Benchmark list

1 Acquisition

1.1 Acquisition

Description: After a number of CS–US pairings, the CS elicits a conditioned response (CR) that corresponds to a consistent increase or decrease in magnitude or frequency of an observable behaviour.

Grade: AA

Domains: universal

1.2 Benefit of spatiotemporal contiguity

Description: CRs are stronger when CS and US are presented close in space and time.

Grade: AA

Domains: universal

1.3 Interstimulus interval (ISI) effects

Description: Strongest CR with relatively short ISIs.

Grade: AA

Domains: universal

1.4 Second-order conditioning

Description: When A–US pairings are followed by B–A pairings, presentation of B generates a CR.

Grade: AA

Domains: universal

1.5 Recency effect

Description: After acquisition, CRs are stronger when the test is closer in time to acquisition.

Grade: AA

Domains: universal

1.6 Post-training US revaluation

Description: After CS-US pairings, revaluation of the US affects the response to the CS.

Grade: AA

Domains: universal

1.7 Timing of the CR

Description: Peak of the CR tends to occur near the end of relatively long ISIs.

Grade: A

Domains: Widely observed but not universal, and boundary conditions unclear

1.8 Negatively accelerated learning curves

Description: Group-averaged CRs tend towards an asymptote during acquisition.

Grade: A

Domains: Widely observed, but deviations occur, with unclear boundary conditions.

1.9 Intertrial interval (ITI) and trial spacing effects

Description: CRs are acquired faster and/or are stronger after acquisition when trials with the same CS are further apart in time during training.

Grade: A

Domains: Widely observed across species, but not universal and with unclear boundary conditions.

1.10 Scalar invariance in response timing

Description: The variability of CR timing scales with mean CR timing.

Grade: A

Domains: Eyeblink conditioning with rabbits and ring doves; reward conditioning, with rats and pigeons; conditioned suppression and instrumental (Sidman) avoidance conditioning with rats; aversive conditioning in gold fish.

1.11 Sensory preconditioning

Description: When B–A pairings are followed by A–US pairings, presentation of B generates a CR.

Grade: A

Domains: Domain- and species-general in Pavlovian conditioning but with circumscribed exceptions.

1.12 Pavlovian-instrumental transfer

Description: After Pavlovian acquisition, the presence of the CS influences instrumental behaviour to obtain/avoid the same US1 (outcome-specific PIT) or a different US2 (outcome-general PIT).

Grade: A

Domains: Appetitive and aversive Pavlovian conditioning.

1.13 Summation

Description: When two CSs independently trained with the same US are tested in combination, CRs are often greater than those exhibited to either CS individually.

Grade: A

Domains: Pavlovian conditioning.

1.14 Conditioned diminution and facilitation of the unconditioned response

Description: After acquisition, the response to the US in a CS-US presentation is diminished or facilitated compared to the US not preceded by CS, depending on the setup and species.

Grade: B

Domains: Human aversive classical conditioning (diminution), rabbit eyeblink conditioning (facilitation).

1.15 Imperfect predictors effect

Description: An association CS-US2 is learned more rapidly after training CS-US1 with partial reinforcement rather than full reinforcement.

Grade: B

Domains: Domain-general in Pavlovian conditioning  across non-human species with unknown boundary conditions.

1.16 Learned predictiveness effect

Description: After training with compound CS such that one component A always predicts US1 while another component B is sometimes coupled with US1 and sometimes with US2, the association A-US3 will be more rapidly learned than B-US3.

Grade: B

Domains: Mainly human contingency learning, some non-human setups with unknown boundary conditions.

1.17 More intense CSs facilitate acquisition

Description: When CSs are further away from a perceptual baseline (usually more intense), acquisition is faster or the CR asymptote is higher.

Grade: B

Domains: Aversive Pavlovian conditioning in several species (rats, rabbits, and humans).

1.18 Backward conditioning elicits CRs

Description: NA

Grade: B

Domains: Mainly rat conditioned suppression, occasionally other preparations in rats, rabbit eyeblink conditioning, human autonomic conditioning.

1.19 Acquisition is context-dependent

Description: A CR is stronger when tested in the same context as opposed to a different context from acquisition.

Grade: B

Domains: Human predictive learning, Pavlovian conditioning in rats.

