Selective attribution and the judgment of causality

Four experiments used a video game procedure to investigate the mechanism by which people are sensitive to the degree of contingency between two events. Subjects were presented with sets of trials on each of which they could perform a particular action and observe whether the action produced a particular outcome in a situation in which there was an alternative potential cause of the outcome. The experiments attempted to show that the process of selective attribution operates during exposure to a particular contingency and mediates the contingency judgment. In Experiment 1 the impact of outcomes occurring in the absence of the action was reduced by changing the location at which outcomes occurred following the action. Experiment 2 replicated this effect and showed that it was not due to simple changes in temporal contiguity, but rather was due to affecting the process of selective attribution. In Experiment 3 judgments were shown to increase when outcomes occurring in the absence of the action, outcomes which otherwise would reduce judgments of action-outcome contingency, were signaled. Finally, in Experiment 4 this effect was replicated, and in addition it was shown that the signaling effect is not simply due to the presence of the signal. For the effect to be shown, the signal must occur when the outcome occurs in the absence of the action.     

Judgement of act-outcome contingency: The role of selective attribution

In the first experiment subjects were presented with a number of sets of trials on each of which they could perform a particular action and observe the occurrence of an outcome in the context of a video game. The contingency between the action and outcome was varied across the different sets of trials. When required to judge the effectiveness of the action in controlling the outcome during a set of trials, subjects assigned positive ratings for a positive contingency and negative ratings for a negative contingency. Furthermore, the magnitude of the ratings was related systematically to the strength of the actual contingency. With a fixed probability of an outcome given the action, judgements of positive contingencies decreased as the likelihood that the outcome would occur without the action was raised. Correspondingly, the absolute value of ratings of negative contingencies was increased both by an increment in the probability of the outcome in the absence of the action and by a decrement in the probability of the outcome following the action. A systematic bias was observed, however, in that positive judgements were given under a non-contingent relationship when the outcome frequency was relatively high. However, this bias could be reduced by giving extended exposure to the non-contingent schedule (Experiment 2).

This pattern of contingency judgements can be explained if it is assumed that a process of selective attribution operates, whereby people are less likely to attribute an outcome to some potential target cause if another effective cause is present. Experiments 2 and 3 demonstrated the operation of this process by showing that initially establishing another agent as an effective cause of the outcome subsequently reduced or blocked the extent to which the subjects attributed the outcome to the action.

Finally, we argue that the pattern and bias in contingency judgements based upon interactions with a causal process can be explained in terms of contemporary conditioning models of associative learning.     

Forward blocking in human learning sometimes reflects the failure to encode a cue–outcome relationship

In two “allergist” causal judgement experiments, participants were trained with a blocking design (A?+?|AB+). The procedure allowed different food cues to be paired with different fictitious allergic reactions. On test, participants were asked to rate the causal efficacy of the target cues and to recall the particular allergic reaction (outcome) that had followed each cue during training. Forward blocking was observed on the causal judgement measure and on the outcome recall measure in both Experiment 1 and Experiment 2. A backward blocking contingency was also trained in Experiment 2 (AB?+?|A+). Backward blocking was not observed either on the causal judgement or on the outcome recall measure. The evidence from the recall measure suggests that forward blocking in this task results from a failure to encode the B–outcome relationship during training. Associative and nonassociative mechanisms of forward blocking are discussed.     

Forward blocking in human learning sometimes reflects the failure to encode a cue-outcome relationship

In two “allergist” causal judgement experiments, participants were trained with a blocking design (A + |AB+). The procedure allowed different food cues to be paired with different fictitious allergic reactions. On test, participants were asked to rate the causal efficacy of the target cues and to recall the particular allergic reaction (outcome) that had followed each cue during training. Forward blocking was observed on the causal judgement measure and on the outcome recall measure in both Experiment 1 and Experiment 2. A backward blocking contingency was also trained in Experiment 2 (AB + |A+). Backward blocking was not observed either on the causal judgement or on the outcome recall measure. The evidence from the recall measure suggests that forward blocking in this task results from a failure to encode the B-outcome relationship during training. Associative and nonassociative mechanisms of forward blocking are discussed.     

Backward and forward blocking in human electrodermal conditioning: Blocking requires an assumption of outcome additivity

Blocking was observed in two human Pavlovian conditioning studies in which colour cues signalled shock. Both forward (Experiment 1) and backward (Experiment 2) blocking was demonstrated, but only when prior verbal and written instructions suggested that if two signals of shock (A+ and B+) were presented together, a double shock would result (AB++). In this case, participants could assume that the outcome magnitude was additive. Participants given non-additivity instructions (A+ and B+ combined would result in the same outcome, a single shock) failed to show blocking. Modifications required for associative models of learning, and normative statistical accounts of causal induction, to account for the impact of additivity instructions on the blocking effect, are discussed. It is argued that the blocking shown in the present experiments resulted from the operation, not of an error-correction learning rule, nor of a simple contingency detection mechanism, but of a more complex inferential process based on propositional knowledge. Consistent with the present data, blocking is a logical outcome of an A+/AB+ design only if participants can assume that outcomes will be additive.     

