Kappa effect¶

A psychophysical illusion in which the perceived duration between two stimuli lengthens when they are farther apart in space and shortens when closer, because the perceptual system reconstructs elapsed time from spatial layout under an implicit constant-velocity prior.

Core Idea¶

The kappa effect is a psychophysical illusion in which the perceived duration of the interval between two successive stimuli is systematically lengthened when the stimuli are farther apart in space and shortened when they are closer together, even though the physical time interval is held constant. The same 500-millisecond gap between two light flashes feels longer when the flashes are separated by a larger spatial distance and shorter when the distance is small — a bias in which the perceived dimension (time) is pulled in the direction predicted by the spatial dimension (distance).

The mechanism is interpreted as a perceptual prior on constant-velocity motion: the visual system models the stimulus sequence as produced by an object moving at some speed, and infers — implicitly and automatically — that greater spatial separation between events requires more elapsed time if velocity is constant. The prior was originally demonstrated by Cohen, Hansel, and Sylvester (1953) using spatially separated light flashes, and has since been replicated across sensory modalities — in audition, where the distance dimension is pitch separation between tones, and in touch, where taps farther apart on the skin lengthen the perceived inter-tap interval — indicating that the constant-velocity prior operates on the common neural currency of event-sequence representation rather than within a single sense. The kappa effect is one half of a bidirectional space–time interaction pair: its mirror, the tau effect, shows the complementary bias in which temporal interval distorts perceived spatial distance.

Structural Signature¶

Sig role-phrases:

the two-dimension event — a stimulus sequence carrying both a temporal interval and a spatial separation (or its modality analog: pitch separation, skin distance)
the physical orthogonality — time and distance vary independently in the actual stimulus, so any coupling cannot be in the input
the judged dimension — the one the observer is asked to estimate (duration, for kappa), left perceptually underdetermined
the constant-velocity prior — the implicit, automatic assumption that the sequence was generated by an object moving at fixed speed
the cross-dimension inference — applying that prior so the under-determined dimension is reconstructed from the other (more space ⇒ more elapsed time at constant velocity)
the directional bias — perceived duration pulled the way the prior predicts: greater separation lengthens it, lesser separation shortens it, with magnitude set by the separation
the integration-stage locus — the coupling arises where the dimensions are combined under the prior, on a common event-sequence representation shared across vision, audition, and touch
the tau mirror — the same prior run the other way, warping perceived distance from temporal interval — the complementary half of one assumption

What It Is Not¶

Not sloppy or noisy time estimation. The distortion is not error or inattention to the clock; it is the systematic signature of an implicit constant-velocity prior reconstructing duration from spatial layout. The perceived interval is exactly the elapsed time an object would have needed to cross that distance at fixed speed — a determinate inference, not scatter.
Not a coupling present in the stimulus. Time and distance vary independently in the input, so the bias cannot be in the input; it arises at the integration stage where the two dimensions are combined under the prior. The orthogonality of the physical dimensions is exactly what locates the effect downstream of perception.
Not a vision-specific illusion. The same bias appears in audition (with pitch separation as the distance dimension) and touch (with skin separation), so it is not a quirk of the visual timing machinery. The prior operates on a common event-sequence representation shared across senses, which is what the modality-spanning evidence reveals.
Not anchoring. It superficially resembles "one dimension biases the estimate of another," but anchoring works through numerical cognitive contagion — a salient number dragging a numerical judgment. The kappa effect is perceptual integration under a motion model; the cross-domain cases that fit the surface description belong to anchoring, not to this prior.
Not independent of the tau effect. Kappa is one half of a bidirectional pair: its mirror, the tau effect, runs the identical constant-velocity prior the other way, warping perceived distance from temporal interval. They are two readings of one assumption, not two unrelated illusions to be catalogued separately.

Scope of Application¶

The kappa effect lives entirely within the psychophysics of time perception, recurring across the sensory modalities that share one neural machinery and one explanatory commitment — an implicit constant-velocity prior operating on a common event-sequence representation; its reach is within that perceptual substrate, since even the broader perceptual_prior / bayesian_brain pattern stays bound to perceiving systems and genuine cross-domain "one estimate pulls another" cases belong to anchoring, not to this prior.