1.20 Cue-to-consequence effect

Description: Some combinations of CS and US elicit stronger CR than others.

Grade: B

Domains: Pavlovian conditioning, mainly involving nausea US in rats.

1.21 Benefit of retrieval practice

Description: Repeated retrievals increase the CR.

Grade: B

Domains: Mainly human predictive learning, rat spatial (instrumental) learning.

1.22 Partial reinforcement acquisition effect

Description: Acquisition is slower or the CR asymptote is lower during partial rather than continuous reinforcement.

Grade: C

Domains: Pavlovian conditioning across species, with a substantial proportion of negative and some opposing reports.

1.23 CS-specific CR

Description: The pattern of the CR depends on the CS.

Grade: C

Domains: Specific appetitive and aversive conditioning procedures in rats.

1.24 Parameter-specific CR

Description: The pattern of a CR depends on the parameters of the conditioning schedule.

Grade: C

Domains: Systematically studied in several relatively narrow Pavlovian conditioning procedures.

1.25 Inverse base-rate effect

Description: When a compound cue AB is paired with US1 more frequently than compound cue AC is with US2, the novel compound cue BC elicits the CR associated with US2.

Grade: C

Domains: Human predictive learning.

2 Generalisation

2.1 Generalisation

Description: A novel cue elicits a CR to the degree that it shares some characteristics with a CS that has been paired with the US.

Grade: AA

Domains: universal

2.2 Peak & area shift

Description: After intradimensional discrimination training, generalisation gradients increase beyond the CS+ in the opposite direction from the CS-.

Grade: AA

Domains: universal

2.3 Asymmetry in generalisation

Description: Generalisation of inhibition is broader than generalisation of excitation.

Grade: A

Domains: Mainly appetitive instrumental conditioning.

2.4 Prototype and exemplar effects in category learning

Description: After training with a set of related CSs, each independently presented with the same US, responding is stronger to trained exemplars than to untrained stimuli of the same category, and stronger to a novel stimulus that represents the average of the trained CS set (the prototype) than to exemplars.

Grade: A

Domains: Mainly appetitive instrumental  conditioning.

2.5 External inhibition

Description: Adding a novel cue B to a trained cue A results in less responding than the trained A by itself.

Grade: C

Domains: Widely observed in Pavlovian and instrumental conditioning but often as an incidental finding without statistical tests, and there are a considerable number of null results with unknown boundary conditions.

3 Discrimination

3.1 Differential/discriminative training

Description: After receiving training in which one CS is followed by a US (CS+) and another CS is not followed by a US (CS-), the difference in responding to the two stimuli is greater than if only CS+ were trained and CS- first encountered at test.

Grade: AA

Domains: universal

3.2 Positive patterning

Description: After training with intermixed presentations of a compound of two stimuli followed by an outcome (AB+) and each of the individual stimuli followed by no outcome (A– / B–), responding to the compound is stronger than responding to the individual stimuli.

Grade: AA

Domains: universal

3.3 Asymmetry in intensity/magnitude discrimination

Description: When a discrimination (A+, B-) involves stimuli that differ in intensity or magnitude, acquisition is faster when A+ is more intense or greater in magnitude.

Grade: AA

Domains: universal

3.4 Negative patterning

Description: After training with intermixed presentations of two individual stimuli followed by an outcome (A+, B+) and a compound of those two stimuli followed by no outcome (AB–), responding to the individual stimuli is stronger than responding to the compound.

Grade: A

Domains: Widely observed but unclear impact of procedural variations.

3.5 Positive patterning is easier than negative patterning

Description: Given a discrimination in which the compound of two stimuli (AB) leads to a different consequence than the two stimuli individually (A / B), the learner acquires the discrimination faster when the compound predicts the outcome (positive patterning) than when the compound predicts the absence of the outcome (negative patterning).

Grade: A

Domains: Widely observed but unclear impact of procedural variations.

3.6 Biconditional discrimination

Description: After training with intermixed presentations of stimulus compounds in which no single stimulus signals the presence or absence of the outcome but each compound predicts the presence or absence of the outcome (AB+ / BC– / CD+ / AD–), responding is higher for the compounds that predict the outcome.

Grade: A

Domains: Widely observed but unclear impact of procedural variations.