Contiguity and the outcome density bias in action-outcome contingency judgements

In cause-outcome contingency judgement tasks, judgements often reflect the actual contingency but are also influenced by the overall probability of the outcome, P(O). Action-outcome instrumental learning tasks can foster a pattern in which judgements of positive contingencies become less positive as P(O) increases. Variable contiguity between the action and the outcome may produce this bias. Experiment 1 recorded judgements of positive contingencies that were largely uninfluenced by P(O) using an immediate contiguity procedure. Experiment 2 directly compared variable versus constant contiguity. The predicted interaction between contiguity and P(O) was observed for positive contingencies. These results stress the sensitivity of the causal learning mechanism to temporal contiguity.     

Toxicosis affects instrumental behaviour in rats

Three experiments examined the effect of toxicosis on instrumental responding. These studies were prompted by Morrison and Collyer's (1974, Experiment 1) finding that the induction of toxicosis after an instrumental conditioning session produces greater response suppression if the response is reinforced by a novel saccharin solution rather than familiar water during conditioning. Experiments 1 and 2 investigated whether this suppression was mediated by the Pavlovian contingency between the contextual cues and the saccharin solution or the instrumental relationship between the response and the reward. A role for the instrumental contingency was indicated by the greater suppression of the response producing novel saccharin rather than water when the context of both responses was equally associated with the saccharin and illness. Experiment 3 found that extinction of the aversion to a novel reinforcer following aversive conditioning would re-establish an action previously associated with that reinforcer, in contrast to an action whose reinforcer remained aversive. This result was a further indication that the instrumental contingency between the response and reward contributes to response suppression.     

Discounting in causality and covariation judgements

In two experiments, humans were asked to judge the strength of a moderate contingency between a cue and an outcome in the presence of a second strong contingency between another cue and the outcome. The first experiment replicated the discounting effect whereby a strong contingency causes subjects to reduce or discount judgements of a moderate contingency. This experiment used a video-game procedure in which subjects camouflaged a tank to make it safe from mines. The second causal cue was the presence or absence of a spotter plane. Experiment 1 also ruled out the possibility that judgements might be determined by the number of co-occurrences of the cue and outcome as opposed to the level of contingency. The second experiment used an abstract scenario in which discounting was demonstrated when subjects were asked to judge the relationship between the occurrence of geometric objects. The instructions were neutral to discourage causal hypotheses. These results support the notion that judgements result from associative or connectionist processes and not from a two-step cognitive retrospective process in which an estimate of covariation is calculated between cue and outcome and then in a second step this information is used in a normative manner when a decision is made.     

Continuous monitoring of human contingency judgment across trials

Recent evidence suggests that people are sensitive to the degree of contingency between their actions and ensuing outcomes, but little is known about the way in which such contingency judgments develop as more and more information about the contingency is provided. Three experiments examined this issue in the context of a video game. In Experiment 1, it was found that contingency judgments follow growth functions: When the contingency was positive, judgments increased toward a positive asymptote, and when the contingency was negative, judgments decreased toward a negative asymptote. When the contingency was zero, judgments themselves remained close to zero but were biased by the overall frequency with which the outcome occurred. In Experiment 2, it was shown that the growth function was not the result of the anchoring of early judgments at the zero point. The bias in judgments when the contingency is zero was investigated in Experiment 3. The results are discussed in terms of rule-based analyses and contemporary theories of conditioning.     

Bidirectional Instrumental Conditioning

Three experiments examined bidirectional instrumental conditioning by training hungry rats to push a pole in one direction for food pellets and in the other for either a sugar or a starch solution. In the first study we examined whether the animals learned about the actionreinforcer relations using a specific satiety procedure. Prefeeding one type of reinforcer before an extinction test selectively depressed the performance of the action that had been paired with this reinforcer during training. The second experiment investigated the sensitivity of the bidirectional actions to variations in the action-reinforcer contingencies. When the instrumental contingency was degraded by presenting unpaired reinforcers, the animals pushed less in the direction that was paired with the reinforcer type that was the same as the non-contiguous one. A third study revealed that increasing the rate of reinforcement for one action enhanced its rate of performance without significantly affecting the performance of the other action. We conclude that the effects of reinforcer devaluation, the action-outcome contingency, and the rate of reinforcement are not mediated by Pavlovian associations between the manipulandum and the reinforcer.     

