Within-compound Associations Research Articles

Retrospective revaluation effects during interpersonal attributions

Three questionnaire studies examined retrospective revaluation during judgments of causal attribution in social situations, that is, judgments about whether a particular individual or environmental cause was responsible for an outcome. An initial phase of training occurred in which two individuals (or one person and an environmental event) were described as possible candidates as causes of an outcome. One cause was the target test cause, and the other cause was manipulated in Phase 2. In Phase 2, new information about the validity of the “nontarget” cause was presented. This “nontarget” cause was either elevated in its causal status (an “inflation” treatment) or decreased in status as a possible cause (a “deflation” treatment). Participants rated the extent that the target cause was the cause of the outcome before and after receiving the inflation or deflation information in Phase 2. The difference in those scores was calculated to examine the potential of the Phase 2 information to influence attributions regarding the target cause (retrospective revaluation). Retrospective revaluation effects were found, specifically in deflation conditions, such that this new Phase 2 information did influence judgments about causal status of the target cause (Study 1); but inflation effects were weak. Deflation occurred when the cause was environmental and when it involved two people as causes. Also, the importance of the strength of the association between two causes (“within-compound associations”) was evaluated in two subsequent studies, revealing that this factor is important for observing retrospective revaluation, particularly when the strength of the association is either weak or strong but not of medium strength.

Aging and associative binding in contingency learning

ABSTRACT Cue competition effects are pervasive in young adults’ learning, but evidence for these effects in older adults’ learning is mixed. For example, although older adults show strong forward blocking, they do not show recovery from overshadowing. We examined whether this could be due to problems with associative binding using a rapid, streamed trial contingency learning task to minimize long-term memory retrieval demands. In a forward blocking paradigm , target cues gained less predictive value when the competing companion cues had high predictive value and this forward blocking effect was similar for younger and older adults. In a backward blocking paradigm, target cues lost more predictive value when the competing companion cues had high predictive value, but this backward blocking effect was greater for younger than older adults. These findings, together with evidence that within-compound associations for companion and target cues mediate backward, but not forward cue competition effects, suggest that a decline in associative binding may be responsible for the absence of backward cue competition effects in older adults’ contingency learning .

Can an augmented flavor augment another flavor? Evidence from second-order augmentation

In 3 experiments with rat subjects, we explored flavor-aversion learning across multiple conditioning stages. Experiment 1 was the first demonstration of taste-taste augmentation: Saccharin (SAC) preconditioning followed by compound conditioning of SAC + saline produced a significantly stronger saline aversion. Notably, taste-taste augmentation was recorded both when the 2 tastes were mixed together or presented separately during compound conditioning. Experiments 2A and 2B involved 3 conditioning phases to determine if an augmented flavor could subsequently augment a third flavor. Specifically, after Flavor 1 preconditioning, Flavors 1 and 2 were conditioned in compound, and then, Flavors 2 and 3 were conditioned in compound. This conditioning regimen yielded a significantly stronger aversion to Flavor 3, and this stronger aversion was not attributable to direct generalization from either Flavor 1 or Flavor 2. It appears that a given flavor can have multiple within-compound associations, and these associations can summate to produce a stronger conditioned response to a third taste.

Over-selectivity decreases with increased training: A role for within-compound associations

Over-selectivity occurs when one element of a complex-stimulus controls behavior at the expense of other equally elements of that stimulus; a phenomenon common in populations subject to cognitive challenge. However, lack of theoretically-based analysis, may have hindered understanding and remediation of the practically-important over-selectivity phenomena. Current studies examined whether associative theories applied to overshadowing, a similar phenomenon in the context of conditioning experiments, could be applied to over-selectivity effects to open theoretical analysis of over-selectivity. Three experiments investigated whether length of training impacts over-selectivity in the same way as overshadowing, which has theoretical implications for understanding that latter phenomenon. All studies employed variants of a judgment procedure in which participants had to judge the relationship between a predictor and an outcome, and the predictors were presented either on their own, or in compound with another predictor. In all studies, the elemental cue (A) was rated similarly to one of the components of the compound (B), but higher than the other component (C). The difference in the extent to which the components of the compound (B and C) were judged as predictors became smaller as levels of training increased, which is an effect that is also seen in discrimination learning studies of over-selectivity. Moreover, it was apparent that as the strength of the within-compound association increased, the level of over-selectivity decreased. These results are similar to those seen for overshadowing, and are discussed with respect to the possible associative mechanisms controlling over-selectivity.

Associative structure of conditioned inhibition produced by inhibitory perceptual learning treatment.

