Amodal Completion

A class of phenomenal-organizational events identified by Kanizsa (1991) in which the perceptual system "completes" a partially occluded object — perceiving it as continuous behind the occluder despite the absence of any continuous distal stimulus. Amodal completion is direct and concrete perception of the invisible: we see an arrow piercing a body, see a single line passing behind a rectangle, see a hexagon as whole though only its visible portions are stimulated. The completion is not inferred, imagined, or conceptually integrated; it is a phenomenally explicit fact whose conditions are observable through experimental phenomenology.

In Taddio 2025 §3 amodal completion is the paradigm case for the autonomy of the phenomenal level from the cognitive level: vision operates by rules independent of conceptual categories. The painter exploits this autonomy — Magritte's Le Blanc-seing (1965) deliberately stages the conflict between cognitive expectation (the lady on horseback should not pass behind the tree) and amodal completion (and yet she does, visibly).

Key Points

  • "Direct and concrete perception of the invisible" (Kanizsa 1991: 51): the partially occluded object is seen as continuous, not inferred. The "surplus" of visibility (MP's term) is the visible-that-is-not-seen-but-still-perceived.
  • Distinguished from modal completion and from cognitive integration (Kanizsa's three-case taxonomy, Fig. 9 in Taddio 2025):
    • (a) Modal completion: a unified figure appears with phenomenal presence (e.g., a perceived ring is fully present even where its arcs are not stimulated).
    • (b) Mental/cognitive integration: two autonomous figures are conceptually joinable but each remains phenomenally distinct.
    • (c) Amodal completion: a partially occluded figure is perceived as a single, unified entity that maintains its phenomenal identity behind the occluder.
  • Operates by phenomenal-invariant conditions, not by past experience: in Kanizsa's panther/goat example (Fig. 8 in Taddio 2025), amodal completion imposes the perceptually-coercive (but logically counterintuitive) reading of one elongated animal across the occluder rather than the more intuitive reading of two distinct animals.
  • Autonomy from thought: "amodal completion illustrates that visual perception operates according to rules independent of the conceptual categories of thought" (Kanizsa 1991; Metzger 1954, cited Taddio §3). Vision and thought "often overlap and cooperate in synergy" but follow different "grammars."
  • Exploited by painters as expressive resource: Magritte's Le Blanc-seing uses the conflict between cognitive integration and amodal completion to produce ambiguity-as-effect. The viewer's gaze is denied "comfortable contemplation"; figure and ground oscillate.
  • A counterexample to sensation-based perception: if perception were built from atomic sensations corresponding to physical stimuli, amodal completion could not occur — there is no continuous distal stimulus to be sensed. The phenomenal level is irreducible.

Details

Wittgenstein on the Pierced Animal

The §3 anchor passage is from Wittgenstein's Philosophical Investigations: "I see that an animal in a picture is transfixed by an arrow. It has struck it in the throat and sticks out at the back of the neck. Let the picture be a silhouette. – Do you see the arrow – or do you merely know that these two bits are supposed to represent part of an arrow?" (Wittgenstein 2003: 173).

Wittgenstein's point — pressed by Taddio — is that the observer is not interpreting; they see the arrow piercing the figure. The two protruding segments are not separately registered and then mentally joined; they are perceived as one arrow whose middle section is occluded by the silhouette. The completion is amodal, in Kanizsa's sense.

Wittgenstein himself draws the comparison to Köhler's interpenetrating hexagons (in Remarks on the Philosophy of Psychology): we do not see three juxtaposed regions but two interpenetrating hexagons. The case is structurally identical.

Kanizsa's Taxonomy (Fig. 9)

Kanizsa (1991: 33) distinguishes three completion types using a U-shaped figure:

  • (a) Modal: the gray U with horizontal black bars suggests a white rectangle in front; the rectangle is present (modally) though not stimulated.
  • (b) Cognitive integration: the U alone, without the rectangle; two arcs that could be conceptually joined into a ring but remain phenomenally autonomous.
  • (c) Amodal: the U with a white rectangle superimposed; the U is perceived as a single ring partially occluded by the rectangle, maintaining its unity despite the obstruction.

Only (c) is amodal in the strict sense. The taxonomy matters because it isolates the phenomenal phenomenon Taddio's argument turns on: completion that is neither fully present nor a matter of conceptual joining but a third, phenomenologically distinct mode of givenness.

The Conflict-Cases (Fig. 8 and Le Blanc-seing)

Two examples make the autonomy of perception from thought vivid:

Kanizsa's panther/goat row (Fig. 8 in Taddio 2025): three panthers, an occluder bar, three goats. The cognitively-natural reading would be: two distinct animal groups (three panthers and three goats) partially obscured. But amodal completion favors the perceptually coercive (but biologically nonsensical) reading: one elongated animal merging panther-front and goat-back across the occluder. The phenomenal-invariant condition (good continuation, edge alignment at right angles) imposes the unified solution against the more "logical" alternative.

