Causal-based property generalization

The role of causal structure in implicit evaluation

Causal relationships, unlike mere co-occurrence, allow humans to obtain rewards and avoid punishments by intervening on their environment. Accordingly, explicit (controlled) evaluations of stimuli encountered in the environment are known to be sensitive to causal relationships above and beyond mere co-occurrence. In this project, we conduct stringent tests of whether implicit (automatic) evaluation also reflects causal relationships and begin to probe the representational mechanisms underlying such sensitivity. Participants (N = 4836) observed causal events during which two stimuli were equally contingent with positive or negative outcomes but only one of them was causally responsible for these outcomes. Across 6 studies, varying in design and amount of verbal scaffolding provided, differences in causal status consistently guided not only explicit measures of evaluation (Likert and slider scales; Bayes Factor meta-analysis: Cohen's d = 0.28, BF₁₀ > 10⁴⁶) but also their implicit counterparts (Implicit Association Tests; Bayes Factor meta-analysis: Cohen's d = 0.22, BF₁₀ > 10²⁹). However, unlike their explicit counterparts, implicit evaluations were not sensitive to causal relationships that had to be flexibly derived by combining disparate past experiences. Taken together, these studies suggest that implicit evaluations are sensitive to causal information. Such sensitivity seems to be mediated via precompiled, causally informed value representations rather than online computations over a causal model.

How Inductive and Deductive Generalization Shape the Guilt-by-Association Phenomenon Among Firms: Theory and Evidence

This study advances and tests the notion that the phenomenon of guilt by association-- whereby innocent organizations are penalized due to their similarity to offending organizations-- is shaped by two distinct forms of generalization. We analyze how and why evaluators’ interpretative process following instances of corporate misconduct will likely include not only inductive generalization (rooted in similarity judgments and prototype-based categorization) but also deductive generalizing (rooted in evaluators’ theories and causal-based categorization). We highlight the role and relevance of this neglected distinction by extending guilt-by-association predictions to include two unique predictions based on deductive generalization. First, we posit a recipient effect: if an innocent organization falls under a negative stereotype that causally links the innocent firm with corporate misconduct, then that innocent firm will suffer a greater negative spillover effect, irrespective of its similarity to the offending firm. Second, we also posit a transmission effect: if the offending firm falls under the same negative stereotype, then the negative spillover effect to other similar firms will be lessened. We also analyze how media discourse can foster negative stereotypes, and thus amplify these two effects. We find support for our hypotheses in an analysis of stock market reactions to corporate misconduct for all U.S. and international firms using reverse mergers to gain publicly traded status in the United States. We discuss the implications of our theoretical perspective and empirical findings for research on corporate misconduct, guilt by association, and stock market prejudice.

Beyond Markov: Accounting for independence violations in causal reasoning

Although many theories of causal cognition are based on causal graphical models, a key property of such models-the independence relations stipulated by the Markov condition-is routinely violated by human reasoners. This article presents three new accounts of those independence violations, accounts that share the assumption that people's understanding of the correlational structure of data generated from a causal graph differs from that stipulated by causal graphical model framework. To distinguish these models, experiments assessed how people reason with causal graphs that are larger than those tested in previous studies. A traditional common cause network (Y₁←X→Y₂) was extended so that the effects themselves had effects (Z₁←Y₁←X→Y₂→Z₂). A traditional common effect network (Y₁→X←Y₂) was extended so that the causes themselves had causes (Z₁→Y₁→X←Y₂←Z₂). Subjects' inferences were most consistent with the beta-Q model in which consistent states of the world-those in which variables are either mostly all present or mostly all absent-are viewed as more probable than stipulated by the causal graphical model framework. Substantial variability in subjects' inferences was also observed, with the result that substantial minorities of subjects were best fit by one of the other models (the dual prototype or a leaky gate models). The discrepancy between normative and human causal cognition stipulated by these models is foundational in the sense that they locate the error not in people's causal reasoning but rather in their causal representations. As a result, they are applicable to any cognitive theory grounded in causal graphical models, including theories of analogy, learning, explanation, categorization, decision-making, and counterfactual reasoning. Preliminary evidence that independence violations indeed generalize to other judgment types is presented.

Inductive reasoning 2.0

Inductive reasoning entails using existing knowledge to make predictions about novel cases. The first part of this review summarizes key inductive phenomena and critically evaluates theories of induction. We highlight recent theoretical advances, with a special emphasis on the structured statistical approach, the importance of sampling assumptions in Bayesian models, and connectionist modeling. A number of new research directions in this field are identified including comparisons of inductive and deductive reasoning, the identification of common core processes in induction and memory tasks and induction involving category uncertainty. The implications of induction research for areas as diverse as complex decision-making and fear generalization are discussed. This article is categorized under: Psychology > Reasoning and Decision Making Psychology > Learning.

Is there an exemplar theory of concepts?

It is common to describe two main theories of concepts: prototype theories, which rely on some form of summary description of a category, and exemplar theories, which claim that concepts are represented as remembered category instances. This article reviews a number of important phenomena in the psychology of concepts, arguing that they have no proposed exemplar explanation. In some of these cases, it is difficult to see how an exemplar theory would be adequate. The article concludes that exemplars are certainly important in some categorization judgments and in category-learning experiments, but that there is no exemplar theory of human concepts in a broad sense.

