The object exceeds the lit surface

Intentionality is often translated as “aboutness”. That is useful only if it does not shrink the problem. A perception, recollection, fear, hope or judgement is more than tagged with a topic. Each is directed in a distinctive way. To fear the storm, remember the storm and inspect the storm are three different acts, even when they concern what we call the same storm. Husserl’s central move was to analyse this correlation between an experience and what it is an experience of, rather than treating consciousness as a container filled with mental items.[6]

Consider the kettle again. The current visual profile is perspectival. It is given from here, under this light, at this distance, with this bodily orientation. The kettle is not given as another profile alongside that profile. It is given through it, as the identical object that can show other sides. Husserl called such perspectival presentations Abschattungen, often translated as profiles, adumbrations or shadings. No finite run of them exhausts a spatial thing.[8]

The familiar arrow from subject to object hides three different problems. An act can be directed towards something that does not exist. Different acts can concern the same object without giving it in the same way. The same sensory pattern can also be taken as different objects, as when a yellow curve is first seen as a kettle edge and then as paint on a flat prop. A causal ray from object to retina does not explain these differences, and an internal picture merely reproduces the puzzle inside the head.

Husserl’s language of correlation is meant to keep both sides in view. We ask how an object appears with a certain sense, and how the act is structured as an act of that object. This does not mean that consciousness manufactures the kettle out of nothing. Nor does it mean that the kettle’s physical existence is irrelevant. It means that physical existence alone does not specify how the thing is encountered as one object, from here, as usable, doubtful, remembered or expected. The explanatory target is the relation between manners of givenness and objectivity.

The same noun can conceal different acts

Suppose you say “the yellow kettle” four times. Once you see it, once you remember it, once you imagine a better version, and once you judge that it belongs to the museum. The nominal object stays verbally stable. The mode of givenness does not. Perception presents the object as bodily there. Memory presents it as having been there. Imagination presents a possibility without positing it as actual. Judgement takes a state of affairs as the case. Husserl’s later vocabulary distinguishes the noetic side, the intending activity or mode, from the noematic side, the intended as intended.[2]

Object, object-as-intended and noema

This is where an easy diagram becomes dangerous. It is tempting to place a “noema” inside the mind as a miniature object between the act and the real kettle. Husserl’s texts do not support such a simple intermediary picture, and the scholarly debate remains live. Some interpretations treat the noema as an ideal sense or way of presentation distinct from the object. Others understand it as the object precisely as intended. Recent work tries to reconcile these emphases by locating them within different methodological contexts.[9]

We therefore need a neutral working expression: object-sense, meaning the object as this act gives it. It is the kettle as seen from here, as remembered from yesterday, as imagined with a copper handle, or as judged to be borrowed. This phrase does not settle the ontology of the noema. It keeps apart three items that careless analysis collapses: the occurrent act, the object as given in that act, and whatever worldly object may or may not satisfy the intention.

flowchart TD
  A["The kettle that was not a kettle"] --> B["The object exceeds the lit surface"]
  B --> C["Four relations inside one<br/>experience"]
  C --> D["Fulfilment makes directedness<br/>testable"]
  D -. "re-test" .-> A
  classDef first fill:#0a1424,stroke:#B96D00,stroke-width:2px,color:#e6f3ff
  classDef step fill:#0a1424,stroke:#3f7da0,stroke-width:1px,color:#e6f3ff
  class A first
  class B,C,D step

Four relations inside one experience

The opening case can now be decomposed. The act is seeing. The object-sense is this usable yellow kettle in the spotlight. The present profile is its illuminated front. The horizon includes a rear, handle, cavity, stable shape under movement, and a surrounding room in which the object can be approached. Fulfilment occurs as later profiles arrive. Disappointment occurs when the prop thins into a sheet.