3.7 Biconditional is harder than component discrimination

Description: Discriminative responding between the outcome-present and outcome-absent trials of a biconditional discrimination (AB+ / BC– / CD+ / AD–) is slower than discriminative responding between the outcome-present and outcome-absent trials of a component discrimination that is matched for complexity but in which single stimuli signal the presence or absence of the outcome (AB+ / BC– / BC– / AD+).

Grade: A

Domains: Appetitive Pavlovian and instrumental conditioning, human causal learning.

3.8 Simultaneous feature-positive discrimination

Description: Simultaneous presentations of two stimuli paired with a US (AX+), intermixed with presentations of one stimulus with no US (X–), result in stronger responding to the compound (AX) than to the single stimulus (X) alone, and strong responding in the presence of the other stimulus (A).

Grade: A

Domains: Observed in multiple procedures across species, but not universal. See the closely related feature discrimination phenomena 3.9-3.11 and 3.16.

3.9 Serial feature-positive discrimination

Description: Serial presentations of two stimuli paired with the US (A.

Grade: A

Domains: Appetitive Pavlovian conditioning in pigeons and rats, appetitive instrumental conditioning in rats, rabbit eyeblink conditioning, human conditioned suppression. See the closely related feature discrimination. phenomena 3.8, and 3.10-3.12.

3.10 Simultaneous feature-negative discrimination

Description: Presentations of one stimulus paired with a US (X+) intermixed with simultaneous presentations of two stimuli without the US (AX–), result in stronger responding to the single stimulus (X) alone than to the compound (AX), while the other stimulus (A) acquires the ability to suppress responding when combined with another stimulus that has been paired with the outcome.

Grade: A

Domains: Observed in multiple procedures across species, but not universal. See the closely related feature discrimination phenomena 3.9-3.11 and 3.16.

3.11 Serial feature-negative discrimination

Description: Presentations of one stimulus paired with a US (X+) intermixed with simultaneous presentations of two stimuli without the US (AX–), result in stronger responding to the single stimulus (X) alone than to the compound (AX), while the other stimulus (A) acquires the ability to suppress responding when combined with another stimulus that has been paired with the outcome.

Grade: A

Domains: Primarily appetitive conditioning. See the closely related feature discrimination phenomena 3.8-3.10 and 3.12.

3.12 Resistance of occasion setting to extinction and counterconditioning

Description: After serial feature-positive or feature-negative discrimination, the ability of the feature to disambiguate the target is resistant to extinction and counterconditioning.

Grade: A

Domains: Appetitive and aversive Pavlovian conditioning in rats. See the closely related feature discrimination phenomena 3.9 and 3.11.

3.13 Interference by a common cue in negative patterning

Description: When training with intermixed presentations of two individual stimuli followed by a US (A+, B+) and a compound of those two stimuli followed by no US (AB–), the discrimination is acquired slower when a common stimulus is added to all trials (AX+ / BX+ / ABX–).

Grade: B

Domains: Pavlovian and instrumental conditioning in rats, pigeons, bees, humans, usually using within-modality stimuli; but opposing findings with unclear boundary conditions.

3.14 Biconditional discriminations are harder than negative patterning

Description: Discriminative responding between the US-present and US-absent trials of a biconditional discrimination (AB+ / BC- / CD+ / AD-) is more difficult than a discrimination between the US-present and US-absent trials of a negative patterning discrimination (A+ /B+/ AB-).

Grade: B

Domains: rats appetitive Pavlovian  conditioning and human causal learning, boundary conditions include the presence of a common cue and the CS duration.

3.15 Intra/extradimensional shift effects

Description: After discrimination training with CSs that differ on two perceptual dimensions, one CS predicting the US and another not, a subsequent discrimination with different CSs is facilitated if the previously predictive dimension continues to be predictive and the other one not, rather than the other way around.

Grade: B

Domains: Many species but only demonstrated in narrow experimental procedures mainly involving instrumental learning or pigeon autoshaping.

3.16 Transfer along a continuum (easy-to-hard effect)

Description: Discrimination training with two CSs that are highly discriminable facilitates subsequent discrimination training with CSs that are more similar to each other.

Grade: B

Domains: Instrumental conditioning across species, eyeblink conditioning in rabbits; contradictory results in conditioned taste aversion and unclear boundary conditions.