Visual perception — the home turf: the original Cohen-Hansel-Sylvester light-flash demonstrations, where greater spatial separation lengthens the perceived temporal gap.
Auditory perception — sequential tones in which pitch separation acts as the distance dimension, larger pitch jumps lengthening the perceived inter-tone interval.
Tactile perception — tap sequences on the skin, where taps farther apart on the body lengthen the perceived inter-tap interval.
Cross-modal integration — audiovisual timing paradigms that draw on the same constant-velocity prior across senses (with the mirror tau effect warping perceived distance from temporal interval).

Clarity¶

Naming the kappa effect makes a specific bias in subjective time judgment legible and, crucially, attributes it: the distortion is not noise in the observer's timing, nor a failure to attend to the clock, but the signature of an implicit constant-velocity prior reconstructing duration from spatial layout. Without the label, the finding that an identical 500-millisecond gap "feels longer" when its endpoints are farther apart looks like sloppy temporal estimation; with it, the analyst reads a system inferring the elapsed time an object would have needed to cross that distance at constant speed. That reframes the question from "how accurate is the perceived interval?" to "what does the perceptual model assume about the motion that generated this event sequence?" — and once the prior is named, its direction is predictable rather than mysterious: greater separation lengthens perceived time, lesser separation shortens it.

The concept sharpens two distinctions a vaguer "time perception is biased" would lose. First, it separates the physically orthogonal dimensions — time and distance vary independently in the stimulus — from the perceptually coupled judgment, so the bias is located in the integration stage, not in the input. Second, it fixes kappa's place as one half of a bidirectional pair: its mirror, the tau effect, runs the same prior the other way (temporal interval distorting perceived distance), and holding the two as complementary readings of a single constant-velocity assumption is what unifies them rather than cataloguing two unrelated illusions. Because the same bias recurs across vision, audition (pitch as the distance dimension), and touch, naming it also tells the perception researcher where the inference lives — on a common event-sequence representation rather than inside any one sense — which is the sharper question the modality-spanning evidence was always pointing at.

Manages Complexity¶

The psychophysics of subjective duration accumulates a scatter of distortions that, taken one at a time, look like a list of separate quirks: identical visual gaps that feel longer when their flashes are spread out and shorter when bunched; auditory intervals that stretch when the bounding tones jump far in pitch; tactile intervals that lengthen when the taps land far apart on the skin; and the converse family in which temporal spacing warps perceived distance. Catalogued as independent illusions, each invites its own ad hoc story about that modality's timing machinery, and the modality-spanning recurrence reads as coincidence. The kappa effect compresses this by attributing the whole family to a single inferential commitment — an implicit constant-velocity prior that reconstructs one perceptual dimension from the other on the assumption that the event sequence was generated by an object moving at fixed speed. Once that prior is named, the analyst no longer re-derives each illusion but reads its direction and sign straight off the assumption: hold velocity constant and greater spatial separation requires more elapsed time, so wider-apart events are judged longer and closer-together events shorter, with the mirror reading (the tau effect) running the identical prior the other way to warp distance from time. A roster of seemingly unrelated biases becomes one assumption applied to whichever dimension is left ambiguous.

What this leaves the perception researcher to track shrinks to a compact structure rather than the full timing profile of each sense. First, which two dimensions of the event are at play and which is being judged — duration inferred from a spatial dimension (kappa) or distance inferred from a temporal one (tau); the dimensions are physically orthogonal in the stimulus, so the bias is known to live in the integration stage, not the input, and that localization is itself part of the compression. Second, the magnitude and sign of the separation along the non-judged dimension, from which the direction of the pull follows by the constant-velocity rule. Third — and this is where the modality-spanning evidence pays off — the recognition that the prior operates on a common event-sequence representation rather than inside vision or audition or touch separately, so the same compact account covers pitch-as-distance and skin-position-as-distance without a fresh model per sense. From these the qualitative outcome is fixed: given a layout, the perceptual model predicts whether perceived time stretches or compresses, and by which direction, the same way across modalities. A high-dimensional pile of perceptual anomalies collapses to a single prior plus the two-dimension bookkeeping needed to read off which way it bends the judgment — and it converts the question from "how accurate is the perceived interval?" to "what motion does the model assume generated this sequence?", which is the form in which the answer is actually determinate.

