Within-compound associations in retrospective revaluation and in direct learning: A challenge for comparator theory

In three human causal learning experiments we investigated the role of within-compound associations in learning about absent cues versus learning about present cues. Different theoretical approaches agree that within-compound associations are essential for learning about absent cues-that is, for retrospective revaluation. They differ, however, with regard to the role of within-compound associations for learning about present cues-that is, for direct learning. A memory test was used to assess within-compound associations. Experiment 1 used a blocking/release from overshadowing design, Experiment 2 used a conditioned inhibition design, and Experiment 3 used a higher-order cue selection design. In all experiments, first-order retrospective revaluation was significantly correlated with within-compound associations, but no significant correlations were found for the direct learning conditions. In addition to this, second-order retrospective revaluation in Experiment 3 was positively correlated to joint knowledge of first-order and second-order within-compound associations. Furthermore, cue selection effects were stronger for direct learning conditions than for retrospective learning conditions. These results are at variance with the comparator hypothesis but are in agreement with a modified associative theory and with the suggestion that retrospective revaluation might be due to rehearsal processes.     

The role of memory for compounds in cue competition

Revisions of common associative learning models incorporate a within-compound association mechanism in order to explain retrospective cue competition effects (e.g., [Dickinson, A., & Burke, J. (1996). Within-compound associations mediate the retrospective revaluation of causality judgements. Quarterly Journal of Experimental Psychology, 49B, pp. 60-80.]). These models predict a correlation between memory for compounds (as a measure for the strength of within-compound associations) and retrospective cue competition but not forward cue competition. This was indeed found in a study of [Melchers, K. G., Lachnit, H., & Shanks, D. (2004). Within-compound associations in retrospective revaluation and in direct learning: A challenge for comparator theory. Quarterly Journal of Experimental Psychology, 57B, pp. 25-54.]. We argue that a higher-order reasoning account of causal learning can also explain the evidence for the role of within-compound associations in cue competition. Moreover, this account predicts that if making inferences during the learning stage is impeded, a correlation between memory for compounds and forward cue competition should be found as well. The results of a new study confirmed this prediction.     

The comparator theory fails to account for the selective role of within-compound associations in cue-selection effects

In a human causal learning experiment, we investigated cue selection effects to test the comparator theory (Denniston, Savastano, & Miller, 2001; Miller & Matzel, 1988). The theory predicts that the occurrence of cue selection is independent of whether the relevant learning trials are presented in a standard forward manner or in a backward manner and that within-compound associations are of equal importance in both cases. We found that the strength of the cue-selection effect was positively correlated with knowledge of within-compound associations in the backward condition but not in the forward condition. Furthermore, cue-selection effects were less pronounced in the former than in the latter condition. These results are at variance with the comparator hypothesis but are in agreement with a modified associative theory and with the suggestion that retrospective revaluation might be due to rehearsal processes.     

Re-examination of the role of within-compound associations in the retrospective revaluation of causal judgements

We investigated blocking and retrospective revaluation of causal judgements using a scenario in which food cues acted as potential causes of an allergic reaction as the outcome. In the blocking contingency,the treatment cues were either paired or unpaired with the outcome prior to a second stage in which sequential compounds of treatment and target cues were paired with the outcome. The order of this compound and treatment training was reversed in retrospective revaluation contingencies. When the interstimulus interval between the treatment and target cues was unfilled on compound trials (Experiments 1 and 3), both blocking and retrospective revaluation were observed in that the target cue trained in compound with the paired treatment cue attracted lower causal ratings than the target cue trained in compound with the unpaired treatment cue. By contrast, performing a mental arithmetic task using numerals presented during the interstimulus interval had no effect on the magnitude of blocking but rendered retrospective revaluation unreliable (Experiments 2 and 3). These results provide further support for accounts of revaluation based upon within-compound associations.     