3.17 Feature-positive discriminations are easier than feature-negative discriminations

Description: Given a discrimination in which the compound of two stimuli (AX) leads to a different outcome than one of the stimuli presented individually (X), discriminative responding develops faster when the compound predicts the US (feature positive) than when the compound predicts the absence of the US (feature negative).

Grade: C

Domains: Mainly rat conditioned suppression, pigeon and human instrumental conditioning; unclear boundary conditions, and opposing results in humans.

3.18 Feature ambiguous occasion setting

Description: In serial discriminations, one stimulus (X) can be trained to concurrently serve as the feature in both a feature-negative discrimination (A+ / X…A–) and a feature-positive discrimination (B– / X…B+) when combined with different stimuli (A / B).

Grade: C

Domains: Rat appetitive (Pavlovian and instrumental) conditioning, only few demonstrations.

4 Extinction

4.1 Extinction

Description: The CR decreases when CS–US pairings are followed by presentations of the CS alone or by unpaired CS and US presentations.

Grade: AA

Domains: universal

4.2 Counterconditioning

Description: CS–US training with an aversive US diminishes an appetitive CR otherwise produced by prior CS–US training with an appetitive US (and conversely).

Grade: AA

Domains: universal

4.3 ABA renewal

Description: If CS-US acquisition takes place in context A and extinction in context B, then testing the CS in context A produces a stronger CR than in context B.

Grade: AA

Domains: universal

4.4 ABC renewal

Description: If CS-US acquisition takes place in context A and extinction in context B, then testing in novel context C produces a stronger CR to the CS than in context B.

Grade: AA

Domains: universal

4.5 Protection from extinction

Description: After pairing stimulus A with US, and conditioned inhibition training with stimulus B, presenting AB-noUS compounds leads to a higher CR to A alone, compared to conventional extinction (A-noUS presentations).

Grade: A

Domains: Pavlovian aversive and appetitive conditioning in rats, pigeons, humans.

4.6 Partial-reinforcement extinction effect

Description: Extinction is slower following partial rather than continuous reinforcement.

Grade: A

Domains: Domain-general with short ITI and between-subjects tests.

4.7 Contingency degradation

Description: Interspersing unsignalled US presentations during CS-US acquisition reduces the CR.

Grade: A

Domains: Domain-general, across several species including rats, mice, pigeons, and humans; but not demonstrated universally.

4.8 Spontaneous recovery

Description: Presentation of the CS some time after successful extinction partially restores the CR.

Grade: A

Domains: Domain-general in rodents and humans, but not universal.

4.9 Reinstatement

Description: After extinction, presentation of the US in the same context partially restores responding.

Grade: B

Domains: Domain-general across species with a proportion of weak results and partly unknown boundary conditions.

4.10 Rapid reacquisition

Description: After CS-US acquisition and CS-noUS extinction, further CS–US presentations result in re-acquisition of the CR in fewer trials than during initial acquisition.

Grade: B

Domains: Across species and domains, but not in all procedures and with unclear boundary conditions.

4.11 Recovery from counterconditioning

Description: After counterconditioning, the initial CR is partially recovered by the passage of time (spontaneous recovery), unsignalled US presentations (reinstatement), or change of context (renewal).

Grade: B

Domains: Across Pavlovian conditioning domains in humans; limited evidence from other species or instrumental conditioning.

4.12 Attenuation of renewal by multiple context extinction

Description: Conducting extinction in multiple contexts, rather than a single context, reduces the CR when testing occurs in the original acquisition context or an entirely novel context.

Grade: B

Domains: Rat and human fear conditioning.

4.13 Attenuation of renewal by retrieval cues

Description: After interspersing a specific stimulus during extinction training, the presence of this stimulus reduces the CR when tested in the original acquisition context or a novel context.

Grade: B

Domains: Rat magazine approach conditioning, human fear conditioning, only few observations but with no opposing results.

4.14 AAB renewal

Description: If both CS-US acquisition and extinction take place in context A, then testing in context B produces a higher CR to the CS than in context A.

Grade: B

Domains: Across species and Pavlvovian as well as instrumental procedures but, but robustness variable.

4.15 Concurrent recovery

Description: After A-US acquisition and A-noUS extinction, B-US acquisition training leads to increased CR to A.