Abstract Reasoning¶

The kappa effect licenses a prior-attribution move that the perception researcher runs to convert an apparent timing error into a determinate inference about the perceptual model. Confronted with the finding that an identical 500-millisecond gap "feels longer" when its endpoints are farther apart, the analyst does not read sloppy temporal estimation or inattention to the clock; the signature is a perceived dimension pulled in the direction the other dimension predicts, and that points to an implicit constant-velocity prior reconstructing duration from spatial layout. The move reasons FROM "perceived time tracks spatial separation at fixed physical interval" TO "the visual system is inferring the elapsed time an object would have needed to cross that distance at constant speed." This reframes the operative question from "how accurate is the perceived interval?" — which has no determinate answer — to "what motion does the perceptual model assume generated this event sequence?", which does. The predictive move then reads the sign and direction straight off the prior rather than measuring it case by case: holding velocity constant, greater spatial separation requires more elapsed time, so the analyst predicts wider-apart events are judged longer and closer-together events shorter, and predicts the magnitude of the pull from the size of the separation along the non-judged dimension. The mirror prediction is part of the same reasoning: the tau effect runs the identical prior the other way, so the analyst forecasts that a temporal interval will warp perceived distance by the same constant-velocity logic, and treats kappa and tau as two readings of one assumption rather than two unrelated illusions.

The localization move pins where the bias lives and uses the orthogonality of the stimulus dimensions as evidence: because time and distance vary independently in the input, the analyst reasons that the coupling cannot be in the input and must arise at the integration stage where the two are combined under the prior. The modality-spanning recurrence drives a second localization inference: the same bias appears in vision, in audition with pitch separation as the distance dimension, and in touch with skin-position separation, so the analyst reasons that the prior operates on a common event-sequence representation shared across senses rather than inside any single sensory channel — and predicts that a newly tested modality with a mappable "distance" dimension should show the same kappa bias. The boundary-drawing move fixes the construct's scope to its preconditions: it requires two perceptual dimensions of one event that the system can plausibly link by a motion model, with one dimension left ambiguous for the prior to reconstruct. So the analyst predicts the effect where an event sequence invites a constant-velocity reading and the judged dimension is underdetermined, and reasons that it should weaken or vanish where the perceptual model has no motion interpretation to impose or where the judged dimension is given unambiguously — the prior having nothing to fill in.

Knowledge Transfer¶

Within the psychophysics of perception the effect transfers as mechanism, across the sensory modalities that share one underlying neural machinery and one explanatory commitment — the implicit constant-velocity prior operating on a common event-sequence representation. The prior-attribution move, the read-the-sign-off-the-prior prediction, the integration-stage localization (justified by the orthogonality of the stimulus dimensions), and the two-dimension bookkeeping all carry intact. In visual perception it is the original Cohen-Hansel-Sylvester flashes. In auditory perception it is sequential tones with pitch separation as the distance dimension. In tactile perception it is taps whose skin-separation lengthens the perceived inter-tap interval. In cross-modal work it underwrites audiovisual timing paradigms. Its mirror, the tau effect, runs the identical prior the other way to warp perceived distance from time, so the two are unified as readings of one assumption rather than catalogued separately. Crucially — and the seed is firm here — these modalities are not three independent domains: they share a single neural mechanism, so this is breadth across senses on one perceptual substrate, and the prediction (a newly tested modality with a mappable distance dimension should show the same kappa bias) follows because the mechanism is the same.

