and how to decide whether a fluent result actually fulfilled it.
Ghosts as Phenomenal Projections suggests a possible organismic extension. The conscious person may likewise inhabit an interface-generated field whose apparent simplicity conceals multiple sources. Different domains may notice different information, weight different histories and contribute different predictions. The apparent world may therefore be less like a raw screen onto reality and more like a coordinated phenomenal construction.
If that proposal proves useful, the first task of individuation is not necessarily to take control.
It may be to improve the diagnostics.
To move gradually from:
“This is what happened.”
toward:
“This is what I experienced happening.”
Then:
“This is the interface through which I experienced it.”
And eventually:
“These are some of the differentiated sources that contributed to what I experienced as one event.”
That progression does not make lived experience unreal.
It makes it more locatable.
The computer user still genuinely experienced pressing Send.
The frightened domain still genuinely experiences danger.
The symbolic figure still genuinely occupies the phenomenal field.
What improves is the ability to distinguish the experience of the interface from a complete account of the architecture behind it.
That is the baby step Part III contributes.
Before asking who guards the Gate, we first have to understand what someone standing in front of it can actually see.
References
Barnett, B. O., Brooks, J. A., & Freeman, J. B. (2021). Stereotypes bias face perception via orbitofrontal–fusiform cortical interaction. Social Cognitive and Affective Neuroscience, 16, 302–314. The study provides evidence that social-conceptual expectations can influence neural representations involved in face perception.
Denny, P., McDonald, F., Empson, R., Kelly, P., & Petersen, A. (2018). Empirical support for a causal relationship between gamification and learning outcomes. Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems. The randomised study found that particular game elements altered voluntary self-testing behaviour and that this behavioural change was associated with learning outcomes.
Dijkstra, N., Mazor, M., Kok, P., & Fleming, S. M. (2021). Mistaking imagination for reality: Congruent mental imagery leads to more liberal perceptual detection. Cognition, 212, 104719.
He, J., & Liu, J. (2026). Seeing to Think? How source transparency design shapes interactive information seeking and evaluation in conversational AI. The controlled study examines how different source-presentation interfaces alter evidence exploration, trust and synthesis.
Pajani, A., Kok, P., Kouider, S., & de Lange, F. P. (2015). Spontaneous activity patterns in primary visual cortex predispose to visual hallucinations. Journal of Neuroscience, 35, 12947–12953.
Powers, A. R., III, Mathys, C., & Corlett, P. R. (2017). Pavlovian conditioning-induced hallucinations result from overweighting of perceptual priors. Science, 357, 596–600.
Subramonyam, H., Pea, R., Pondoc, C. L., Agrawala, M., & Seifert, C. (2024). Bridging the Gulf of Envisioning: Cognitive challenges in prompt-based interactions with LLMs. Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems.
Walji, M. F., & Zhang, J. Expert Review and Human-Factors Guidance for User Interface Design. NIST-supported usability material. The guidance distinguishes feedback, data-availability, interpretation and integrity errors and emphasises visibility of system status and meaningful feedback.
Theoretical antecedent
Ghosts as Phenomenal Projections: From Externalised Complexes to the Construction of Lived Reality (2026). The article provides the immediate theoretical basis for the present extension through its proposals concerning polyphenomenal fields, phenomenal provenance, the Human Symbolic Interface, phenomenal sovereignty and individuation as differentiation of the locally constructed fields contributing to lived reality.
0 Comments