Crespi Effect

Read operant behavior after a reward change as a discrepancy-driven transient rather than an absolute-value response: animals shifted up overshoot same-reward controls (elation) and animals shifted down undershoot them (depression), calibrated to the signed gap between delivery and a recalibrating expectation.

Core Idea

The Crespi effect names the contrast-based overshoot in operant behavior that follows a sudden change in reward magnitude: animals shifted to a higher reward respond at a level above that of animals trained on the higher reward all along (positive contrast, or "elation effect"), while animals shifted to a lower reward respond at a level below that of animals trained on the lower reward all along (negative contrast, or "depression effect"). In both cases the behavioral output is calibrated not to the absolute value of the current reward but to the discrepancy between that reward and the prior expectation — overshooting the new steady state in the direction of the shift before eventually recalibrating to match animals with no shift history.

Leo Crespi established the phenomenon in 1942 with rats running a straight runway for sucrose pellets: animals shifted from 1 pellet to 16 ran faster than the stable-16 group for several sessions, then converged; animals shifted from 16 to 1 ran slower than the stable-1 group, again transiently. The structural commitment is a four-part sequence: a history of reward at one magnitude sets an expectation; a sudden shift produces a signed discrepancy between expectation and delivery; that discrepancy drives a transient response amplified beyond what the new absolute magnitude would sustain; repeated exposure to the new magnitude recalibrates the reference and the amplification decays. The asymmetry between elation and depression effects — negative contrast is typically larger and more persistent — anticipates the asymmetry formalised decades later in prospect theory's loss-aversion parameter.

Structural Signature

Sig role-phrases:

• the reinforced behavior — an operant response (running, lever-pressing) maintained by a reward
• the reward history — a prior run at one magnitude that sets a recalibrating expectation
• the magnitude shift — a sudden change to a higher or lower reward, the perturbation
• the signed discrepancy — the gap between the new delivery and the prior expectation, with a direction (up = positive, down = negative)
• the contrast overshoot — a transient response amplified beyond what the new absolute magnitude alone would sustain, in the direction of the shift (elation up, depression down)
• the same-reward control — the stable group trained on the new magnitude throughout, against which the overshoot is read as a control-relative gap
• the recalibration decay — repeated exposure moves the reference toward the new magnitude, shrinking the discrepancy to zero and relaxing the overshoot to the no-shift level
• the loss-side asymmetry — negative contrast typically runs larger and more persistent than positive contrast, the behavioral ancestor of loss aversion
• the regime boundary — contrast governs only around change points where a discrepancy exists; at the recalibrated steady state the absolute-value account suffices

What It Is Not

• Not a response calibrated to the absolute reward. The shifted animal's behavior is keyed to the signed discrepancy between delivery and a prior expectation, not to how much reward it now receives — which is exactly why it over- or under-shoots a same-reward control. The current magnitude alone under-predicts the response around a change point.
• Not a permanent new response rate. The overshoot is a transient: as repeated exposure moves the reference toward the new magnitude, the discrepancy shrinks to zero and behavior relaxes to the no-shift level. The steady state, once the reference catches up, is calibrated to the new absolute reward; the contrast lives only in the settling.
• Not regression to the mean or a measurement artifact. The elated runner outpacing controls and the depressed runner lagging them are real behavioral overshoots produced by a reference mechanism, not statistical drift toward an average or noise in the apparatus. Crespi's two-group design isolates the effect by holding the current reward fixed and varying only history.
• Not the step-response overshoot of a physical system. An underdamped circuit or sprung mass that overshoots a step input produces a curve that looks like the Crespi transient but is governed by inductance, momentum, or reactance — with no reward, no expectation, and no reference point. The resemblance is shape only; the motivational mechanism that defines the effect is absent.
• Not a symmetric pair of effects. Elation (positive contrast) and depression (negative contrast) are one discrepancy mechanism read at opposite signs, but they are not equal: negative contrast typically runs larger and more persistent — a loss against expectation weighs heavier than an equal gain, the behavioral ancestor of loss aversion.

Scope of Application

The Crespi effect lives across the reward-driven-behavior subfields of psychology — animal learning, human motivation, and the applied behavioral sciences — wherever a system represents reward magnitude and forms a recalibrating expectation about it; that precondition bounds its reach, and the look-alike step-response overshoot of an underdamped physical system (governed by inductance or momentum, with no reference point) is resemblance only, not a habitat.

