Search Results

This chapter focuses on the role of causal models in judgment. It is argued that causal modeling plays a central role in the process of judgment when the object of judgment can be construed as a causal effect. Such a construal is almost always appropriate in the legal domain where both crimes and accidents are effects of individual actions. It is also appropriate in scientific domains. Scientists are also in the business of building causal models, in their case to understand how the world works in general rather than to understand the circumstances of a specific event. Causal models are relevant to judgment in any domain in which physical, social, or abstract events cause other events. Causal models may well be the primary determinant of what is considered relevant when reasoning, when making judgments and predictions, and when taking action within such domains.

This chapter offers a selection of theories of causal learning. Some of the theories come out of psychology, while others come out of rational analyses of causal learning. All tend to focus on how people use correlations — information about which events go together — to figure out what is causing what. A number of other, supporting pieces of information about what causes what — a number of other cues to causal structure — that people are responsive to are described. The costs and benefits of allowing learners to actively intervene on the system they are learning are discussed.

This chapter focuses on how we induce properties of the world. Some psychologists have been coming back to the view that induction is mediated by causal models — causal models that are often generated online through the application of causal principles, abstract causal relations that have general applicability. Such causal models help to explain how people make inductive inferences when the inference can be conceived as a causal effect, as in the ‘bananas have it; therefore, monkeys have it’ example. In other cases, inference involves analogy: a predicate is applied to one category because it is known to apply to an analogous category, as in the ‘tigers do it; therefore, hawks do it’ example. In such cases, the analogy seems to be between causal structures. Finally, causal models give a psychologically plausible way to think about why people sometimes show sensitivity to statistical information. Instead of assuming that people calculate statistics like measures of the variability of a property, the requisite information can be interpreted as a property's centrality in a causal model.

This concluding chapter attempts to draw out the general lessons of causal and interventional principles for an understanding of people, their situations, their problems, and solutions to their problems. Topics discussed include the role of causal models in how the mind works, fundamental claims of the causal modelling framework, what causal models can contribute to human welfare, and the causal modelling metaphor.

Developments in M/EEG analysis allows for models that are sophisticated enough to capture the full richness of the data. This chapter focuses on dynamic causal modeling (DCM) for M/EEG, which entails the inversion of informed spatiotemporal models of observed responses. The idea is to model condition-specific responses over channels and peristimulus time with the same model, where the differences among conditions are explained by changes in only a few key parameters. The face and predictive validity of DCM have been established, which makes it a potentially useful tool for group studies.

This chapter describes repertoire-based methods for detecting and quantifying the social transmission of behavior based on a “snapshot” of the behavioral repertoires of individuals or groups. Repertoire-based methods often take the form of a group contrast approach, where the researcher attempts to ascertain whether different groups have different behavioral repertoires, which might be caused by a higher rate of social transmission within groups than between them. The chapter first considers approaches that can be applied to determine whether group differences in behavior exist, including the group contrasts approach and the method of exclusion. In particular, it discusses methods for assessing the genetic hypothesis and the ecological hypothesis. It also presents a model-fitting approach and a causal modeling framework. Finally, it highlights the limitations of studying social learning based solely on differences in repertoires.

This chapter compares two causal models used in qualitative and quantitative research: an additive-linear model and a set-theoretic model. The additive-linear causal model is common in the statistical culture, whereas the set-theoretic model is often used (implicitly) in the qualitative culture. After providing an overview of the two causal models, the chapter considers the main differences between them. It then gives an example to illustrate how a set-theoretic causal model is implicitly used in the within-case analysis of a specific outcome. It also explains how the form of causal complexity varies across the quantitative and qualitative paradigms. Finally, it examines another difference between the causal models used in quantitative and qualitative research, one that revolves around the concept of “equifinality” or “multiple causation.” The chapter suggests that while the two causal models are quite different, neither is a priori correct.

This chapter discusses how we use aspects of language. It shows that we use causal models to produce and understand speech even when we have no idea that we are doing so. Two aspects of language are considered that illustrate this usage: what pronouns refer to and the meaning of conjunctions like and if. The meaning of if is especially complicated, and the details of the causal model framework are particularly helpful for understanding it.

Causal induction has two components: learning about the structure of causal models and learning about causal strength and other quantitative parameters. This chapter argues for several interconnected theses. First, people represent causal knowledge qualitatively, in terms of causal structure; quantitative knowledge is derivative. Second, people use a variety of cues to infer causal structure aside from statistical data (e.g. temporal order, intervention, coherence with prior knowledge). Third, once a structural model is hypothesized, subsequent statistical data are used to confirm, refute, or elaborate the model. Fourth, people are limited in the number and complexity of causal models that they can hold in mind to test, but they can separately learn and then integrate simple models, and revise models by adding and removing single links. Finally, current computational models of learning need further development before they can be applied to human learning.