These are not five objects stored side by side. They are moments in one dynamic correlation. The object is experienced as more than the profile because the profile belongs to a horizon-governed trajectory of possible givenness. Husserl’s descriptions of horizon stress that an object is always further determinable. What is absent is not simply missing. It is co-intended with a degree and style of determinacy.[7]

Horizon is constrained openness

A horizon is sometimes described as “everything in the background”. That is too indiscriminate. When you see the kettle, its internal horizon includes unseen sides and further properties that belong to this thing. Its external horizon includes the table, room, museum, practical setting and wider world within which the thing appears. Not every imaginable future is equally implicated. A handle is more strongly prefigured than a pair of wings.

Horizon therefore combines determinacy and indeterminacy. You may expect something handle-like without expecting the exact scratch pattern. You may anticipate that walking left will reveal more of the spout without calculating a pixel-perfect image. Husserl’s phrase “determinable indeterminacy” captures this useful middle. The absent side is not specified in full, yet it is not a blank.[8]

The horizon also has a practical topology. Some possibilities are near because one small movement would disclose them. Others are remote, such as the kettle’s manufacturing history. Some are tied to bodily capacities: the handle is graspable from a certain orientation, the lid liftable with a certain force, the vessel inspectable only by changing posture. The lived body is not an optional sensor mount added to an already complete picture. It helps define which appearances are available from where and through what movement. Work on fulfilment has therefore emphasised its embodied dimension rather than treating intuition as passive receipt.[10]

This is one reason a horizon should not be equated with a list of explicit predictions. Most of its structure can be operative without being stated. Before moving, you may never formulate “the ellipse will widen by twelve degrees”. Yet a wildly discontinuous change can still surprise you as inappropriate to the same thing. The horizon is broader than forecast, but narrower than unconstrained possibility. It is a patterned readiness for how this object may continue to show itself.

The horizon contract

To make the mechanism usable, we can express directed experience as a horizon contract. This is a practitioner synthesis, not a term from Husserl. A contract records four obligations that are usually compressed into the statement “the subject experiences X”.

The mode clause states how X is given. The identity clause states what must remain invariant for later profiles to belong to the same X. The access clause states which action or transition should disclose what. The revision clause states what would refine, reclassify or defeat the original object-sense. Directedness is thin when these clauses are absent or interchangeable. It becomes analytically rich when different interventions have different expected consequences.

A small formalism, and what it does not claim

Formal notation can compress the contract, provided we do not mistake compression for explanation. Let the current profile at time t be pt. Let the horizon contain anticipated transitions. Each transition has an action ai, an anticipated profile p̂i, a weight wi, and a criticality flag ci. After an action, the observed match mi ranges from −1 for contradiction to +1 for fulfilment. Untested anticipations receive zero.

F

an illustrative fulfilment balance, not a measure of consciousness or truth;

wᵢ

the declared importance of an anticipation for this analysis;

mᵢ

the outcome after a specified intervention;

cᵢ

whether contradiction threatens the identity of the object rather than a peripheral property.

The equation encodes three assumptions. Anticipations are expressible, weights are declared before outcome, and fulfilment can be assessed relative to an intervention. Remove those assumptions and the score loses meaning. Even when they hold, the score says only that a description has a coherent pattern of anticipation and revision. It does not show that a person consciously entertained explicit probabilities. It does not establish that the brain computes this formula. It does not resolve the metaphysics of constitution.

The criticality flag prevents a common error in quantitative evaluation. Averaging treats every mismatch as exchangeable. Directed experience often does not. A red handle instead of a black one may leave kettle identity intact. A flat printed rear may not. The distinction is not fixed for all purposes. A conservator authenticating a historical kettle may treat the exact glaze as identity-critical, while a thirsty visitor does not. The horizon is therefore indexed not only to an object type, but also to the act, interest and practical situation in which the object is given.

Fulfilment makes directedness testable

Husserl’s Logical Investigations analyses knowledge through the movement from empty or merely signitive intention towards intuitive fulfilment. You hear that a rare kettle is in the gallery. The sentence directs you towards it without presenting it bodily. You enter the room and see something that seems to answer the intention. You approach, rotate and inspect it. Fulfilment can deepen by degrees, yet the spatial object still exceeds every profile.[1]

This gives directedness a temporal structure. An intention is not validated merely because the next sensation resembles an expectation. It is fulfilled when a later givenness presents the relevant object or state of affairs in a way appropriate to the act. A testimony can answer a question differently from perception. A photograph can support identity without presenting the object as bodily present. A measurement can fulfil a quantitative judgement while leaving colour untouched.