Grade: C

Domains: Rabbit eyeblink conditioning; only replicated within a single laboratory.

4.16 Retrieval-extinction effect

Description: Following conditioning with a CS, extinction following retrieval of that CS alone reduces reinstatement, spontaneous recovery, and renewal, compared to standard extinction.

Grade: C

Domains: Appetitive Pavlovian conditioning in rodents, some evidence in aversive Pavlovian conditioning in rodents and humans, but there are considerable null and opposing findings without clear knowledge of the boundary conditions.

4.17 Secondary extinction

Description: After conditioning with A-US and B-US, extinguishing A also reduces CR to B.

Grade: C

Domains: Mainly rat fear conditioning with intermixed CSs.

4.18 Mediated extinction

Description: If two CSs have been presented together, non-reinforced presentation of one of them yields a reduction in responding to the other.

Grade: C

Domains: Widely observed but with highly variable results and unclear boundary conditions.

5 Inhibitory conditioning

5.1 Conditioned inhibition/Simultaneous feature-negative discrimination

Description: After interspersed AB-noUS presentations and B-US pairings, A reduces the CR when paired with a US-predictive C (summation test), and when A is paired with the US, acquisition of a CR is retarded (retardation test).

Grade: A

Domains: Multiple Pavlovian conditioning domains, but not universal.

5.2 Conditioned inhibition/Negative contingency training

Description: After interspersed A-noUS and unsignalled US presentations, A reduces the CR when paired with a US-predictive B (summation test), and when A is paired with the US, acquisition of a CR is retarded (retardation test).

Grade: A

Domains: Multiple Pavlovian conditioning domains, but small effect size suggests heterogeneity between individuals.

5.3 Conditioned inhibition/Differential conditioning

Description: After interspersed A-noUS presentations and B-US pairings, A reduces the CR when paired with a US-predictive C (summation test), and when A is paired with the US, acquisition of the CR is retarded (retardation test).

Grade: B

Domains: Eyeblink conditioning and lick suppression in rats; autonomic conditioning in humans.

5.4 Extinction resistance of conditioned inhibition

Description: After interspersed A-noUS presentations and B-US pairings, A reduces the CR when paired with a US-predictive C (summation test), and when A is paired with the US, acquisition of the CR is retarded (retardation test).

Grade: B

Domains: Conditioned suppression in rats, human predictive learning.

5.5 Retrospective revaluation of conditioned inhibition

Description: After interspersed AB-noUS presentations and B-US pairings, further B-US presentations increase, and B-noUS presentations reduce, the conditioned inhibitory properties of A in summation and retardation tests.

Grade: C

Domains: Conditioned suppression in rats.

6 Stimulus competition/potentiation

6.1 Overshadowing

Description: Conditioning to the compound AB results in a weaker CR to B than is attained with B–US pairings.

Grade: A

Domains: Several Pavlovian conditioning tasks across domains and species with well-known boundary conditions, but the opposite effect (potentiation) is sometimes observed for reasons not yet fully characterized.

6.2 Forward blocking

Description: Conditioning to a compound AB results in weaker conditioning to B when the former is preceded by conditioning to A than when it is not.

Grade: A

Domains: Frequently observed across species and conditioning tasks with well-known boundary conditions, but the opposite effect (augmentation) is sometimes observed for reasons not yet fully identified.

6.3 Superconditioning

Description: Reinforced AB presentations following inhibitory conditioning of A increase the CR to B as compared with when B is trained in the absence of A.

Grade: A

Domains: Varied Pavlovian conditioning  procedures and species, but a limited number of demonstrations, and the potential boundary conditions are not known.

6.4 Relative validity

Description: The CR to CS A is weaker when training consists of AB-US pairings alternated with AC-noUS trials than when it consists of AB trials alternated with AC trials, with each type reinforced half of the time.

Grade: B

Domains: Conditioned suppression, taste aversion, magazine-approach conditioning (all rats); eyeblink conditioning (rabbits); autoshaping (pigeons) with no systematic investigation of procedural variation.

6.5 Overexpectation

Description: Reinforced AB presentations following independent reinforced presentations of A and B result in a decrement in the CR to A and B separately.

Grade: B

Domains: Multiple procedures in multiple species, but only few demonstrations and negative results.