• Animal learning — the founding case: Crespi's runway rats shifted between sucrose-pellet magnitudes, showing elation (positive contrast) and depression (negative contrast) overshoots relative to same-reward controls.
• Human operant and motivational research — the same signed-discrepancy transient wherever a reward shifts against an established baseline, foundational to reference-point models of motivation.
• Workplace psychology — pay-cut studies show depression effects deeper than the new wage alone predicts, and pay-raise studies show transient elation, exactly the contrast overshoot keyed to a prior expectation.
• Consumer behavior — response to discount-then-restore pricing and post-upgrade or post-downgrade satisfaction reports, where the prior reference, not the absolute level, sets the reaction.
• Behavioral economics and prospect theory — the elation/depression asymmetry (negative contrast running larger and more persistent) is the behavioral ancestor of loss aversion and reads as the asymmetric form of the negative-contrast component.

Clarity

The Crespi effect makes legible a distinction that absolute-value accounts of reinforcement blur: the difference between an animal's response to how much reward it now receives and its response to how the current reward compares with what it had come to expect. Without the concept, the elated runner who outpaces a same-reward control and the depressed runner who lags a same-reward control look like noise or like animals that simply differ — until one sees that response level is keyed to the signed discrepancy between delivery and expectation, not to the magnitude on its own. Crespi's two-group design is what forces this apart, holding the current reward fixed while varying only the history that precedes it, so the contrast can be read off directly as the gap between shifted and stable animals.

This sharpens the questions a learning researcher can ask around any change in reinforcement. Rather than asking only "what response will this reward magnitude sustain?", the analyst now asks "what was the prior expectation, what is the sign and size of the discrepancy, and how long until the reference recalibrates and the overshoot decays?" — recognizing the response as a transient governed by expectation rather than a new steady state. The effect also makes the elation–depression asymmetry a legible object in its own right: that negative contrast tends to run larger and longer than positive contrast frames the question of why losses relative to expectation bite harder than equivalent gains, a directional puzzle within reinforcement learning that the symmetric absolute-value picture cannot even pose.

Manages Complexity

An absolute-value account of reinforcement leaves a learning researcher with a residue of unexplained behavior around every change in reward. Animals shifted to a richer reward run faster than animals trained on that reward throughout; animals shifted to a leaner one run slower than animals trained on the lean reward throughout; the gaps appear, then fade; and they differ in size depending on which way the shift went. Item by item these look like noise, individual variation, or a tangle of separate runway, lever, and consumption findings, each needing its own ad hoc patch onto the response-versus-magnitude curve.

The effect compresses that residue by relocating the controlling variable from the reward's absolute magnitude to a signed discrepancy between the delivered reward and a recalibrating expectation. The analyst then stops tracking the full reward-history-to-response map and tracks four quantities instead: the prior expectation set by reward history, the sign of the shift, the size of the discrepancy, and the recalibration rate at which the reference moves to the new magnitude. The behavioral output around any change point is read off these — not as a new steady state but as a transient whose direction follows the sign of the discrepancy, whose amplitude scales with its size, and whose decay follows the recalibration as the overshoot relaxes to the no-shift level.

The branch structure falls out of the sign. A positive discrepancy (shift up) predicts elation, the overshoot above the same-reward control; a negative discrepancy (shift down) predicts depression, the undershoot below it; a zero discrepancy (no shift) predicts the stable baseline with no transient at all — so the two surface phenomena, elation and depression, collapse into one mechanism read at opposite signs, and a designer can predict from history alone whether a given manipulation will overshoot up, overshoot down, or stay put. One further regularity is folded in rather than added on: the asymmetry by which negative contrast runs larger and longer than positive contrast becomes a single property of the discrepancy response — losses against expectation weigh heavier than equal gains — letting the analyst carry the magnitude and persistence of the overshoot, not just its direction, on the same compact parameter set.