Exposure to a set of complex stimuli yields an enhanced ability to discriminate between these stimuli. In previous experimental studies, two distinguishable stimuli, X and A, were each repeatedly paired with a common Stimulus B to create compound Stimuli XB and AB. Prior evidence suggests that unique Features X and A form mutually inhibitory associations. This was evidenced by pairing Feature A with a biologically relevant stimulus (i.e., an unconditioned stimulus [US]) and observing that Stimulus X alone later serves to inhibit anticipatory behaviors for that US. These observations may reflect the mutually inhibitory nature of the two Features X and A. However, by assessing the influence of X on behavior that anticipates the US rather than Feature A, these experiments tested inhibition only indirectly. In the present experiments, a more direct measure of inhibition is proposed and tested with rats. We found evidence of retardation and negative summation of associations between unique Features X and A in their capacity to serve as competing cues during overshadowing treatments. Stimulus X was less susceptible to overshadowing by A (which is indicative of retardation of the establishment of an X-A within-compound association) and was able to suppress overshadowing by A of another stimulus (Y) when X was presented with Y at test (which is indicative of negative summation of the representation of A by X). Thus, XB/AB trials were seen to establish an inhibitory relationship between X and A.

The effect of stimulus duration on over-selectivity: Evidence for the role of within-compound associations.

The phenomenon whereby behavior becomes controlled by one aspect of the environment at the expense of other aspects of the environment (stimulus overselectivity) is widespread across many intellectual and developmental disabilities. However, the theoretical mechanisms underpinning overselectivity are not understood. Given similarities between overselectivity and overshadowing, exploring overselectivity using associative learning paradigms might allow better theoretical understanding of the phenomenon. Three experiments explored overselectivity using a simultaneous discrimination task with typically developing participants undergoing a cognitively demanding task. Experiment 1 investigated whether stimulus duration effects found within the overshadowing literature also occurred in an overselectivity paradigm, and demonstrated that greater overselectivity was observed when stimuli were presented for short durations (2s and 5s) compared with longer durations (10s). Experiment 2 demonstrated that a posttraining revaluation procedure resulted in retrospective revaluation for stimuli presented at shorter durations (2s) and mediated extinction for stimuli presented at longer durations (10s). Such results replicate findings from the overshadowing literature that have been interpreted in terms of within-compound associations while also supporting assumptions made by an extended comparator hypothesis. Experiment 3 uses an additional control condition to further demonstrate that the retrospective revaluation is a genuine revaluation effect. Additionally, the experiment provides further evidence for the within-compound association explanation of the results through manipulating the consistency with which elements of a compound were paired during training. Taken together, the findings highlight the necessity to consider the role of within-compound associations in overselectivity, allowing for a better understanding of overselectivity effects. (PsycINFO Database Record

Effects of extinction of a nontarget CS on performance to a target CS

When a target conditioned stimulus (CS A) is paired with an unconditioned stimulus in the presence of a second, conditioned stimulus (CS B) during compound conditioning trials, the associative strength of CS B can influence the magnitude of the conditioned response (CR) to CS A. For example, extinction of the competing, nontarget CS B can influence the CR to CS A. An enhancement of the CR to the target CS A due to extinction of the nontarget CS B after compound conditioning is sometimes referred to as “recovery from overshadowing” – a type of retrospective revaluation. The present experiments examined retrospective revaluation effects using a conditioned taste aversion procedure. The experiments obtained an effect on the CR to CS A following extinction of CS B. The results are discussed with respect to the comparator hypothesis, within-compound associations, and retrieval as well as other relationships between the target CS and nontarget CS.

Augmentation’s boundary conditions? Investigation of spatial contiguity, temporal contiguity, and target flavor familiarity

When a preconditioned flavor (A) is conditioned in compound with a novel target flavor (X), the aversion to the target X is increased; this enhanced aversion to X is called augmentation. In 6 experiments with rat subjects, we manipulated the spatial contiguity of cues during compound conditioning (AX+), the temporal contiguity of cues during compound conditioning (AX+), and the familiarity of the target. In all 6 studies, augmentation was recorded with spatially separated flavors. In Experiments 2–4, augmentation was not detected if the two flavors were temporally discontiguous, but augmentation still occurred if the cues were partially contiguous (i.e., the flavors co-occurred for 2 min of a 4-min exposure). Even though stimulus preexposure can often weaken subsequent conditioning, augmentation was observed following 1 or 4 preexposures to the target taste (Experiment 5A) or target odor (Experiment 5B). In sum, manipulations that should weaken, but not eliminate, the within-compound association formed during AX+ conditioning did not prevent augmentation, suggesting the robustness of the within-compound association when one of the elements is a preconditioned flavor.