Outcome additivity and outcome maximality influence cue competition in human causal learning

Recent research suggests that outcome additivity pretraining modulates blocking in human causal learning. However, the existing evidence confounds outcome additivity and outcome maximality. Here the authors present evidence for the influence of presenting information about outcome maximality (Experiment 1) and outcome additivity (Experiment 2) on subsequent forward blocking. The results of Experiment 3 confirm that, with outcome maximality controlled, outcome additivity affects backward blocking but not release from overshadowing. Finally, the results of Experiment 4 demonstrate that information about outcome additivity has a similar effect on forward blocking if presented after the blocking training instead of before. The results are compatible with the idea that blocking results from inferential processes at the time of testing and not from a failure to acquire associative strength during training.     

Judgement of two causal candidates from contingency information: II. Effects of information about one cause on judgements of the other cause

When judgements are being made about two causes there are eight possible kinds of contingency information: occurrences and nonoccurrences of the outcome when both causes are present, when Cause 1 alone is present, when Cause 2 alone is present, and when neither cause is present. It is proposed that contingency information is used to some extent to judge proportionate strength, which is the proportion of occurrences of the outcome that each cause can account for. This leads to a prediction that judgements of one cause will be influenced by information about occurrences, but not nonoccurrences, of the outcome when only the other cause is present. In six experiments consistent support was found for this prediction when the cause being judged had a positive relation with the outcome, but no consistent tendency was found when the cause being judged had a negative relation with the outcome. The effects found for causes with positive contingency cannot be explained by the Rescorla-Wagner model of causal judgement nor by the hypothesis that causal judgements are based on conditional contingencies.     

Parallel acquisition of awareness and differential delay eyeblink conditioning

There is considerable debate about whether differential delay eyeblink conditioning can be acquired without awareness of the stimulus contingencies. Previous investigations of the relationship between differential-delay eyeblink conditioning and awareness of the stimulus contingencies have assessed awareness after the conditioning session was finished using a post-experimental questionnaire. In two experiments, the point at which contingency awareness developed during the conditioning session was estimated from a button-press measure of expectancy of the unconditioned stimulus (US). In both experiments, knowledge of the stimulus contingencies and acquisition of differential delay eyeblink conditioning developed approximately in parallel. In Experiment 1 it was shown that predicting the US facilitated eyeblink conditioning compared with predicting the eyeblink response. In Experiment 2, a masking task was used that slowed down the emergence of awareness, and it was shown that differential conditioning only occurred in participants who were able to predict the US. The current findings challenge the hypothesis that differential delay eyeblink conditioning is entirely mediated by a functionally and neurally distinct nondeclarative learning system.     

The sad truth about depressive realism

In one form of a contingency judgement task individuals must judge the relationship between an action and an outcome. There are reports that depressed individuals are more accurate than are nondepressed individuals in this task. In particular, nondepressed individuals are influenced by manipulations that affect the salience of the outcome, especially outcome probability. They overestimate a contingency if the probability of an outcome is high—the “outcome-density effect”. In contrast, depressed individuals display little or no outcome-density effect. This apparent knack for depressives not to be misled by outcome density in their contingency judgements has been termed “depressive realism”, and the absence of an outcome-density effect has led to the characterization of depressives as “sadder but wiser”. We present a critical summary of the depressive realism literature and provide a novel interpretation of the phenomenon. We suggest that depressive realism may be understood from a psychophysical analysis of contingency judgements.*     

Insensitivity to stimulus validity in human pavlovian conditioning

Four experiments investigated the role of information value, or stimulus validity, in human electrodermal conditioning. Conditioned stimuli (CSs) were long, variable duration (10-50-sec) slides or sounds, the unconditioned stimulus (US) was shock, and the primary measures were change in tonic skin conductance level and self-report expectancy of shock. In Experiment 1 electrodermal responding to a target stimulus was marginally lower in a blocking group than in a superconditioning group. This difference failed to occur in Experiment 2, despite increased sensitivity of the electrodermal measure. Explicit instructions to pay attention to the relationship between each stimulus and shock improved learning, but did not lead to group differences (Experiment 3). In Experiment 4 expectancy ratings in a blocking group were lower than in an overshadowing group, but no electrodermal differences occurred. The results were interpreted in terms of non-additive learning processes such as occasion-setting, in addition to a general lack of transfer of learning about a stimulus from one context to another (element to compound or vice versa). It was suggested that sensitivity to stimulus validity might be observed if conditioning were embedded in a causally familiar task, as in contingency judgement research.     