Fulfilment is mode-relative, partial and revisable. It is also distinct from confidence. Husserl’s notion of evidence, Evidenz, concerns the self-givenness of what is intended, not a psychological feeling of certainty. A person can feel completely certain while the horizon is poorly tested. A cautious observer can receive strong fulfilment while remaining verbally hesitant.[8]

Four outcomes are worth keeping separate. Confirmation gives what was anticipated in the expected style. Enrichment adds determinations that were left open, such as finding a maker’s mark on the base. Revision preserves the object while altering a predicate, as when the rear is orange rather than yellow. Defeat breaks an identity-critical commitment, as when the apparent vessel has no interior. These outcomes form no universal ladder. A single encounter can confirm shape, revise colour and defeat ownership at once.

This layered result matters for knowledge. The fulfilment of one intention does not automatically validate every judgement attached to it. Seeing the kettle can fulfil “there is a vessel-shaped object here” while leaving “it is authentic”, “it is safe to use” and “it belongs to the museum” empty. Evidence must be matched to the exact object-sense and act it is meant to answer. Otherwise a vivid perception lends borrowed authority to claims it never presented.

Disappointment is more informative than easy agreement

A static label can agree with thousands of images without revealing what the system treats as the same object. A targeted intervention is more diagnostic. Rotate the object while preserving colour. Change the handle while preserving silhouette. Break spatiotemporal continuity while preserving all later pixels. Each intervention asks whether identity depends on profile similarity, trajectory, manipulability, prior category or declared evidence.

The horizon contract predicts a characteristic pattern. Peripheral violations should cause local revision. Identity-critical violations should cause reclassification. Untested features should remain open rather than silently counted as true. A system that simply repeats “kettle” after every contradiction has not displayed stable directedness. It has displayed output persistence.

The perfect hallucination boundary

Empirical findings can constrain nearby questions without deciding Husserl’s transcendental claims. Change-blindness research shows that observers can miss large alterations across disruptions, which challenges the intuition that a fully detailed scene is continuously represented.[14] Sensorimotor theories argue that seeing depends on active mastery of lawful relations between movement and sensory change.[13] Research on amodal completion studies how partly occluded objects are experienced as continuing behind an occluder.[15]

These programmes are compatible with parts of the horizon picture, especially perspectival access and active exploration. They do not prove that horizonality is reducible to sensorimotor contingency, prediction or completion. Husserl is analysing the structure of givenness and objectivity. Cognitive science seeks causal and computational explanations. Compatibility is not identity of explanatory level.

A productive bridge must state what is preserved. In both cases, a current view can be related to possible changes under movement. What is not preserved is equally important. A computational model may encode transition probabilities without anything being given to a subject. A first-person description may reveal a stable style of appearance without identifying the responsible neural dynamics. The experimental insight is to design matched interventions across levels, then type the conclusions separately: phenomenological structure, behavioural sensitivity, computational state and physical cause.

The bridge also yields a prediction that is weaker, and therefore more defensible, than a theory of consciousness. When an observer or system genuinely relies on action-conditioned access, an impossible transition should disrupt object identity more than an equally large but horizon-compatible change. A kettle that becomes slightly darker under shadow may remain the same kettle. A front that rotates without revealing any side may trigger a different judgement. That pattern can be tested without claiming that the tested mechanism is intentionality itself.

stateDiagram-v2
  [*] --> Claim
  state "The same noun can conceal<br/>different acts" as Claim
  state "Object, object-as-intended and<br/>noema" as Test
  state "Horizon is constrained openness" as Revision
  Claim --> Test: expose mechanism
  Test --> Decision: evidence survives
  Test --> Revision: rival survives
  Revision --> Claim: narrow or rebuild
  Decision --> [*]

Run an intentional analysis

Phenomenology becomes vague when “describe the experience” means free association. A useful analysis constrains attention without pretending that the resulting report is incorrigible. The protocol below isolates one episode, records the act mode, names the object as given, separates present profile from horizon, and specifies an intervention before classifying fulfilment.