Beyond perception the honest reading is that the effect essentially does not transfer — this is one of the most strongly home-confined entries. The seed is blunt that it "does not reorganise how non-perceptual domains see their work": cosmologists, project managers, and economists gain no traction on duration estimation by importing it, and it gives engineers, lawyers, and biologists no new way to think about their domains. What little is portable is shared abstract mechanism of a thin, still-perception-bound kind: the broader perceptual prior / bayesian_brain pattern — a perceptual system filling in an underdetermined dimension from an implicit generative model — of which kappa (and tau, and other motion-related biases) are worked-example children. But even that parent is bound to perceiving systems with Bayesian priors over sensory input; it is not a substrate-neutral prime that reaches physics or institutions. What stays home-bound is the entire specific apparatus: the constant-velocity prior, the time-and-distance dimension pair, the common event-sequence representation it runs on, and the modality-spanning evidence. A tempting surface analogy — "one dimension biases the estimate of another" — does not license exporting kappa, because the cross-domain cases that fit that surface (a salient number dragging a numerical estimate) are anchoring, a different mechanism operating through numerical cognitive contagion rather than perceptual integration under a motion model. So invoking "a kappa effect" outside perception is metaphor built on a surface resemblance whose actual cross-domain instances belong to anchoring, not to this prior. The discipline is to carry the perceptual_prior / bayesian_brain parent only within perceiving systems, to route genuine cross-domain "one estimate pulls another" cases to anchoring, and to reserve "kappa effect" for the perceptual time-from-space inference its constant-velocity prior actually produces (see Structural Core vs. Domain Accent).

Examples¶

Canonical¶

The founding demonstration is Cohen, Hansel, and Sylvester's 1953 experiment with three successive light flashes. Three lamps, call them A, B, and C, are flashed in sequence along a line. The physical time intervals between A–B and B–C are set equal, but the spatial gaps are made unequal — B is placed nearer A and farther from C. Observers report that the interval on the wider spatial gap (B–C) feels longer than the interval on the narrower gap (A–B), even though the two occupied identical clock time. The perceived duration is dragged in the direction of the spatial separation. The standard reading is that the visual system models the flashes as a single object in motion and, assuming constant speed, infers that covering the larger distance must have taken more time.

Mapped back: The flash sequence is the two-dimension event, carrying independent time and space intervals — the physical orthogonality that forces the coupling downstream. Duration is the judged dimension; the assumption of a fixed-speed moving object is the constant-velocity prior, and the longer-felt wide gap is the directional bias it produces at the integration-stage locus.

Applied / In Practice¶

The effect is a real design constraint in haptics — the engineering of vibrotactile displays that communicate through patterns of vibration on the skin, as used in wearable alerts, sensory-substitution aids for blind users, and touch feedback in devices. Because a tactile kappa effect operates on the skin (taps spaced farther apart on the body lengthen the perceived interval between them), the spatial layout of a stimulator array distorts the perceived timing of the signals it delivers. A designer who spaces actuators for anatomical convenience can inadvertently make evenly timed pulses feel irregular, corrupting rhythm-based tactile codes. Haptic researchers therefore account for the space-to-time bias when laying out actuators and timing pulse trains, so that intended temporal patterns are perceived as intended rather than warped by the skin's constant-velocity inference.

Mapped back: A pulse train across skin locations is the two-dimension event; perceived timing is the judged dimension the array's spatial layout perturbs via the constant-velocity prior. Evenly timed pulses felt as irregular because the actuators are far apart is the directional bias, and correcting the layout works on the integration-stage locus where space and time are combined.

Structural Tensions¶

T1: Illusion versus veridical inference (a bias only where the world violates the prior). The construct insists the kappa effect is not sloppy timing but the correct output of a constant-velocity prior — the perceived interval is exactly the time an object would need to cross that distance at fixed speed. But that reframing cuts both ways: in the natural world, where event sequences usually are generated by objects moving at roughly constant speed, the prior yields veridical perception, and the "illusion" appears only because the experimenter artificially decouples time from distance. So the same inference is adaptive accuracy in the ecology it evolved for and a systematic error in the orthogonalized lab stimulus. Calling it an illusion imports a defect connotation that the prior's ordinary correctness contradicts; calling it inference understates the real, measurable distortion under decoupling. Diagnostic: In this stimulus, does the constant-velocity assumption hold (making the percept veridical), or has time and distance been decoupled so the prior's usually-correct fill-in becomes an error?

