We present a framework for the rational analysis of elemental causal induction – learning about the existence of a relationship between a single cause and effect – based upon causal graphical models. This framework makes precise the distinction between causal structure and causal strength: the difference between asking whether a causal relationship exists and asking how strong that causal relationship might be. We show that two leading rational models of elemental causal induction, ∆P and causal power, both estimate causal strength, and introduce a new rational model, causal support, that assesses causal structure. Causal support predicts several key phenomena of causal induction that cannot be accounted for by other rational models, which we explore through a series of experiments. These phenomena include the complex interaction between ∆P and the base-rate probability of the effect in the absence of the cause, sample size effects, inferences from incomplete contingency tables, and causal learning from rates. Causal support also provides a better account of a number of existing datasets than either ∆P or causal power.

Inducing causal relationships from observations is a classic problem in scientific inference, statistics, and machine learning. It is also a central part of human learning, and a task that people perform remarkably well given its notorious difficulties. People can learn causal structure in various settings, from diverse forms of data: observations of the co-occurrence frequencies between causes and effects, interactions between physical objects, or patterns of spatial or temporal coincidence. These different modes of learning are typically thought of as distinct psychological processes and are rarely studied together, but at heart they present the same inductive challenge—identifying the unobservable mechanisms that generate observable relations between variables, objects, or events, given only sparse and limited data. We present a computational-level analysis of this inductive problem and a framework for its solution, which allows us to model all these forms of causal learning in a common language. In this framework, causal induction is the product of domain-general statistical inference guided by domain-specific prior knowledge, in the form of an abstract causal theory. We identify 3 key aspects of abstract prior knowledge—the ontology of entities, properties, and relations that organizes a domain; the plausibility of specific causal relationships; and the functional form of those relationships—and show how they provide the constraints that people need to induce useful causal models from sparse data.

Similarities between Pavlovian conditioning in nonhumans and causal judgment by humans suggest that similar processes operate in these situations. Notably absent among the similarities is backward blocking (i.e., retrospective devaluation of a signal due to increased valuation of another signal that was present during training), which has been observed in causal judgment by humans but not in Pavlovian responding by animals. The authors used rats to determine if this difference arises from the target cue being biologically significant in the Pavlovian case but not in causal judgment. They used a sensory preconditioning procedure in Experiments 1 and 2, in which the target cue retained low biological significance during the treatment, and obtained backward blocking. The authors found in Experiment 3 that forward blocking also requires the target cue to be of low biological significance. Thus, low biological significance is a necessary condition for a stimulus to be vulnerable to blocking. In recent years, numerous researchers have remarked on the similarity of the conditions that encourage the acquisition of causal relationships in humans and those that foster

Conditioned suppression is a useful technique for assessing whether subjects have learned a CS ± US association, but it is dif ® cult to use in humans because of the need for an aversive US. The purpose of this research was to develop a non-aversive procedure that would produce suppression. Subjects learned to press the space bar of a computer as part of a video game, but they had to stop pressing whenever a visual US appeared, or they would lose points. In Experiment 1, we used an A+/B2 discrimination design: The US always followed Stimulus A and never followed Stimulus B. Although no information about the existence of CSs was given to the subjects, suppression ratio results showed a discrimination learning curveÐ that is, subjects learned to suppress responding in anticipation of the US when Stimulus A was present but not during the presentations of Stimulus B. Experiment 2 explored the potential of this preparation by using two different instruction sets and assessing post-experimental judgements of CS A and CS B in addition to suppression ratios. The results of these experiments suggest that conditioned suppression can be reliably and conveniently used in the human laboratory, providing a bridge between experiments on animal conditioning and experiments on human judgements of causality.

