Probability and Contiguity Trade-Offs in Causal Induction

… Journal of Experimental Psychology: Section B, 1998

Two experiments on human causal induction with multiple candidate causes are reported. Experiment 1 investigated the in¯uence of a perfect preventive cause on the ratings of a less contingent cause. Whereas the Rescorla± Wagner model (RWM) and Cheng' s probabilistic contrast model predict that the less contingent cause should be completely discounted, the Pearce model predicts, in most cases, an enhancement of that cause' s perceived importance. Results corresponded more closely to the predictions of the Pearce model. T he predictions of both the RWM and the Pearce model rely on a constant context cue acquiring associative strength, yet no such cue was explicitly identi® ed in the task scenario employed in Experiment 1. Experiment 2 replicated a number of key conditions of Experiment 1 with a task scenario that afforded ratings of the causal importance of the context in which the effectiveness of the discrete candidate causes was evaluated. In addition, the number of trials was increased to test the possibility that the ratings in Experiment 1 were the product of incomplete learning. T he results of the ® rst experiment were replicated and the ratings of the effectiveness of the context cue were anticipated by both the RWM and the Pearce model. Overall, the Pearce model offers a more comprehensive account of the causal inferences recorded in this study. In a causal induction task a number of candidate causes may precede the occurrence or absence of a target outcome. T he nature of the inductive inferences supported by one such candidate cause is in part determined by its predictive value as de® ned by the difference between the probability of the outcome given its presence, P(O | C), and given its absence, P(O | 2

Memory & Cognition, 2007

2009

Contemporary theories of Human Causal Induction assume that causal knowledge is inferred from observable contingencies. While this assumption is well supported by empirical results, it fails to consider an important problem-solving aspect of causal induction in real time: In the absence of well structured learning trials, it is not clear whether the effect of interest occurred because of the cause under investigation, or on its own accord.

2002

How do humans discover causal relations when the effect is not immediately observable? Previous experiments have uniformly demonstrated detrimental effects of outcome delays on causal induction. These findings seem to conflict with everyday causal cognition, where humans can apparently identify long-term causal relations with relative ease.

Memory & Cognition, 2000

The temporal relations among candidate causes were studied in a causal induction task using a design that is known to produce occasion setting in animal learning preparations. For some subset of the observations, one event, the occasion setter, was accompanied by another event, the conditional cause; for another subset ofthe observations, the conditional cause occurred alone. The efficacy of the conditional cause depended on whether it was or was not accompanied by the occasion setter. Participants used the occasion setter to modulate their effect expectancy to the conditional cause when the events were presented serially, but not simultaneously. Current causal induction models are unable to account for the full range of effects that we observed; the relative roles of time, attention, and cue distinctiveness are discussed.

2011

People often makes inductive inferences that go beyond the data that are given. In order to generate these inferences, people must rely on inductive biases - constraints on learning that guide conclusion from limited data. This thesis presents a survey of three topics concerning people's inductive biases.The first part of this thesis examines people's expectations about the strengths of causes in elemental causal induction - learning about the relationship between a single cause and effect. These expectations are formalized as prior probabilities in a Bayesian model. We estimate people's prior beliefs concerning the variables involved in such causal systems using the technique of iterated learning and demonstrate that a Bayesian model using the priors which are produced by this experiment performs well in predicting human behavior.The second part attempted to capture people's inductive biases in causal relationships by expressing them in logical rules, and assign pri...

Journal of Experimental Psychology: General, 2010

Temporal predictability refers to the regularity or consistency of the time interval separating events. When encountering repeated instances of causes and effects, we also experience multiple cause–effect temporal intervals. Where this interval is constant it becomes possible to predict when the effect will follow from the cause. In contrast, interval variability entails unpredictability. Three experiments investigated the extent to which temporal predictability contributes to the inductive processes of human causal learning. The authors demonstrated that (a) causal relations with fixed temporal intervals are consistently judged as stronger than those with variable temporal intervals, (b) that causal judgments decline as a function of temporal uncertainty, and (c) that this effect remains undiminished with increased learning time. The results therefore clearly indicate that temporal predictability facilitates causal discovery. The authors considered the implications of their findings for various theoretical perspectives, including associative learning theory, the attribution shift hypothesis, and causal structure models.