Inference-based retrospective revaluation in human causal judgments requires knowledge of within-compound relationships

In an allergist causal-judgment task, food compounds were followed by an allergic reaction (e.g., AB+), and then 1 cue (A) was revalued. Experiment 1, in which participants who were instructed that whatever was true about one element of a causal compound was also true of the other, showed a reverse of the standard retrospective revaluation effect. That is, ratings of B were higher when A was causal (A+) than when A was safe (A-). This effect was taken to reflect inferential reasoning, not an associative mechanism. In Experiment 2, within-compound associations were found to be necessary to produce this inference-based revaluation. Therefore, evidence that within-compound associations are necessary for retrospective revaluation is consistent with the inferential account of causal judgments.     

Causal Learning: Association Versus Computation

Causal learning enables humans and other animals not only to predict important events or outcomes, but also to control their occurrence in the service of needs and desires. Computational theories assume that causal judgments are based on an estimate of the contingency between a causal cue and an outcome. However, human causal learning exhibits many of the characteristics of the associative learning processes thought to underlie animal conditioning. One problem for associative theory arises from the finding that judgments of the causal power of a cue can be revalued retrospectively after learning episodes when that cue is not present. However, if retrieved representations of cues can support learning, retrospective revaluation is anticipated by modified versions of standard associative theories.     

Within compound Associations Mediate the Retrospective Revaluation of Causality Judgements

The role of within-compound associations in the retrospective revaluation of causality judgements was investigated in a two-stage procedure in which the subjects were asked to learn whether or not different food stimuli caused an allergic reaction in hypothetical patients. In the compound-cue stage a number of compound cues, each consisting of a competing stimulus and a target stimulus, were associated with the reaction across a series of trials, whereas in the single-cue stage the subjects had the opportunity to learn which of the competing cues, when presented alone, caused the reaction. Each target stimulus was presented with the same competing cue across all compound trials in the consistent condition, but with a different competing cue on each trial in the varied condition. In a forward procedure, in which the single-cue stage preceded compound cue training, judgements of the causal effectiveness of the target stimuli were reduced or blocked by training them in compound with a competing cue that had been previously paired with the reaction. Moreover, the magnitude of this reduction was comparable in the consistent and varied conditions. This was not true, however, when the single- and compound-cue stages were reversed in the backward procedure. Judgements for target cues compounded with competing cues that were subsequently paired with the reaction were reduced only in the consistent condition. If it is assumed that stronger associations were formed between the competing and target stimuli during the compound-cue stage in the consistent condition than in the varied condition, this pattern suggests that the retrospective revaluation of causality judgements can be mediated by the formation of within-compound associations.     

A comparison between elemental and compound training of cues in retrospective revaluation

Associative learning theories assume that cue interaction and, specifically, retrospective revaluation occur only when the target cue is previously trained in compound with the to-be-revalued cue. However, there are recent demonstrations of retrospective revaluation in the absence of compound training (e.g., Matute & Pine?o, 1998a, 1998b). Nevertheless, it seems reasonable to assume that cue interaction should be stronger when the cues are trained together than when they are trained apart. In two experiments with humans, we directly compared compound and elemental training of cues. The results showed that retrospective revaluation in the elemental condition can be as strong as and, sometimes, stronger than that in the compound condition. This suggests that within-compound associations are not necessary for retrospective revaluation to occur and that these effects can possibly be best understood in the framework of general interference theory.     

Retrospective revaluation in humans: Learning or memory?

The phenomenon of retrospective revaluation has posed considerable problems for many associative learning theories as it involves a change in the associative strength of a cue on trials on which that cue is absent. The present series of experiments pursues this idea of changes in associative strength between evoked representations of cues, in an effort to establish, de novo, an excitatory connection between two cues simultaneously activated in memory. Given the finding of Dwyer, Mackintosh, and Boakes (1998) that simultaneous activation of absent cues in the memory of rats resulted in learning comparable to that seen inretrospective revaluation, we expected that if retrospective revaluation was found in humans, then excitatory learning due to simultaneous activation would also be seen. This was not the case. The implications of our results are discussed in terms of Dickinson and Burke's (1996) modified SOP model and a version of McLaren's (1993) APECS network. We conclude that many of the effects attributed to learning in retrospective revaluation studies are better thought of as due to changes in the retrievability of items in memory.     