Extinction of the initial within-compound association established in a blocked preexposure to two compound flavours

Abstract Extinction of the A↔X association after blocked preexposure to AX-BX was studied in two experiments. In Experiment 1, two groups of rats received long (14 trials) or short (4 trials) blocked preexposure to AX-BX and subsequent conditioning of X. The results showed that the AX association was equally preserved after long and short preexposure. Experiment 2 studied the effect of blocked preexposure to 0, 1 or 2 ruptures of the AX association on extinction. In this experiment a “rupture” is produced when, in subsequent blocks, one element of the original compound is presented in compound with a different element. A significant extinction was observed only when the AX association was broken twice

The sometimes competing retrieval and Van Hamme & Wasserman models predict the selective role of within-compound associations in retrospective revaluation

The correlation between blocking and within-compound memory is stronger when compound training occurs before elemental training (i.e., backward blocking) than when the phases are reversed (i.e., forward blocking; Melchers et al., 2004, 2006). This trial order effect is often interpreted as problematic for performance-focused models that assume a critical role for within-compound associations in both retrospective revaluation and traditional cue competition. The present manuscript revisits this issue using a computational modeling approach. The fit of sometimes competing retrieval (SOCR; Stout & Miller, 2007) was compared to the fit of an acquisition-focused model of retrospective revaluation and cue competition. These simulations reveal that SOCR explains this trial order effect in some situations based on its use of local error reduction.

Preservation of within-compound associations after blocked preexposure to two compound flavors

Three experiments investigated the extinction of the within-compound A↔X association established when two compound flavors, AX and BX, are preexposed in blocks (i.e., AX, AX, AX, ... BX, BX, BX). In Experiment 1, a group of rats received preexposure to a block of AX trials followed by a second block of BX trials (AX–BX), while a second group received blocked preexposure to the same stimuli in the opposite order (BX–AX). Subsequently, flavor A was paired with lithium chloride. This conditioning resulted in a similar reduction of consumption of flavor X in both groups. In Experiment 2 four groups of rats received blocked preexposure to AX–BX, AX–B, A–BX, or A–X. After aversive conditioning of X, consumption of A and B was significantly lower for the groups which received these flavors paired with X than for the groups for which these flavors were presented isolated. In Experiment 3 a group of rats was preexposed to a block of SaltX presentations followed by a block of BX presentations (SaltX–BX), and a second group received blocked preexposure to (BX–SaltX). After subsequent sodium depletion, consumption of X was high and similar for the SaltX–BX and for the BX–SaltX groups. These results indicate that the within-compound association established in the first block of a blocked preexposure is not extinguished when the preexposure phase is concluded.

Within-subjects assessment of the within-compound associations resulting from intermixed and blocked preexposure schedules.

Nonhungry rats received training consisting of intermixed presentations of a compound flavor and an element of that compound (AX, X, AX, X, . . .), and then a separate block of presentations of another compound (BX, BX, BX, . . .). Stimuli A and B were two odor solutions (almond and vanilla), and stimulus X was a highly concentrated solution of sucrose. After training, a state of hunger was induced in the rats, and their consumption levels of A and B alone were tested. We found higher test consumption of B than of A (Exp. 1). We interpreted these differences as indicating that the B-X association had become stronger than the A-X association as a result of the training. In Experiment 2, we demonstrated that the presence of X during training was necessary for that effect to appear (Exp. 2). These results give support to the recent proposal that within-compound associations are maintained better by blocked than by intermixed preexposure (Rodríguez & Alonso, 2014). We discuss the implications of this difference for explaining the intermixed-blocked perceptual-learning effect.

Within-compound associations mediate augmented flavor aversions

Five experiments with rat subjects explored flavor-aversion conditioning in the A+/AX+ design. In Experiment 1, odor preconditioning prior to taste+odor compound conditioning significantly strengthened a taste aversion, and taste preconditioning prior to taste+odor conditioning significantly augmented an odor aversion. Experiments 2A and 2B eliminated the alternative explanation that the augmentation effect was produced by differential exposure to the compound during conditioning. Next, we examined a possible mechanism of these augmentation effects by employing post-conditioning extinction of the augmenting stimulus (CS A); previous research had also used this technique to explore augmentation, but those studies yielded conflicting results. In both Experiments 3A and 3B, post-conditioning extinction of CS A led to a significant decrease in the strength of the augmented aversion to CS X, regardless of stimulus modality or stimulus salience. Collectively, these results suggest augmented flavor aversions are produced via within-compound associations between CS A and CS X.

Revaluation of geometric cues reduces landmark discrimination via within-compound associations.