T2: The constant-velocity prior versus alternative couplings (the phenomenon underdetermines the mechanism). Attributing kappa and tau to one implicit motion prior is what unifies them as two readings of a single assumption rather than two catalogued illusions — an elegant compression. But the observed bias (perceived time tracks separation, and vice versa) is also consistent with more general magnitude-coupling accounts in which any two prothetic dimensions contaminate each other's estimates, no motion model required. The construct commits to the constant-velocity interpretation, and much of its predictive machinery (read the sign off "more space needs more time at fixed speed") inherits that commitment. If the true coupling is a generic magnitude interaction, the motion-prior story is a plausible over-specification that happens to fit. Diagnostic: Does the case require a motion model to predict the bias, or would a dimension-agnostic magnitude-coupling account produce the same sign and magnitude without the constant-velocity assumption?

T3: Automatic prior versus correctable perception (design works around it, never removes it). The haptics application shows the effect is actionable: space actuators to pre-compensate, and evenly timed pulses can be made to feel even. But this works only by distorting the stimulus to cancel the bias — the prior itself is automatic, implicit, and cognitively impenetrable, so it cannot be switched off, learned away, or attended past. The designer's leverage is real but strictly indirect: every mitigation is a pre-distortion that assumes the bias's magnitude is known and stable, and a layout tuned for one observer or context may over- or under-correct for another. The construct offers control over the input and none over the inference. Diagnostic: Is the remedy pre-distorting the stimulus to offset a known, stable bias magnitude — or is it (impossibly) trying to make the perceptual system stop applying the prior?

T4: Orthogonality localization versus the common-representation leap (how far the evidence reaches). The construct's cleanest argument is that because time and distance vary independently in the stimulus, the coupling cannot live in the input and must arise at the integration stage — a genuinely tight inference. It then extends this, from the modality-spanning recurrence (vision, pitch, skin), to the stronger claim that one common event-sequence representation shared across senses carries the prior. That second step is supported only correlationally: the same-shaped bias in three modalities is consistent with a shared representation but also with three parallel modality-specific priors of the same form. The localization argument that is airtight for "downstream of input" is suggestive-not-decisive for "one shared substrate." Diagnostic: Does the evidence establish only that the bias is post-input, or does it actually license the stronger claim of a single cross-modal representation rather than parallel same-form priors?

T5: Autonomy versus reduction (a perceptual prior, with anchoring as a false friend). The kappa effect is a genuinely named, canonically studied illusion with irreducible home cargo — the constant-velocity prior, the time-and-distance pair, the tau mirror, the modality-spanning evidence — and it transfers as mechanism across the senses that share its neural machinery. But it is among the most home-confined entries: even its parent, the perceptual_prior / bayesian_brain pattern (a system filling an underdetermined dimension from an implicit generative model), stays bound to perceiving systems and reaches no non-perceptual domain. Its tempting surface — "one dimension biases the estimate of another" — is a trap, because the cross-domain cases that fit that description (a salient number dragging a numerical estimate) are anchoring, a different mechanism entirely. The tension is between a vivid named illusion and the recognition that its only genuine generalization is a perception-bound prior, with anchoring poaching its surface elsewhere. Diagnostic: Resolve toward perceptual_prior / bayesian_brain only within perceiving systems; route any non-perceptual "one estimate pulls another" to anchoring; and reserve "kappa effect" for the perceptual time-from-space inference the constant-velocity prior actually produces.