Predictive versus diagnostic causal learning: Evidence from an overshadowing paradigm

Causal directionality belongs to one of the most fundamental aspects of causality that cannot be reduced to mere covariation. This paper is part of a debate between proponents of associative theories, which claim that learners are insensitive to the causal status of cues and outcomes, and proponents of causal-model theory, which postulates an interaction of assumptions about causal directionality and learning. Some researchers endorsing the associationist view have argued that evidence for the interaction between cue competition and causal directionality may be restricted to two-phase blocking designs. Furthermore, from the viewpoint of causal-model theory, blocking designs carry the potential problem that the predicted asymmetries of cue competition are partly dependent on asymmetries of retrospective inferences. The present experiments use a one-phase overshadowing paradigm that does not allow for retrospective inferences and therefore represents a more unambiguous test of sensitivity to causal directionality. The results strengthen causal-model theory by clearly demonstrating the influence of causal directionality on learning. However, they also provide evidence for boundary conditions for this effect by highlighting the role of the semantics of the learning task.     

Associative history affects the associative change undergone by both presented and absent cues in human causal learning

R. A. Rescorla (2000) noted that a number of influential theories of associative learning do not take the associative history of cues (i.e., the prior training that they have received) into account when calculating the associative change undergone by those cues. The authors tested this assumption in a human causal learning paradigm and found associative history to be an important determinant of the learning undergone by cues that are presented on a trial. Moreover, associative history was also found to influence the amount of retrospective revaluation undergone by absent cues. These findings conflict with models of causal learning in which the associative change undergone by an element of a cue compound is governed by a summed error term (e.g., R. A. Rescorla & A. R. Wagner, 1972).     

Asymmetries in cue competition in forward and backward blocking designs: Further evidence for causal model theory

A hallmark feature of elemental associative learning theories is that multiple cues compete for associative strength when presented with an outcome. Cue competition effects have been observed in humans, both in forward and in backward blocking procedures (e.g., Shanks, 1985) and are often interpreted as evidence for an associative account of human causal learning (e.g., Shanks & Dickinson, 1987). Waldmann and Holyoak (1992), however, demonstrated that cue competition only occurs in predictive, and not diagnostic, learning paradigms. While unexplainable from an associative perspective, this asymmetry readily follows from structural considerations of causal model theory. In this paper, we show that causal models determine the extent of cue competition not only in forward but also in backward blocking designs. Implications for associative and inferential accounts of causal learning are discussed.     

Aging and retrospective revaluation of causal learning

In a 2-stage causal learning task, young and older participants first learned which foods presented in compound were followed by an allergic reaction (e.g., STEAK-BEANS→ REACTION) and then the causal efficacy of 1 food from these compounds was revalued (e.g., BEANS→ NO REACTION). In Experiment 1, unrelated food pairs were used, and although there were no age differences in compound- or single-cue-outcome learning, older adults did not retrospectively revalue the causal efficacy of the absent target cues (e.g., STEAK). However, they had weaker within-compound associations for the unrelated foods, and this may have prevented them from retrieving the representations of these cues. In Experiment 2, older adults still showed no retrospective revaluation of absent cues even though compound food cues with pre-existing associations were used (e.g., STEAK-POTATO), and they received additional learning trials. Finally, in Experiment 3, older adults revalued the causal efficacy of the target cues when small, unobtrusive icons of these cues were present during single-cue revaluation. These findings suggest that age-related deficits in causal learning for absent cues are due to ineffective associative binding and reactivation processes.     

Sometimes-competing retrieval (SOCR): a formalization of the comparator hypothesis

Cue competition is one of the most studied phenomena in associative learning. However, a theoretical disagreement has long stood over whether it reflects a learning or performance deficit. The comparator hypothesis, a model of expression of Pavlovian associations, posits that learning is not subject to competition but that performance reflects a complex interaction of encoded associative strengths. That is, subjects respond to a cue to the degree that it signals a change in the likelihood or magnitude of reinforcement relative to that in the cue's absence. Initially, this performance-focused view was supported by studies showing that posttraining revaluation of a competing cue often influences responding to the target cue. However, recently developed learning-focused accounts of retrospective revaluation have revitalized the debate concerning cue competition. Further complicating the picture are phenomena of cue facilitation, which have been addressed less frequently than cue competition by formal models of conditioning of either class. The authors present a formalization and extension of the comparator hypothesis, which results in sharpened differentiation between it and the new learning-focused models.     