Abstract Reasoning

The Crespi effect licenses a set of reasoning moves by which the learning researcher reads operant behavior around a reward shift as a discrepancy-driven transient rather than an absolute-value response, all grounded in the signed gap between delivery and a recalibrating expectation. The foundational move is predicting an overshoot from the sign of the discrepancy. Knowing an animal's reward history and the new magnitude, the researcher reasons not to a new steady state but to a transient whose direction follows the sign of the shift: a positive discrepancy (shift up) predicts elation — running or responding above the same-reward control — while a negative discrepancy (shift down) predicts depression, responding below it, and a zero discrepancy predicts the stable baseline with no transient at all. The inference runs from the relation between expectation and delivery to the direction of the behavioral overshoot, so the researcher forecasts whether a manipulation will overshoot up, overshoot down, or stay put from the history alone.

A second move is diagnostic inference of the reference point from a control-relative anomaly. Confronted with an animal that outpaces or lags a same-reward control, the researcher does not treat the gap as noise or individual difference but reasons backward to a hidden expectation: the response is keyed to the signed discrepancy between delivery and what the animal had come to expect, so the anomaly reveals the prior reference rather than a defect in the current reward. Crespi's two-group design is the instrument that licenses this — holding the current reward fixed while varying only the preceding history — so the contrast is read directly off the gap between shifted and stable animals as evidence of the expectation each carries.

A third move is predicting the decay of the transient from the recalibration rate. Because repeated exposure to the new magnitude moves the reference toward it, the researcher reasons that the overshoot is temporary and will relax to the no-shift level as the discrepancy shrinks to zero, and predicts the time course of that relaxation from how fast the reference recalibrates. This reframes the analyst's question around any change point from "what response will this magnitude sustain?" to "what was the prior expectation, what is the sign and size of the discrepancy, and how long until the reference recalibrates and the overshoot decays?" — treating the post-shift behavior as a settling transient with a predictable endpoint rather than a permanent new rate.

A fourth move is predicting amplitude and asymmetry from the discrepancy's size and direction. The researcher reasons that the overshoot's magnitude scales with the size of the discrepancy, and folds in the directional regularity that negative contrast typically runs larger and more persistent than positive contrast — that a loss relative to expectation weighs heavier than an equal gain. This lets the researcher predict not merely the direction of the transient but its size and persistence, anticipating that a downward shift of a given magnitude will produce a deeper and longer-lasting deviation than an upward shift of the same magnitude.

A fifth move is boundary-drawing on where contrast governs versus absolute level. The researcher reasons that history-dependence is what produces contrast, so contrast effects are expected specifically around change points where a discrepancy exists, while the eventual steady state, once the reference has caught up, is calibrated to the new absolute magnitude. This tells the analyst when to expect the absolute-value account to under-predict the response (immediately after a shift, where the overshoot lives) and when it suffices (at the recalibrated steady state), bounding the regime in which the discrepancy mechanism, rather than the magnitude alone, controls behavior.

Knowledge Transfer

Within reward-driven behavior the effect transfers as mechanism, because its precondition — a system that represents reward magnitude and forms a recalibrating expectation about it — is met across the relevant subfields. The diagnostics (read a control-relative over- or under-shoot as evidence of a hidden reference point), the predictions (direction from the sign of the discrepancy, amplitude from its size, decay from the recalibration rate, extra weight on the downward side), and the vocabulary (positive/negative contrast, elation/depression, reference point) all carry intact. In animal learning it is the founding case — Crespi's runway rats. In human operant and motivational work the same contrast appears wherever reward shifts against an established baseline. In workplace psychology, pay-cut studies show depression effects deeper than the new wage alone predicts and pay-raise studies show transient elation, exactly the signed-discrepancy transient. In consumer behavior, discount-then-restore pricing and post-upgrade/downgrade satisfaction reports are the same overshoot keyed to a prior reference. Across these the mechanism is the same object — only the reward currency and the response measure change — so the analysis ports without translation.