Interference with judgements of control and learning as a result of prior exposure to controllable and uncontrollable feedback during concept-learning tasks

Two experiments examined whether exposure to an uncontrollable relationship between an action and its outcome during a nonaversive pretreatment phase would attenuate subsequent ratings of control given to actions emitted by subjects. In Experiment 1, such an interference effect was demonstrated relative to a group that received prior training with a controllable action- outcome relationship, and relative to a group not exposed to any prior relationship. In Experiment 2, these effects were replicated, and interference was also found to occur when learning a maze task. Thus, the effects of helplessness were shown to be quite general, to be produced by a nonaversive induction procedure, and to occur most readily when the current contingency between action and outcome was weakest.     

Causal powers and preventers: An explanatory account of cue interaction effects in human causal judgement

According to the causal powers theory, all causal relations are understood in terms of causal powers of one thing producing an effect by acting on liability of another thing. Powers can vary in strength, and their operation also depends on the presence of preventers. When an effect occurs, there is a need to account for the occurrence by assigning sufficient strength to produce it to its possible causes. Contingency information is used to estimate strengths of powers and preventers and the extent to which they account for occurrences and nonoccurrences of the outcome. People make causal judgements from contingency information by processes of inference that interpret evidence in terms of this fundamental understanding. From this account it is possible to derive a computational model based on a common set of principles that involve estimating strengths, using these estimates to interpret ambiguous information, and integrating the resultant evidence in a weighted averaging model. It is shown that the model predicts cue interaction effects in human causal judgement, including forward and backward blocking, second and third order backward blocking, forward and backward conditioned inhibition, recovery from overshadowing, superlearning, and backward superlearning.     

Instrumental judgment and performance under variations in action-outcome contingency and contiguity

The intentional theory of instrumental performance proposes that performance of an action is determined in part by a belief about its causal effectiveness in producing a desired outcome. At variance with this notion, previous implicit learning experiments appear to have yielded dissociations between subjects' performance and beliefs. In two experiments, subjects were given an opportunity to perform an action--pressing a key on a computer keyboard--which was associated with an outcome on the computer screen according to a free-operant contingency. The subjects in one group were asked to judge the effectiveness of the action in causing the outcome, while those in a second group were asked to maximize their points score under a payoff schedule. In the first study, the effect of varying the contingency between the action and outcome was examined by keeping the probability of an outcome contiguous with an action constant and varying the probability of an outcome in the absence of an action. Performance and judgments showed a comparable sensitivity to variations of the instrumental contingency. In the second study, the delay between the action and the resultant outcome was varied. Increasing the action-outcome delay from 0 sec up to 4 sec produced a systematic decline in both causal judgments and performance relative to noncontingent, control conditions. These results are in accord with the intentional theory of performance, but they present difficulties for the notion of implicit learning.     

Factors that affect action possibility judgements: Recent experience with the action and the current body state

It has been suggested that action possibility judgements are formed through a covert simulation of the to-be-executed action. We sought to determine whether the motor system (via a common coding mechanism) influences this simulation, by investigating whether action possibility judgements are influenced by experience with the movement task (Experiments 1 and 2) and current body states (Experiment 3). The judgement task in each experiment involved judging whether it was possible for a person's hand to accurately move between two targets at presented speeds. In Experiment 1, participants completed the action judgements before and after executing the movement they were required to judge. Results were that judged movement times after execution were closer to the actual execution time than those prior to execution. The results of Experiment 2 suggest that the effects of execution on judgements were not due to motor activation or perceptual task experience—alternative explanations of the execution-mediated judgement effects. Experiment 3 examined how judged movement times were influenced by participants wearing weights. Results revealed that wearing weights increased judged movement times. These results suggest that the simulation underlying the judgement process is connected to the motor system, and that simulations are dynamically generated, taking into account recent experience and current body state.     

Alcohol seeking by rats: Action or habit?

In two experiments, we examined the relative susceptibility to outcome devaluation of lever pressing by rats for either a 10% ethanol solution or food pellets. The rats were trained to press different levers for these two reinforcers using a sucrose-substitution procedure. An aversion was then conditioned from either the ethanol solution or the food pellets by pairing consumption with illness induced by lithium chloride. When instrumental performance was subsequently tested in extinction, the rats pressed less on the pellet lever if the pellets, rather than the ethanol, had been devalued by aversion conditioning. By contrast, performance on the ethanol lever was unaffected by whether the ethanol or pellets were devalued. Moreover, noncontingent presentations of the devalued reinforcer had no impact on test performance. The differential resistance to outcome devaluation suggests that, in contrast to food seeking, alcohol seeking is a stimulus-response habit rather than a goal-directed action mediated by a representation of the action-outcome contingency.