6.6 Unblocking by increasing (or decreasing) the US

Description: Increasing (or decreasing) the US during Phase 2 AB training increases responding to the blocked B.

Grade: B

Domains: Conditioned suppression and appetitive Pavlovian conditioning in rats, autoshaping in pigeons.

6.7 Backward blocking

Description: When AB-US pairings in Phase 1 are followed by A-US pairings in Phase 2, the CR to B is weaker than without the A-US pairings.

Grade: B

Domains: Conditioned suppression in rats, contingency judgements in humans.

6.8 Unequal changes in responding by elements as a result of compound pairings (compound conditioning)

Description: Unequal changes in CR to two CSs trained in compound occur when the CSs start with different response potentials.

Grade: B

Domains: Conditioned suppression, magazine approach, instrumental learning in rats and pigeons, but only few demonstrations.

6.9 Recovery from overshadowing

Description: Extinction of the overshadowing cue results in increased responding to the overshadowed cue.

Grade: B

Domains: Conditioned suppression in rats; human causal judgement.

6.10 Overshadowing is stronger than external inhibition

Description: Adding a CS to a trained compound of two CS results in a smaller decrease in CR than does removing a CS from the same compound.

Grade: B

Domains: Rabbit eyeblink conditioning, rat fear conditioning, human predictive learning.

6.11 Redundancy effect

Description: Stronger responding to the blocked stimulus B after A-US/AB-US pairings than to an irrelevant stimulus B after AB-US/BC-noUS pairings.

Grade: B

Domains: Magazine approach in rats, predictive learning in humans.

6.12 Recovery from forward blocking

Description: Several different post-blocking manipulations result in increased responding to the blocked cue.

Grade: C

Domains: Conditioned suppression in rats.

6.13 Recovery from backward blocking

Description: Extinction of the blocker A results in increased responding to the blocked B.

Grade: C

Domains: Conditioned suppression in rats.

7 CS/US preexposure effects

7.1 US pre-exposure effect

Description: Presentation of the US in a training context prior to CS–US pairings retards the acquisition of the CR to the CS.

Grade: AA

Domains: universal

7.2 CS pre-exposure effect (latent inhibition)

Description: Pre-exposure to a CS followed by CS–US pairings retards the acquisition of the CR.

Grade: A

Domains: Fairly general in several species, but inconsistent or narrower in humans, pigeons, fish, and invertebrates.

7.3 A change of context disrupts latent inhibition

Description: NA

Grade: A

Domains: Domain-general in Pavlovian  conditioning with partly unknown boundary conditions.

7.4 Pre-exposure to a discriminant CS facilitates learning (perceptual learning)

Description: Exposure to similar stimuli, CS1 and CS2, leads to faster subsequent acquisition of a discrimination between them.

Grade: A

Domains: Instrumental conditioning in rats and birds, conditioned taste aversion and appetitive Pavlovian conditioning in rats, evaluative conditioning and predictive learning.

7.5 Context pre-exposure facilitates aversive CRs to the context

Description: NA

Grade: B

Domains: Rat contextual conditioning with shock or nausea US.

7.6 Hall–Pearce effect

Description: Training CS–weak US leads to slower acquisition of CS–strong US.

Grade: B

Domains: Aversive Pavlovian conditioning in rats.

7.7 Learned irrelevance

Description: Random interspersed presentations of the CS alone and the US alone retard conditioning even more than combined latent inhibition and US pre-exposure.

Grade: B

Domains: Robust across species, but with a limited range of Pavlovian  procedures.

7.8 Following CS-pre-exposure and conditioning, a long relative to a short delay before testing decreases CRs (super latent inhibition)

Description: NA

Grade: C

Domains: Rat conditioned taste aversion and conditioned suppression, human predictive learning.

7.9 Presentation of a different CS before conditioning disrupts latent inhibition

Description: NA

Grade: C

Domains: Conditioned suppression and conditioned taste aversion in rats; contingency/predictive judgments in humans.

7.10 Recovery from latent inhibition (LI) induced by context extinction

Description: LI is attenuated by extensive exposure to the training context between CS pre-exposure and the CS–US pairings or between the CS-US pairings and testing.

Grade: C

Domains: Conditioned suppression and conditioned taste aversion in rats; predictive learning in humans; robustness unclear.