Beyond cognitive substrates the picture has two strands that must not be confused. The first is a genuine shared abstract mechanism, and it belongs to the parent, not to the Crespi effect's named machinery. The deep, substrate-spanning structure here is response calibrated to a reference point, with contrast-based overshoot around a shift, and this really does recur across reward-driven domains — it is the same structure prospect theory formalized, and the elation/depression asymmetry is the behavioral ancestor of loss aversion. But that recurring object is already in the catalog as reference_point (and the broader contrast pattern); the cross-domain lesson "expect contrast, not absolute-level effects, around any change point in a history-dependent system" should be carried under those headings. What stays home-bound is everything specifically Crespi: the operant runway paradigm, the two-group elation/depression demonstration, the motivational reading of the overshoot. The second strand is a false analogy worth flagging explicitly, because the surface invites it: a physical system that overshoots after a step input — an underdamped circuit, a sprung mass, a control loop — produces a curve that looks like the Crespi transient but is governed by entirely different mechanisms (inductance, momentum, reactance), with no reward, no expectation, and no reference point in sight. Calling that "a Crespi effect" borrows the shape while dropping the motivational mechanism that defines the original, and should be marked as resemblance, not transfer. The honest boundary, then: as mechanism the effect reaches exactly as far as systems that represent reward and form expectations; the genuine cross-domain insight is the reference-point/contrast parent's to carry, not this named effect's; and step-response overshoot in physical systems is a look-alike, not an instance (see Structural Core vs. Domain Accent).

Examples

Canonical

Leo Crespi's 1942 runway experiment is the founding demonstration, and its two-group design is what makes the effect visible. Rats ran a straight alley for sucrose-pellet reward. One group had been trained on a small reward (roughly 1 pellet) and was then abruptly shifted to a large reward (roughly 16 pellets); another had been trained on the large reward all along. The shifted-up rats did not merely rise to the large-reward level — they ran faster than the stable-large group for several sessions before converging down to it (positive contrast, "elation"). Symmetrically, rats shifted from the large to the small reward ran slower than rats trained on the small reward throughout, again transiently, before converging up (negative contrast, "depression").

Mapped back: The prior training magnitude is the reward history setting an expectation; the abrupt change is the magnitude shift, and the gap between new delivery and expectation is the signed discrepancy. Running faster or slower than the matched group is the contrast overshoot read against the same-reward control. Convergence over sessions is the recalibration decay as the reference catches up — and that the downward shift bit harder is the loss-side asymmetry.

Applied / In Practice

Field studies of pay cuts show the depression effect in a human workplace. In Greenberg's 1990 study of manufacturing plants, a temporary company-wide pay reduction (about 15%, following a lost contract) was imposed at some sites while a comparable site kept full pay. During the pay-cut period, employee theft rose sharply at the reduced-pay plants and then returned to baseline once pay was restored, while the full-pay site stayed stable. Workers' behavior tracked not the absolute (still positive) wage but the drop relative to the wage they had come to expect — a transient reaction that faded as pay was restored.

Mapped back: The established wage is the reward history fixing the reference; the cut is a negative magnitude shift producing a negative signed discrepancy. Behavior worse than the still-substantial new wage alone would predict is the contrast overshoot on the depression side, and its fading once pay was restored is the recalibration decay. That a loss against the expected wage produced so strong a response illustrates the loss-side asymmetry Crespi first observed in rats.

Structural Tensions

T1: Signed discrepancy versus absolute reward (behavior that depends on history, not level). The effect's founding move is to relocate the controlling variable from how much reward is delivered to the signed gap between delivery and a prior expectation — which is what explains the elated runner outpacing, and the depressed runner lagging, a same-reward control. But that relocation means the same reward magnitude produces different behavior depending only on what came before, so there is no stable reward-to-response curve around a change point: the absolute-value account systematically under-predicts exactly where history has created a discrepancy. The tension is that the insight which dissolves the anomalies also denies reinforcement a fixed dose-response reading, making behavior contingent on an expectation that must itself be reconstructed. Diagnostic: Is the response being predicted from the current reward magnitude, or from the signed discrepancy between that magnitude and the reference the animal's history established?

T2: The overshoot transient versus the recalibrated steady state (which response is the real one). The contrast lives only in the settling: as repeated exposure moves the reference toward the new magnitude, the discrepancy shrinks to zero and behavior relaxes to the no-shift level, which is calibrated to the new absolute reward. So the effect names a transient that eventually erases itself, and the "true" response to the reward is genuinely ambiguous — the overshoot is real and consequential, but so is the steady state it decays to. Which one governs depends entirely on when behavior is measured relative to the shift. The tension is that a single reward magnitude has two legitimate behavioral answers, the contrast-amplified one near the change point and the absolute-calibrated one after recalibration, and the concept's value is precisely in refusing to let either stand for the whole. Diagnostic: Is the behavior in view being read near the change point (contrast-governed transient) or after the reference has caught up (absolute-calibrated steady state)?