Rats were trained in a triangular water maze in which a compound of geometric and landmark cues indicated the position of a submerged platform. Rats that then underwent revaluation of the geometric cues in the absence of the landmarks subsequently failed to discriminate between the landmarks. In contrast, those animals that received geometry training consistent with their previous experience of the geometry-landmark compound continued to discriminate the landmark cues. The experiment showed that within-compound associations had formed between the geometry and landmarks, and that representations of absent geometric cues could be evoked via presentation of the landmark cues alone. We argue that these evoked representations of the absent geometry cues can counteract any overshadowing of the landmark by geometry cues, and may sometimes result in potentiation. The results of this study do not support theories of cue-competition failure based on independent cue processing, but remain readily explicable by appeal to an account based on within-compound associations.

Erratum to: An elemental model of retrospective revaluation without within-compound associations

Erratum to: An elemental model of retrospective revaluation without within-compound associations

Differential effect of the intermixed and blocked preexposure schedules on the strength of within-compound associations.

In Experiment 1, we demonstrated an intermixed-blocked effect where intermixed preexposure to a flavor compound and to an element of that compound (AX, X, AX, X . . .) reduced generalization between them more than equivalent blocked preexposure (AX, AX . . . X, X . . . , or X, X, AX, AX). Then we used sensory preconditioning (Experiment 2) and conditioned flavor preference (Experiment 3) procedures to assess the strength of the X-A within-compound association resulting from those preexposure schedules. In both experiments, we observed that the within-compound association was stronger after blocked than intermixed preexposure. We suggest that these differences in strength produce more mediated generalization in the blocked than intermixed preexposure.

An elemental model of retrospective revaluation without within-compound associations

When retrospective revaluation phenomena (e.g., unovershadowing: AB+, then A-, then test B) were discovered, simple elemental models were at a disadvantage because they could not explain such phenomena. Extensions of these models and novel models appealed to within-compound associations to accommodate these new data. Here, we present an elemental, neural network model of conditioning that explains retrospective revaluation apart from within-compound associations. In the model, previously paired stimuli (say, A and B, after AB+) come to activate similar ensembles of neurons, so that revaluation of one stimulus (A-) has the opposite effect on the other stimulus (B) through changes (decreases) in the strength of the inhibitory connections between neurons activated by B. The ventral striatum is discussed as a possible home for the structure and function of the present model.

Within-compound associations explain potentiation and failure to overshadow learning based on geometry by discrete landmarks.

In three experiments, rats were trained to locate a submerged platform in one of the base corners of a triangular arena above each of which was suspended one of two distinctive landmarks. In Experiment 1, it was established that these landmarks differed in their salience by the differential control they gained over behavior after training in compound with geometric cues. In Experiment 2, it was shown that locating the platform beneath the less salient landmark potentiated learning based on geometry compared with control rats for which landmarks provided ambiguous information about the location of the platform. The presence of the more salient landmark above the platform for another group of animals appeared to have no effect on learning based on geometry. Experiment 3 established that these landmark and geometry cues entered into within-compound associations during compound training. We argue that these within-compound associations can account for the potentiation seen in Experiment 2, as well as previous failures to demonstrate overshadowing of geometric cues. We also suggest that these within-compound associations need not be of different magnitudes, despite the different effects of each of the landmarks on learning based on geometry seen in Experiment 2. Instead, within-compound associations appear to mitigate the overshadowing effects that traditional theories of associative learning would predict.

Enhanced unblocking from sustained post-trial surprise.

Two appetitive conditioning experiments with rats investigated the mechanisms and properties of unblocking that results from the surprising omission of an expected post-trial unconditioned stimulus (US). Experiment 1 demonstrated unblocking under circumstances in which differences in the contribution of generalization decrement and within-compound associations are equated across experimental and control groups. Following Stage 1 training in which a conditioned stimulus (CS) A was followed by a US and a post-trial US, in Experiment 2 we arranged for the post-trial US to be present on some trials with AX but not others. Under these circumstances an enhancement of unblocking to X was observed, relative to a group who received standard unblocking by US omission. The implications of these results for attentional and US-processing theories of associative learning are discussed.

Inferential dependencies in causal inference: A comparison of belief-distribution and associative approaches.

Causal evidence is often ambiguous, and ambiguous evidence often gives rise to inferential dependencies, where learning whether one cue causes an effect leads the reasoner to make inferences about whether other cues cause the effect. There are 2 main approaches to explaining inferential dependencies. One approach, adopted by Bayesian and propositional models, distributes belief across multiple explanations, thereby representing ambiguity explicitly. The other approach, adopted by many associative models, posits within-compound associations--associations that form between potential causes--that, together with associations between causes and effects, support inferences about related cues. Although these fundamentally different approaches explain many of the same results in the causal literature, they can be distinguished, theoretically and experimentally. We present an analysis of the differences between these approaches and, through a series of experiments, demonstrate that models that distribute belief across multiple explanations provide a better characterization of human causal reasoning than models that adopt the associative approach.