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.

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

In this article we describe some of the experimental software we have developed for the study of associative human learning and memory. All these programs have the appearance of very simple video games. Some of them use the participants ’ behavioral responses to certain stimuli during the game as a dependent variable for measuring their learning of the target cue-outcome associations. Some others explicitly ask participants to rate the degree of relationship they perceive between the cues and the outcomes. These programs are implemented in Web pages using JavaScript, which allows their use both in traditional laboratory experiments as well as in Internet-based experiments. The psychology of learning is a research area that has usually been investigated with nonhuman animals and in which, traditionally, there existed too many procedural and ethical problems to conduct experiments with humans. However, human learning is today a flourishing research area in which many interesting effects are being reported around the world (see, e.g., De Houwer &

Several studies have shown that predictive and causal judgments vary depending on whether the question used to assess the relationship between events is presented after each piece of information or only after all the available information has been observed. This effect could be understood by assuming that in the two cases people perceive that the test question requires that different sets of evidence be taken into account. This hypothesis is tested in the present experiments through contextual manipulations that take place at the time of training and at the time of test. Our results show that people use this contextual information to infer which set of events should be considered when making their subjective assessments. The results are at odds with current theoretical approaches, but it is possible to develop mechanisms that would allow these models to account for the observed evidence. Learning to predict future events from present events is one of the most powerful adaptive tools, since it allows an organism to find the necessary resources for survival and to avoid dangerous situations. Given its importance, this kind of predictive learning was the central focus of animal behavior research throughout the twentieth century. During the last decades, predictive learning has also become important in the area of human cognition, where it has given rise to a great amount of empirical and theoretical research. The vast amount of evidence provided by this research has sometimes turned out to be quite difficult to explain by the available theoretical approaches. Many variables usually neglected by theoretical models influence the process of human learning of predictive relations among events or the way in which humans use the acquired information. Among other things, it has been shown that the probe question used to assess participants ’ judgment (Matute,

Abstract not found

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 intuitive distinction between causal structure and causal strength: the difference between asking whether or not a causal relationship exists, and asking how strong that causal relationship might be. We show that the 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 provides a better account of a large number of existing datasets than either #P or causal power. It also predicts several phenomena that cannot be accounted for by other 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.