Structural–Framed Character¶

The kappa effect sits toward the structural end of the spectrum but stops short of the pole — mixed-structural, closely parallel to the irrelevant speech effect: a genuine, evaluatively neutral perceptual mechanism whose vocabulary is locked to one substrate. Four of the five criteria read structural. Evaluative_weight is nil: the effect is a determinate perceptual inference (perceived time reconstructed from spatial layout under a motion prior), neither good nor bad; the words "illusion" and "bias" carry a faint defect connotation, but the entry's own T1 shows the inference is veridical in the ecology it evolved for, so it names a mechanism, not a fault. Institutional_origin is none: the constant-velocity prior is a fact of how perceptual systems reconstruct an underdetermined dimension, not an artifact of any survey, agency, or convention — Cohen, Hansel, and Sylvester discovered a regularity the perceptual system already ran. It is not human-practice-bound in the constitutive sense: the effect runs inside a perceiving brain, observer-free, whenever an event sequence carries a spatial and a temporal dimension and one is left ambiguous — no social practice constitutes it, and it dissolves only if there is no perceiving system, not if there is no human institution. And within its proper range, cross-modal reuse is recognition rather than import: the same prior is recognized intact across vision, audition (pitch as distance), and touch (skin separation), because they share one neural machinery.

What keeps it off the structural pole — and makes it one of the most home-confined entries — is vocab_travels, which it fails decisively, together with a tight substrate-lock. The operative vocabulary — constant-velocity prior, event-sequence representation, time-and-distance pair, the tau mirror — is irreducibly perceptual and floats free of no non-perceiving substrate; beyond perception the effect does not transfer at all, and the surface phrase "one dimension biases another" is a false friend whose genuine cross-domain instances belong to anchoring, a different mechanism, not to this prior. Even its parent stays perception-bound. The portable structural skeleton is a perceptual system reconstructing an underdetermined dimension from an implicit generative model — the perceptual_prior / bayesian_brain pattern — genuinely recurrent, but only across perceiving systems, and exactly what the kappa effect instantiates from that umbrella rather than what makes "kappa effect" itself travel: the reach belongs to the perceptual-prior parent (within perception), while the constant-velocity, time-from-space specifics stay home. Its character: structural in skeleton — a real, evaluatively neutral, recognized-in-perception inference mechanism — but stated in a constant-velocity-prior vocabulary that locks it to perceiving systems and does not reach beyond them, leaving it mixed-structural rather than a free-floating prime.

Structural Core vs. Domain Accent¶

This section decides why the kappa effect is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity in one place.

What is skeletal (could lift toward a cross-domain prime). Strip the time-perception content and a thin relational structure survives: a system reconstructs an underdetermined dimension of an event from another dimension by applying an implicit generative model, so the judged dimension is pulled in the direction the model predicts from the other. The portable pieces are abstract — an event carrying two dimensions, one left ambiguous, an implicit prior linking them, and a directional fill-in of the ambiguous dimension from the observed one. That skeleton is genuinely substrate-portable within perceiving systems — it is the shape of Bayesian perceptual inference — which is exactly why the entry instantiates the catalog's perceptual_prior / bayesian_brain pattern, of which kappa (and its tau mirror) are worked-example children. That recurrence is mechanism, but it is the core the kappa effect shares, not what makes it distinctive — and notably even this parent stays bound to perceiving systems, not a substrate-neutral prime.

What is domain-bound. Everything that makes the effect the kappa effect in particular is time-perception furniture and does not survive extraction. The implicit model is specifically the constant-velocity prior (a sequence read as an object moving at fixed speed); the two dimensions are specifically time and distance (or their modality analogs, pitch separation and skin separation); the judged dimension is perceived duration; the coupling is located at the integration stage on a common event-sequence representation shared across vision, audition, and touch; and the effect comes paired with its tau mirror running the identical prior the other way. The decisive test: remove the constant-velocity motion model — take a stimulus the perceptual system has no motion interpretation for, or give the judged dimension unambiguously — and the effect weakens or vanishes, because the prior has nothing to fill in. And a tempting surface phrase, "one dimension biases the estimate of another," is a false friend: its genuine cross-domain instances (a salient number dragging a numerical estimate) are anchoring, a different mechanism operating through numerical cognitive contagion, not this perceptual integration under a motion model.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose transfer is recognition of the same mechanism, not analogy. The kappa effect is among the most home-confined entries. Within the psychophysics of perception it transfers as mechanism intact — visual, auditory (pitch as distance), tactile (skin separation), and cross-modal timing — but these are not independent domains; they share one neural machinery, so this is breadth across senses on a single perceptual substrate, not cross-substrate travel. Beyond perception it essentially does not transfer at all: it reorganizes no non-perceptual domain's work, and invoking "a kappa effect" outside perception is metaphor built on a surface resemblance whose real cross-domain instances belong to anchoring. And when a genuinely portable lesson is wanted, the most that carries is the perceptual_prior / bayesian_brain parent — and only within perceiving systems — while genuine non-perceptual "one estimate pulls another" cases route to anchoring. The cross-domain reach that exists belongs to the perceptual-prior parent (bounded to perception); "the kappa effect," as named, packs the constant-velocity prior and the time-from-space specifics that stay home in perceiving systems.