Magritte's Le Blanc-seing (1965): the lady on horseback passing through trees in a forest. Some trees seem to leap to the foreground, splitting the horse; some appear to disappear as the lady passes. The painting "exploits the perceptual disorientation between figure and background caused by the discord between mental inference and amodal completion" (Taddio §3). We know (cognitively) that the woman cannot remain visible while the horse passes behind the tree; yet the unity of lady-on-horseback persists. The painting "is animated by an internal dynamic that prevents our gaze from settling into comfortable contemplation."

Magritte's painting is structurally an intentional staging of the phenomenon Kanizsa's panther/goat row reveals experimentally: amodal completion overrides cognitive integration, and the painter can use this override as expressive material.

Cézanne and Amodal Completion (Fig. 12)

Taddio extends amodal completion to Cézanne's Mont Sainte-Victoire (Fig. 13 in Taddio 2025). The Kanizsa figure (Fig. 12) shows a zigzag line partially hidden by a rectangle; among multiple logically-possible completions, only one is perceptually realized. MP's claim — "the sight of the pictures... provides me with the only existing Cézanne" (PhP 174) — receives a phenomenal-experimental gloss: the painting offers one perceptual solution among many possibilities. Perception cannot be imagined or narrated; it must be lived hic et nunc.

The application is not metaphorical. Cézanne's mountain is constituted by amodal-completion factors among others: edges, occlusion relations, color modulations all combine to fix the perceptual outcome. The painter has structurally limited the viewer's interpretive freedom by satisfying the conditions for one specific completion rather than another.

Why Sensation-Based Theories Fail

The negative pendant of the positive thesis. If perception were built from "sensation" — atomic correspondence between proximal stimulus and conscious content — then where the distal stimulus is missing (the occluded portion of the line, the hidden body of the arrow), no sensation should occur, and we should perceive only the visible fragments as separate units. But we don't: we perceive a single arrow, a single line, a single hexagon. Sensation cannot deliver this unity. Therefore the explanatory base is at the level of phenomenal organization, not at the level of stimulus correspondence.

Taddio: "If we were to adopt the concept of 'sensation' as the basis for perception, we would miss its meaning. By adhering to this notion, we could never claim to see a single line. Sensation presupposes a correspondence between what we see, the proximal stimulus, and the distal stimulus. In this case, since no such correspondence exists, we should perceive five segregated and autonomous units rather than a single entity (Fig. 6)" (§3).

Amodal completion is therefore a paradigm case for the irreducibility of the phenomenal level — and one of the three classes (with the Kanizsa triangle and transparency) that anchor Taddio's broader anti-physicalism.

Connections

  • is a class of phenomenal-invariants — the conditions for amodal completion are dependent/independent variables of the experimental-phenomenology tradition; they exemplify the broader structural class.
  • exemplifies gestalt-principles-of-unification — particularly good continuation, closure, and Prägnanz; amodal completion is built on these unification factors operating through the occluder.
  • operates within the *système d'équivalences* — amodal completion conditions are shared between worldly and pictorial perception.
  • exemplifies phenomenal autonomy from cognition — vision operates by rules independent of thought (Kanizsa, Metzger).
  • exploited as expressive resource by Magritte (Le Blanc-seing, L'idée, Le pèlerin, Paysage de Baucis).
  • exemplified in Cézanne's Mont Sainte-Victoire — the painting fixes one phenomenal solution among many logically possible completions.
  • contrasts with sensation-based theories of perception — the fragments-into-unity move cannot be built from atomic sensations.
  • parallels figure-ground-relationship as a precondition of phenomenal unity — both establish the unity of an object against alternatives.

Open Questions

  • Cross-source weight: this is currently a single-source page. Strengthening requires re-ingesting MP's PhP (where the "surplus of visibility" structure is implicit in the chapter on perception of the body and chapters on space) or future ingests of Kanizsa's primary texts.
  • Limits of amodal completion: when does the system fail to complete? Conflict-cases (Magritte) are deliberate stagings of partial failure. Are there structural factors (gap size, edge orientation, color discontinuity) that determine the threshold? Taddio cites Kanizsa but does not develop a threshold theory.
  • Relation to predictive-coding accounts: post-Marr cognitive science models perception as Bayesian inference. Amodal completion would be modeled as a high-prior posterior. Is this an alternative explanation or a different-level description? Phenomenology and Bayesian inference may describe the same regularities at different levels.

Sources

  • taddio-2025-art-and-psychology — §3 ("Amodal Completion") is the section-length treatment. Wittgenstein's pierced-animal example, Kanizsa's three-case taxonomy (Fig. 9), the panther/goat row (Fig. 8), Magritte's Le Blanc-seing, the Cézanne Mont Sainte-Victoire application via Fig. 12. Primary site for the concept on the wiki.
  • Indirect anchors via Taddio: Kanizsa 1991 Vedere e pensare; Wittgenstein 2003 Philosophical Investigations; Wittgenstein 1980 Remarks on the Philosophy of Psychology §§ 641, 644, 645; Metzger 1954 Psychologie. None ingested as primary sources yet.