Higher-order retrospective revaluation in human causal learning

Previous studies demonstrated that participants will retrospectively adjust their ratings about the relation between a target cue and an outcome on the basis of information about the causal status of a competing cue that was previously paired with the target cue. We demonstrate that such retrospective revaluation effects occur not only for target cues with which the competing cue was associated directly, but also for target cues that were associated indirectly with the competing cue. These second-order and third-order retrospective revaluation effects are compatible with certain implementations of the probabilistic contrast model and with a modified, extended comparator model, but cannot be explained on the basis of a revised Rescorla-Wagner model or a revised SOP model.     

Predictions and causal estimations are not supported by the same associative structure

Studies performed by different researchers have shown that judgements about cue-outcome relationships are systematically influenced by the type of question used to request those judgements. It is now recognized that judgements about the strength of the causal link between a cue and an outcome are mostly determined by the cue-outcome contingency, whereas predictions of the outcome are more influenced by the probability of the outcome given the cue. Although these results make clear that those different types of judgement are mediated by some knowledge of the normative differences between causal estimations and outcome predictions, they do not speak to the underlying processes of these effects. The experiment presented here reveals an interaction between the type of question and the order of trials that challenges standard models of causal and predictive learning that are framed exclusively in associative terms or exclusively in higher order reasoning terms. However, this evidence could be easily explained by assuming the combined intervention of both types of process.     

Causal beliefs and conditioned responses: retrospective revaluation induced by experience and by instruction

The author tested causal beliefs and conditioned responses in a task involving retrospective revaluation of the causal status of a target cue with respect to electric shock. Successful revaluation was observed on both self-report shock expectancy and skin conductance, whether the training trials were directly experienced, described, or partly experienced and partly described. The results contradict models that link anticipatory conditioned responses to a separate or earlier process from that underlying explicit causal knowledge. They suggest instead that a single learning process gives rise to propositional knowledge that (a) drives anticipatory responding, (b) forms the basis for self-reported causal beliefs, and (c) can be combined with other knowledge, provided either by experience or symbolically, to generate inferences such as retrospective revaluation.     

A comparison of forward blocking and reduced overshadowing in human causal learning

In this study, we directly compared forward blocking with reduced overshadowing in a human causal learning study using an A+, B- (first learning stage), AX+, BY+, KL+ (second learning stage) design. The results showed that reduced overshadowing was significantly stronger than forward blocking. These results are problematic for at least some associative learning models but were predicted on the basis of higher order reasoning accounts of cue competition in human causal learning.     

An inhibitory within-compound association attenuates overshadowing

According to the comparator hypothesis (Miller & Matzel, 1988), cue competition depends on the association between a target stimulus (X) and a competing cue (e.g., an overshadowing cue [A]). Thus, it was expected that overshadowing would be reduced by establishing an inhibitory-like relationship between X and A before compound conditioning. In three lever press suppression experiments with rats, this expectation was supported. Experiment 1 showed that establishing an inhibitory X-A relationship reduced overshadowing. In Experiment 2, degrading the inhibitory-like relationship before conditioning allowed reinforced AX compound trials to result in overshadowing. Experiment 3 replicated the results of Experiment 2 when the inhibitory relationship was degraded after compound conditioning. The results support the view that within-compound associations are necessary not only for retrospective revaluation, but also for conventional cue competition.     

The hide-and-seek of retrospective revaluation: recovery from blocking is context dependent in human causal learning

In a typical blocking procedure, pairings of a compound consisting of 2 stimuli, A and X, with the outcome are preceded by pairings of only A with the outcome (i.e., A+ then AX+). This procedure is known to diminish responding to the target cue (X) relative to a control group that does not receive the preceding training with blocking cue A. We report 2 experiments that investigated the effect of extinguishing a blocking cue on responding to the target cue in a human causal learning paradigm (i.e., A+ and AX+ training followed by A- training). The results indicate that extinguishing a blocking cue increases conditioned responding to the target cue. Moreover, this increase appears to be context dependent, such that increased responding to the target is limited to the context in which extinction of the blocking cue took place. We discuss these findings in the light of associative and propositional learning theories.