T3: One mechanism at two signs versus an asymmetry it cannot derive (elation and depression are not mirror images). The compression that makes the effect elegant is that elation and depression are a single discrepancy mechanism read at positive and negative sign — overshoot up, overshoot down, from one rule. But the mechanism is not sign-symmetric in its consequences: negative contrast typically runs larger and more persistent than positive contrast, a loss against expectation weighing heavier than an equal gain. That asymmetry is folded in as an observed property rather than derived from the signed-discrepancy rule, so the clean "one mechanism, two signs" picture must carry an extra directional fact it does not explain from within. The tension is that the unification is real yet incomplete — the same rule generates both effects but not their unequal magnitudes, which point past the effect toward loss aversion. Diagnostic: Is the prediction resting only on the sign and size of the discrepancy, or does it also require the loss-side asymmetry that the bare signed-discrepancy rule does not itself produce?

T4: The overshoot signature versus its mechanism (a curve shared with systems that have no reference point). A response that overshoots after a step change is the effect's visible fingerprint — but the identical curve is produced by underdamped physical systems (a sprung mass, an RLC circuit, a control loop) governed by inductance, momentum, or reactance, with no reward, no expectation, and no reference point anywhere in them. So the observable signature underdetermines the mechanism: seeing an overshoot after a shift is not sufficient to diagnose a Crespi effect, and the motivational machinery (a represented reward, a recalibrating expectation) must be independently established. The tension is that the effect is defined by a mechanism but recognized by a shape, and the shape is promiscuous — shared with dynamical look-alikes that instantiate none of what makes the effect what it is. Diagnostic: Does the overshooting system actually represent reward and form a recalibrating expectation, or is it a physical step-response look-alike producing the same curve by a different mechanism?

T5: Autonomy versus reduction (an operant effect or an instance of reference-point contrast). "Crespi effect" carries home-domain cargo — the runway paradigm, the two-group elation/depression demonstration, the motivational reading of the overshoot — and across reward-representing systems (animal learning, workplace pay studies, consumer pricing) it transfers as mechanism, only the reward currency changing. But the deep structure it exposes — response calibrated to a reference point, with contrast-based overshoot around a shift — is the same object prospect theory formalized, with the elation/depression asymmetry the behavioral ancestor of loss aversion, and that object already lives in the catalog as reference_point and the broader contrast pattern. The cross-domain lesson ("expect contrast, not absolute-level effects, around any change point in a history-dependent system") belongs under those headings. The tension is between a named operant effect whose paradigm earns its own study and a reference-point/contrast parent that carries the portable insight. Diagnostic: Resolve toward reference_point / contrast when the lesson is the general history-dependence of response around a change point; toward the named Crespi effect when the operant paradigm, reward representation, and elation/depression demonstration are doing the work.

Structural–Framed Character

The Crespi effect sits toward the structural end of the spectrum but stops short of the pole — best read as mixed-structural: a genuine, evaluatively neutral behavioral mechanism wearing reward-learning vocabulary, closely analogous to how isostasy is placed and to the Baldwin-effect character of a real, recognized-in-nature process. On four of the five criteria its structural credentials are strong. Its evaluative weight is nil: "elation" and "depression" are technical labels for over- and under-shoot, not praise or blame, and a rat running faster or slower than a same-reward control is neither good nor bad — the effect names a value-free discrepancy dynamic, exactly the neutrality of a mechanism that renders no verdict. It is not human-practice-bound: Crespi's rats produced the contrast overshoot in 1942 whether or not anyone theorized it, and the effect runs in any organism that represents reward and forms a recalibrating expectation, so it needs a cognitive-motivational substrate but not a judging observer or a constituting social practice — remove the psychologists and the reference-point transient still fires. Its institutional origin is none: the phenomenon is a fact of reward-driven behavior, discovered rather than invented (the two-group runway design is a human method for detecting it, but the overshoot it isolates is nature's, not the paradigm's). And within its proper range cross-domain reuse is recognition, not import: moving from runway rats to human operant work to workplace pay cuts to consumer discount-then-restore pricing, the same signed-discrepancy mechanism is recognized intact, only the reward currency and response measure changing — and the entry's insistence that a physical step-response look-alike is a false analogy (T4: same curve, no reward, no expectation, no reference point) sharpens exactly this point, since the effect is defined by its mechanism and merely recognized by its shape.