This chapter first discusses experimental findings showing that multisensory perception encompasses a spectrum of phenomena ranging from full integration (or fusion), to partial integration, to complete segregation. Next, it describes two Bayesian causal-inference models that can account for the entire range of combinations of two or more sensory cues. It shows that one of these models, which is a hierarchical Bayesian model, is a special form of the other one (which is a nonhierarchical model). It then compares the predictions of these models with human data in multiple experiments and shows that Bayesian causal-inference models can account for the human data remarkably well. Finally, a study is presented that investigates the stability of priors in the face of drastic change in sensory conditions.

Pearl's work on causation has helped focus new attention on the nature of causal reasoning and causal inference in behavioural science. Pearl takes an axiomatic approach, presenting axioms as first principles, but these may be better understood as boundary conditions for the application of the theory. Pearl adopts a non‐eliminative but instrumental approach to causation which creates some tension with the tradition of ruling out rival hypotheses in the behavioural sciences. Finally, much causal reasoning in the behavioural sciences involves reasoning across possible worlds that differ in their causal structure, which becomes awkward within the basic architecture of Pearl's system. A neighbourhood semantics approach could represent this type of reasoning more naturally. Consideration of these issues may be helpful both to behavioural scientists working to incorporate Pearl's work and also to those working outside the behavioural sciences attempting to explain causal reasoning within those sciences.

The numbers published in physics are steppingstones for further inquiry, because they are few and other researchers often make use of them. The numbers published in social sciences are endpoints, because they are profligate and, once printed, hardly anyone makes use of them. Astronomy could not develop without overcoming the Ptolemaic syndrome of reducing all motion to circular. Social sciences must overcome their syndrome of reducing all relationships to linear. Physicists, like people in general, start with causal models, while social scientists stand apart by starting with empirical models.

This chapter shows that the quantitative and qualitative cultures differ on the issue of symmetry. Whereas quantitative research tends to analyze relationships that are symmetric, qualitative research focuses on relationships that have asymmetric qualities. Causal models and explanations can be asymmetric in a variety of ways. This chapter deals mainly (though not exclusively) on the so-called “static causal asymmetry,” in which the explanation of occurrence is not the mirror image of that of nonoccurrence. After comparing symmetric and asymmetric models, the chapter looks at examples of asymmetric explanations using set-theoretic causal models. It highlights the difficulty of translating the fundamental symmetry of standard statistical models into the basic asymmetry of set-theoretic models, as well as the difficulty of capturing the asymmetry of set-theoretic models with the standard symmetric tools of statistics.

In this chapter, we distinguish between two ways in which counterfactual and causal judgements might be linked. According to a psychological relatedness view, counterfactual and causal judgements are viewed as psychologically related and expected to be consistent with each other, whereas according to a counterfactual process view, counterfactual thought is thought to be actually involved in the process of making causal judgements. Our research with young children is discussed in terms of whether it provides support for either of these views. The findings of studies in which children were asked to make counterfactual judgements about the effects of intervening on a causal system suggest that causal and counterfactual judgements are not necessarily consistent in children. However, the findings of our studies in which children judge whether an object possesses a causal power provided some evidence for a link between causal and counterfactual judgements. We discuss whether counterfactual reasoning may actually be involved in the process of making certain types of simple causal judgements, in tasks examining cue competition effects.

How do people acquire and use causal knowledge? This chapter argues that causal learning and reasoning are intertwined, and recruit similar representations and inferential procedures. In contrast to covariation‐based approaches to learning, this chapter maintains that people use multiple sources of evidence to discover causal relations, and that the causal representation itself is separate from these informational sources. The key roles of prior knowledge and interventions in learning are also discussed. Finally, this chapter speculates about the role of mental simulation in causal inference. Drawing on parallels with work in the psychology of mechanical reasoning, the notion of a causal mental model is proposed as a viable alternative to reasoning systems based in logic or probability theory alone. The central idea is that when people reason about causal systems they utilize mental models that represent objects, events or states of affairs, and reasoning and inference is carried out by mental simulation of these models.

Many different, seemingly mutually exclusive, theories of categorization have been proposed in recent years. The most notable theories have been those based on prototypes, exemplars, and causal models. This chapter provides “representation theorems” for each of these theories in the framework of probabilistic graphical models. More specifically, it shows for each of these psychological theories that the categorization judgments predicted and explained by the theory can be wholly captured using probabilistic graphical models. In other words, probabilistic graphical models provide a lingua franca for these disparate categorization theories, and so we can quite directly compare the different types of theories. These formal results are used to explain a variety of surprising empirical results, and to propose several novel theories of categorization.

This chapter applies the ideas about intervention and invariance developed in previous chapters to so-called causal models of the sort used in the social, behavioral, and biomedical sciences. Both regression models and structural equation models are explored and a modularity condition is defended in connection with the latter. Connections with the Causal Markov condition are also briefly discussed.