This article proposes an adaptive connectionist model that implements an attributional account of cognitive dissonance. The model represents an attitude as the connection between the attitude object and behavioral-affective outcomes. Dissonance arises when circumstantial constraints induce a mismatch between the model's (mental) prediction and discrepant behavior or affect. Reduction of dissonance by attitude change is accomplished through long-lasting changes in the connection weights using the error-correcting delta learning algorithm. The model can explain both the typical effects predicted by dissonance theory as well as some atypical effects (i.e., reinforcement effect), using this principle of weight changes and by giving a prominent role to affective experiences. The model was implemented in a standard feedforward connectionist network. Computer simulations showed an adequate fit with several classical dissonance paradigms (inhibition, initiation, forced compliance, free choice & misattribution), as well as novel studies that underscore the role of affect. A comparison with an earlier constraint satisfaction approach (Shultz & Lepper, 1996) indicates that the feedforward implementation provides a similar fit with these human data, while avoiding a number of shortcomings of this previous model.

Associative and statistical theories of causal and predictive learning make opposite predictions for situations in which the most recent information contradicts the information provided by older trials (e.g., acquisition followed by extinction). Associative theories predict that people will rely on the most recent information to best adapt their behavior to the changing environment. Statistical theories predict that people will integrate what they have learned in the two phases. The results of this study showed one or the other effect as a function of response mode (trial by trial vs. global), type of question (contiguity, causality, or predictiveness), and postacquisition instructions. That is, participants are able to give either an integrative judgment, or a judgment that relies on recent information as a function of test demands. The authors concluded that any model must allow for flexible use of information once it has been acquired. Learning to predict the events in our environment is critical for survival. Both humans and other animals are known to learn predictive and causal relations between the events in their environment, and the question of how they do it has preoccupied philosophers and psychologists for many years.

Abstract not found

When participants assess the relationship between two variables, each with levels of presence and absence, the two most robust phenomena are that: (a) observing the joint presence of the variables has the largest impact on judgment and observing joint absence has the smallest impact, and (b) participants’ prior beliefs about the variables ’ relationship influence judgment. Both phenomena represent departures from the traditional normative model (the phi coefficient or related measures) and have therefore been interpreted as systematic errors. However, both phenomena are consistent with a Bayesian approach to the task. From a Bayesian perspective: (a) joint presence is normatively more informative than joint absence if the presence of variables is rarer than their absence, and (b) failing to incorporate prior beliefs is a normative error. Empirical evidence is reported showing that joint absence is seen as more informative than joint presence when it is clear that absence of the variables, rather than their presence, is rare.

According to the two-stage model of voluntary action, the ability to perform voluntary action is acquired in two sequential steps. Firstly, associations are acquired between representations of movements and of the e#ects that frequently follow them. Secondly, the anticipation or perception of an acquired action effect primes the movement that has been learnt to produce this effect; the acquired action-e#ect associations thus mediate the selection of actions that are most appropriate to achieve an intended action goal. If action-effect learning has an associative basis, it should be influenced by factors that are known to a#ect instrumental learning, such as the temporal contiguity and the probabilistic contingency of movement and effect. In two experiments, the contiguity or the contingency between key presses and subsequent tones was manipulated in various ways. As expected, both factors affected the acquisition of action-effect relations as assessed by the potency of action effects to prime the corresponding action in a later behavioral test. In particular, evidence of action-effect associations was obtained only if the effect of the action was delayed for no more than 1 s, if the effect appeared more often in the presence than in the absence of the action, or if action and effect were entirely uncorrelated but the effect appeared very often. These findings support the assumption that the control of voluntary actions is based on action-effect representations that are acquired by associative learning mechanisms.

How is information transmitted in a group? A multi-agent connectionist model is proposed that combines features of standard recurrent models to simulate the process of information uptake, integration and memorization within individual agents, with novel aspects that simulate the communication of beliefs and opinions between agents. A crucial aspect in belief updating based on information from other agents is the trust in the information provided, implemented as the consistency with the receiving agents’ existing beliefs. Trust leads to a selective propagation and thus filtering out of less reliable information, and implements Grice’s (1975) maxims of quality and quantity in communication. By studying these communicative aspects within the framework of standard models of information processing, the unique contribution of communicative mechanisms beyond intra-personal factors was explored in simulations of key phenomena involving persuasive communication and polarization, lexical acquisition, spreading of stereotypes and rumors, and a lack of sharing unique information in group decisions.