What keeps it off the structural pole is the remaining criterion, vocab-travels, which it fails. Its operative vocabulary is irreducibly reward-learning vocabulary — operant response, reinforcement, positive/negative contrast, elation/depression, reward magnitude, the recalibrating expectation — and none of it floats free of a reward-representing substrate the way a differential equation or "growing quantity" would in a pure prime. Within animal learning, human motivation, and the applied behavioral sciences those terms carry their full content; outside reward-representing systems they have no referent, which is why the physical overshoot can share the curve while instantiating none of the concept. The single portable structural skeleton it does carry — response calibrated to a reference point, with contrast-based overshoot around a shift, amplitude scaling with the signed discrepancy and asymmetric on the loss side — is genuinely substrate-neutral, but it is exactly what the catalog already holds as reference_point (and the broader contrast pattern), and that skeleton is what the Crespi effect instantiates, not what makes "Crespi effect" itself travel: the cross-domain reach (the object prospect theory formalized, the ancestor of loss aversion) belongs to the reference-point/contrast parent, while the runway paradigm, the elation/depression demonstration, and the motivational reading stay home. Its character: a real, evaluatively neutral, recognized-in-nature reference-point-contrast mechanism, structural in skeleton but stated in reward-learning vocabulary that pins it to reward-representing substrates — mixed-structural rather than a free-floating prime.

Structural Core vs. Domain Accent

This section decides why the Crespi effect is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity — so it is worth being exact about what could lift and what stays home.

What is skeletal (could lift toward a cross-domain prime). Strip the reward learning and a thin relational structure survives: response calibrated to a reference point, with contrast-based overshoot around a shift — amplitude scaling with the signed discrepancy and asymmetric on the loss side. The portable pieces are abstract — a recalibrating reference, a perturbation, a signed gap between input and reference, a transient overshoot in the direction of the gap, and a decay as the reference catches up. That skeleton is genuinely substrate-neutral and is already carried by the catalog as reference_point (and the broader contrast pattern); it is the same object prospect theory formalized, with the elation/depression asymmetry the behavioral ancestor of loss aversion. That skeleton is the core the Crespi effect shares, not what makes it the Crespi effect.

What is domain-bound. Almost everything that makes the entry the Crespi effect in particular is reward-learning furniture, and none of it survives extraction. Its subject is an operant response (running, lever-pressing) maintained by a reward; its perturbation is a reward-magnitude shift against a reward history; its signature is positive/negative contrast read as elation/depression against a same-reward control; and its instrument is Crespi's two-group runway paradigm. The decisive test the entry itself supplies: an underdamped physical system that overshoots a step input — an RLC circuit, a sprung mass, a control loop — produces a curve that looks like the Crespi transient but is governed by inductance, momentum, or reactance, with no reward, no expectation, and no reference point anywhere in it; calling that "a Crespi effect" borrows the shape while dropping the motivational mechanism that defines the original. So the effect is defined by its mechanism but recognized only by its shape, and the shape is promiscuous — which pins the named effect to systems that actually represent reward and form expectations. Remove that cognitive-motivational substrate and only the reference-point/contrast parent remains.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same mechanism, not analogy. The Crespi effect's transfer is bimodal. Within reward-driven behavior it travels intact as mechanism — the diagnostics (read a control-relative over- or under-shoot as evidence of a hidden reference), the predictions (direction from the sign of the discrepancy, amplitude from its size, decay from the recalibration rate, extra weight on the loss side), and the vocabulary carry across animal learning, human operant and motivational research, workplace pay studies, and consumer discount-then-restore pricing, only the reward currency and response measure changing. Beyond reward-representing systems the named effect does not travel: the physical step-response look-alike is a false analogy that instantiates none of the mechanism. When the genuine cross-domain lesson — expect contrast, not absolute-level effects, around any change point in a history-dependent system — is needed, it is already carried, in more general form, by reference_point and the broader contrast pattern (the object prospect theory formalized). The cross-domain reach belongs to that parent; "the Crespi effect," as named, carries the runway paradigm, the two-group elation/depression demonstration, and the motivational reading of the overshoot that keep it an operant-learning effect rather than a free-floating prime.