The protocol can be used in a first-person exercise, an interview, a design review or a machine evaluation. The evidential meaning changes by use. A first-person report is evidence about reported experience. A system trace is evidence about configured behaviour and mechanism. Neither automatically transfers to the other.

Disciplined first-person work does not require private certainty. Two observers can be asked to inspect the same kind of episode, undergo the same variation and report where their descriptions converge or diverge. The result is not a vote on reality. It is a structured record of which distinctions survive repeated attention. Reports can then be compared with behaviour, timing, eye movement, neural measures or system traces. The sources constrain one another only when the bridge between them is stated.

The protocol also guards against retrospective storytelling. Ask for the horizon before the revealing action whenever possible. If the observer declares every outcome expected after seeing it, disappointment disappears by definition. In machine evaluation, version the model, prompt, context, tools and world state before intervention. A revision that depends on hidden evaluator feedback is not evidence that the original system carried the relevant horizon.

Six moves

Freeze the episode. Choose a brief, concrete event: “seeing the kettle before walking around it”, not “my experience of objects”. Type the act. Record perception, memory, imagination, judgement, desire, fear or another mode. Name the object-sense. State the object as given, including relevant predicates: “this usable kettle here”, rather than only “kettle”.

Separate profile and horizon. Record what is presently given, then what is co-intended as further accessible. Specify an intervention. Move, question, inspect, compare, retrieve or wait. It must have a discriminating expected result. Classify the outcome. Mark fulfilment, revision, defeat or remaining untested. Do not turn “untested” into agreement.

What the lab tests. It tests whether an episode description contains discriminating anticipations, declared importance and revision rules. What a positive result permits. You may say that the account has a coherent, intervention-sensitive intentional structure. What a negative result means. The object talk may be under-specified, indifferent to action or unable to revise under contradiction.

What it cannot establish. It cannot prove veridicality, consciousness, neural implementation, metaphysical idealism or formal equivalence between a machine trace and human experience. A dream may exhibit a coherent internal horizon while lacking the same worldly correction path. A language model may produce excellent horizon vocabulary without maintaining object identity through action.

Strongest boundaries and rival explanations

The horizon contract could succeed for reasons that have little to do with phenomenology. Better object memory, stronger causal models, active perception or explicit state estimation may produce the same intervention pattern. That is not a defect. It is a reminder that the contract is a comparative test of configured organisation, not an ontology detector.

The proposal would be strengthened if targeted violations produced selective revision: colour changes adjust a property, trajectory breaks lower identity, and failed affordances trigger reclassification. It would be weakened if a static classifier, given no horizon representation, matched every result. A rival explanation would be generic anomaly detection. To distinguish it, test whether the system predicts which action will disclose which property, rather than merely noticing surprise after the fact.

Another rival is narrative repair. A language model may explain every contradiction fluently by inventing a new story, while retaining no stable identity criteria across turns. The defence is not to demand confidence. It is to require preregistered break conditions and replayable receipts. Which property was expected? Which action was authorised? What was observed? Which part of the object-sense changed? A system that cannot answer those questions may still be useful, but its apparent intentional coherence should not be overread.

The protocol should not be applied as a universal score. Emotions, mathematical insight and social perception have horizons that are not reducible to turning an object in space. Cultural familiarity can make possibilities salient for one observer and invisible to another. Trauma, expertise, fatigue and disability can reshape access. The contract is therefore a scaffold for local analysis. Its value lies in forcing the analyst to declare what is being preserved and what would count as a real change.

function assess(rows) {
  let weighted = 0;
  let totalWeight = 0;
  const criticalDefeats = [];

  for (const row of rows) {
    const weight = Number(row.weight);
    const outcome = Number(row.outcome); // -1, 0, 0.35, or 1
    weighted += weight * outcome;
    totalWeight += weight;

    if (row.critical && outcome === -1) {
      criticalDefeats.push(row.label);
    }
  }

  return {
    balance: totalWeight ? weighted / totalWeight : 0,
    criticalDefeats
  };
}