Journal of Experimental Psychology: Animal Behavior Processes, 2002

In predictive causal inference, people reason from causes to effects, whereas in diagnostic inference, they reason from effects to causes. Independently of the causal structure of the events, the temporal structure of the information provided to a reasoner may vary (e.g., multiple events followed by a single event vs. a single event followed by multiple events). The authors report 5 experiments in which causal structure and temporal information were varied independently. Inferences were influenced by temporal structure but not by causal structure. The results are relevant to the evaluation of 2 current accounts of causal induction, the Rescorla-Wagner

International Journal …, 2009

We thank Jacky Boivin and Todd Bailey for helpful discussion of statistical methods to analyze choice data, and Mike Oaksford and Ulrike Hahn for comments on an earlier draft.

Current Directions in Psychological Science, 2006

The philosopher David Hume's conclusion that causal induction is solely based on observed associations still presents a puzzle to psychology. If we only acquired knowledge about statistical covariations between observed events without accessing deeper information about causality, we would be unable to understand the differences between causal and spurious relations, between prediction and diagnosis, and between observational and interventional inferences. All these distinctions require a deep understanding of causality that goes beyond the information given. We report a number of recent studies that demonstrate that people and rats do not stick to the superficial level of event covariations but reason and learn on the basis of deeper causal representations. Causal-model theory provides a unified account of this remarkable competence.

Memory & Cognition, 1997

reported three experiments in which they attempted to test whether causal order affects cue selection, and concluded that it does not. Their study provides an opportunity to highlight some basic methodological criteria that must be met in order to test whether and how causal order influences learning. In particular, it is necessary to (1) ensure that participants consistently interpret the learning situation in terms of directed cause-effect relations; (2) measure the causal knowledge they acquire; (3) manipulate causal order; and (4) control the statistical relations between cause and effect. With respect to these criteria, each experiment reported by Shanks and Lopez fails on multiple counts. Moreover, several aspects of the results reported by Shanks and Lopez are explained by causal-model theory, but not by associative accounts. Their study thus adds to a growing body of evidence from different laboratories indicating that human contingency learning can be guided by causal interpretation.

Memory & Cognition, 2016

A probabilistic causal chain A -> B -> C may intuitively appear to be transitive: If A probabilistically causes B, and B probabilistically causes C, A probabilistically causes C. However, probabilistic causal relations are only transitive if the so-called Markov condition holds. In two experiments, we examined how people make probabilistic judgments about indirect relationships A -> C in causal chains A -> B -> C that violate the Markov condition. We hypothesized that participants would make transitive inferences in accordance with the Markov condition although they were presented with counterevidence showing intransitive data. For instance, participants were successively presented with data entailing positive dependencies A -> B and B -> C. At the same time, the data entailed that A and C were statistically independent. The results of two experiments show that transitive reasoning via a mediating event B influenced and distorted the induction of the indirect relation between A and C. Participants’ judgments were affected by an interaction of transitive, causal-model-based inferences and the observed data. Our findings support the idea that people tend to chain individual causal relations into mental causal chains that obey the Markov condition and thus allow for transitive reasoning, even if the observed data entail that such inferences are not warranted.

2000

Three experiments investigated the impact of delay on human causal learning. We present a new paradigm based on the presentation of continuous event streams, and use it to test two hypotheses drawn from associative learning theories of causal inference. Unlike free-operant procedures traditionally used to study temporal aspects of causal learning (Shanks, Pearson,

Memory & Cognition, 1995

Psychology and Aging, 2009

Age differences in causal judgment are consistently greater for preventative/negative relationships than for generative/positive relationships. We used a feature analytic procedure to determine whether this effect might be due to differences in young and older adults' integration of contingency evidence during causal induction. To reduce the impact of age-related changes in learning/memory we presented contingency evidence for preventative, non-contingent, and generative relationships in summary form and to induce participants to integrate greater or lesser amounts of this evidence, we varied the meaningfulness of the causal context. Young adults showed greater flexibility in their integration processes than older adults. In an abstract causal context, there were no age differences in causal judgment or integration, but in meaningful contexts, young adults' judgments for preventative relationships were more accurate than older adults' and they assigned more weight to the contingency evidence confirming these relationships. These differences were mediated by age-related changes in processing speed. The decline in this basic cognitive resource may place boundaries on the amount or the type of evidence that older adults can integrate for causal judgment.