Relationships to Other Abstractions

Current abstraction Crespi Effect Domain-specific

Parents (3) — more general patterns this builds on

• Crespi Effect is a kind of Transient Response Prime

  The Crespi Effect is a Transient Response specialized to a reward-magnitude step whose history-set prediction error drives temporary control-relative over- or under-shoot.

• Crespi Effect is part of Reward Prediction Error Prime

  Reward Prediction Error is a constituent of the Crespi Effect because the signed gap between received and expected reward drives both the transient response and recalibration of the expectation.

• Crespi Effect is a decomposition of Contrast Prime

  The Crespi Effect decomposes to Contrast because behavior is organized by the emphasized difference between delivered reward and its history-set expectation rather than by absolute reward alone.

Hierarchy paths (3) — routes to 2 parentless roots

• Crespi Effect → Transient Response → Temporal Dynamics → Time

Show alternative paths (2)

Not to Be Confused With

• An absolute-value / rate-of-reinforcement account. The picture that response level is keyed to how much reward is delivered now. The Crespi effect shows behavior keyed to the signed discrepancy between delivery and a prior expectation — which is exactly why a shifted animal over- or under-shoots a same-reward control. The absolute account under-predicts precisely around change points. Tell: Does the response depend only on the current magnitude (absolute account), or on the current magnitude relative to what the animal came to expect (Crespi effect)?
• Regression to the mean / measurement artifact. Statistical drift toward an average, or apparatus noise. The elated runner outpacing and the depressed runner lagging controls are real behavioral overshoots produced by a reference mechanism; Crespi's two-group design isolates them by holding current reward fixed and varying only history. Tell: Does the gap disappear when history is controlled (artifact/regression), or is it produced by the history difference with current reward held equal (Crespi effect)?
• Behavioral (simultaneous) contrast. The multiple-schedule phenomenon (Reynolds) in which changing reinforcement on one concurrent component shifts responding on an unchanged component. The Crespi effect is successive incentive contrast — one behavior's reward shifted over time, read against a same-reward control. Simultaneous-across-components versus successive-over-time. Tell: Is the contrast between two concurrently-available components (behavioral contrast), or between before-and-after a magnitude shift in one behavior (Crespi effect)?
• Loss aversion / prospect theory. The descendant, not the same thing: prospect theory formalized reference-dependence and loss aversion in human decision-making decades later. The Crespi effect's elation/depression asymmetry (negative contrast larger and more persistent) is the behavioral ancestor of loss aversion, observed in operant runway behavior. Tell: Is the claim about reference-dependent choice under risk (prospect theory/loss aversion), or about operant response rate transients after a reward shift (Crespi effect)?
• Underdamped step-response overshoot. A physical system (RLC circuit, sprung mass, control loop) that overshoots a step input, producing a curve that looks like the Crespi transient but is governed by inductance, momentum, or reactance — with no reward, no expectation, no reference point. Resemblance of shape only; the defining motivational mechanism is absent. Tell: Does the overshooting system represent reward and form a recalibrating expectation (Crespi effect), or is it a dynamical look-alike producing the same curve by a different mechanism (step-response overshoot)?
• Reference point / contrast (the parent). The substrate-neutral skeleton the Crespi effect instantiates — response calibrated to a recalibrating reference with contrast-based overshoot around a shift, asymmetric on the loss side, carried by reference_point and the broader contrast pattern. This is the object prospect theory formalized and what carries the cross-domain lesson. Tell: Is the substrate a reward-representing operant system with the runway/elation-depression apparatus (Crespi effect), or any history-dependent system showing contrast around a change point (reference_point/contrast, which carries the cross-domain lesson)?

Neighborhood in Abstraction Space

Crespi Effect sits in a moderately populated region (46^th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Attention, Memory & Automaticity (13 abstractions)

Nearest neighbors

• Near-Miss Effect — 0.86
• Deep Engagement — 0.85
• Hedonic Treadmill — 0.84
• Retrieval-Induced Forgetting — 0.84
• Task-Switching Cost — 0.84

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