Relationships to Other Abstractions¶

Current abstraction Kappa effect Domain-specific

Parents (1) — more general patterns this builds on

Kappa effect is a decomposition of Model Assumption Failure Prime

The Kappa Effect is the perceptual form of Model Assumption Failure in which a constant-velocity prior is load-bearing but false for independently placed stimuli, distorting inferred duration.

Hierarchy paths (2) — routes to 2 parentless roots

Kappa effect → Model Assumption Failure → Representation → Abstraction

Show alternative path (1)

Not to Be Confused With¶

Tau effect. The mirror sibling, not a separate illusion: it runs the identical constant-velocity prior the other way, so that the temporal interval between events distorts their perceived spatial distance (events closer in time are judged closer in space). Kappa infers time from space; tau infers space from time. They are two readings of one assumption. Tell: which dimension is judged and which does the pulling — perceived duration pulled by spatial separation (kappa), or perceived distance pulled by temporal interval (tau)?
Anchoring. The false friend: it superficially fits "one dimension biases the estimate of another," but works through numerical cognitive contagion — a salient number dragging a numerical judgment — not perceptual integration under a motion model. Cross-domain cases that resemble kappa's surface (a quoted figure skewing an estimate) are anchoring, a different mechanism. Tell: is the bias a perceptual reconstruction of an underdetermined sensory dimension under a motion prior (kappa), or a numerical estimate pulled toward a salient reference value (anchoring)?
Generic magnitude coupling / dimensional interference. The rival explanation of the same data: an account in which any two prothetic (magnitude) dimensions contaminate each other's estimates, with no motion model required. It predicts the same sign of bias as kappa but denies the constant-velocity story. The kappa effect specifically commits to the motion-prior interpretation; magnitude coupling is dimension-agnostic. Tell: does predicting the bias require a constant-velocity motion assumption (kappa proper), or would a dimension-blind magnitude interaction produce the same pull without any motion model (generic coupling)?
Chronostasis / other subjective-duration illusions. Not every distortion of perceived time is kappa: chronostasis (the "stopped clock," where the first glance at a clock seems to last unusually long after a saccade) and time dilation for oddball or looming stimuli arise from saccadic remapping or attention/arousal mechanisms, not from a spatial-separation prior. Kappa specifically reconstructs duration from a spatial (or pitch/skin) separation dimension. Tell: is the perceived-duration distortion driven by spatial layout of the event sequence under a motion prior (kappa), or by saccades, attention, or arousal with no distance dimension in play (other timing illusions)?
Perceptual prior / Bayesian brain (the parent). The umbrella the kappa effect instantiates — a perceptual system reconstructing an underdetermined dimension from an implicit generative model — of which kappa and tau are worked-example children. Crucially this parent is itself bounded to perceiving systems; it is not a substrate-neutral prime that reaches physics or institutions. Tell: strip away the constant-velocity prior and the time-and-distance pair and what remains — "a perceptual system fills an ambiguous dimension from an implicit prior" — is the perceptual-prior parent, treated more fully elsewhere; but keep even that within perceiving systems, and route any non-perceptual "one estimate pulls another" to anchoring.

Neighborhood in Abstraction Space¶

Kappa effect sits in a sparse region of the domain-specific corpus (75^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Psychophysical Laws of Perception (10 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12