Reasoning With Causal Cycles

This article assesses how people reason with categories whose features are related in causal cycles. Whereas models based on causal graphical models (CGMs) have enjoyed success modeling category-based judgments as well as a number of other cognitive phenomena, CGMs are only able to represent causal structures that are acyclic. A number of new formalisms that allow cycles are introduced and evaluated. Dynamic Bayesian networks (DBNs) represent cycles by unfolding them over time. Chain graphs augment CGMs by allowing the presence of undirected links that model feedback relations between variables. Unfolded chain graphs are chain graphs that unfold over time. An existing model of causal cycles (alpha centrality) is also evaluated. Four experiments in which subjects reason about categories with cyclically related features provided evidence against DBNs and alpha centrality and for the two types of chain graphs. Chain graphs-a mechanism for representing the equilibrium distribution of a dynamic system-may thus be good candidates for modeling how people reason causally with complex systems. Applications of chain graphs to areas of cognition other than category-based judgments are discussed.

Categories, concepts, and conditioning: how humans generalize fear

During the past century, Pavlovian conditioning has served as the predominant experimental paradigm and theoretical framework to understand how humans learn to fear and avoid real or perceived dangers. Animal models for translational research offer insight into basic behavioral and neurophysiological factors mediating the acquisition, expression, inhibition, and generalization of fear. However, it is important to consider the limits of traditional animal models when applied to humans. Here, we focus on the question of how humans generalize fear. We propose that to understand fear generalization in humans requires taking into account research on higher-level cognition such as category-based induction, inferential reasoning, and representation of conceptual knowledge. Doing so will open the door for productive avenues of new research.

Rational hypocrisy: a Bayesian analysis based on informal argumentation and slippery slopes

Moral hypocrisy is typically viewed as an ethical accusation: Someone is applying different moral standards to essentially identical cases, dishonestly claiming that one action is acceptable while otherwise equivalent actions are not. We suggest that in some instances the apparent logical inconsistency stems from different evaluations of a weak argument, rather than dishonesty per se. Extending Corner, Hahn, and Oaksford's (2006) analysis of slippery slope arguments, we develop a Bayesian framework in which accusations of hypocrisy depend on inferences of shared category membership between proposed actions and previous standards, based on prior probabilities that inform the strength of competing hypotheses. Across three experiments, we demonstrate that inferences of hypocrisy increase as perceptions of the likelihood of shared category membership between precedent cases and current cases increase, that these inferences follow established principles of category induction, and that the presence of self-serving motives increases inferences of hypocrisy independent of changes in the actions themselves. Taken together, these results demonstrate that Bayesian analyses of weak arguments may have implications for assessing moral reasoning.

Causal knowledge and the development of inductive reasoning

We explored the development of sensitivity to causal relations in children's inductive reasoning. Children (5-, 8-, and 12-year-olds) and adults were given trials in which they decided whether a property known to be possessed by members of one category was also possessed by members of (a) a taxonomically related category or (b) a causally related category. The direction of the causal link was either predictive (prey→predator) or diagnostic (predator→prey), and the property that participants reasoned about established either a taxonomic or causal context. There was a causal asymmetry effect across all age groups, with more causal choices when the causal link was predictive than when it was diagnostic. Furthermore, context-sensitive causal reasoning showed a curvilinear development, with causal choices being most frequent for 8-year-olds regardless of context. Causal inductions decreased thereafter because 12-year-olds and adults made more taxonomic choices when reasoning in the taxonomic context. These findings suggest that simple causal relations may often be the default knowledge structure in young children's inductive reasoning, that sensitivity to causal direction is present early on, and that children over-generalize their causal knowledge when reasoning.

Mechanistic beliefs determine adherence to the Markov property in causal reasoning

What kind of information do people use to make predictions? Causal Bayes nets theory implies that people should follow structural constraints like the Markov property in the form of the screening-off rule, but previous work shows little evidence that people do. We tested six hypotheses that attempt to explain violations of screening off, some by asserting that people use mechanistic knowledge to infer additional latent structure. In three experiments, we manipulated whether the causal relations among variables within a causal structure were supported by the same or different mechanisms. The experiments differed in the type of causal structures (common cause vs. chain), the way that causal structures were presented (verbal description vs. observational learning), how the mechanisms were presented (explicit description vs. implicit description vs. visual hint), and the number of predictions requested (2 vs. 24). The results revealed that the screening-off rule was violated more often when the mechanisms were the same than when they were different. The findings suggest that people use knowledge about underlying mechanisms to infer latent structure for prediction.

Conceptual influences on category-based induction

One important function of categories is to permit rich inductive inferences. Prior work shows that children use category labels to guide their inductive inferences. However, there are competing theories to explain this phenomenon, differing in the roles attributed to conceptual information vs. perceptual similarity. Seven experiments with 4- to 5-year-old children and adults (N=344) test these theories by teaching categories for which category membership and perceptual similarity are in conflict, and varying the conceptual basis of the novel categories. Results indicate that for non-natural kind categories that have little conceptual coherence, children make inferences based on perceptual similarity, whereas adults make inferences based on category membership. In contrast, for basic- and ontological-level categories that have a principled conceptual basis, children and adults alike make use of category membership more than perceptual similarity as the basis of their inferences. These findings provide evidence in favor of the role of conceptual information in preschoolers' inferences, and further demonstrate that labeled categories are not all equivalent; they differ in their inductive potential.

An integrated account of generalization across objects and features

Humans routinely make inductive generalizations about unobserved features of objects. Previous accounts of inductive reasoning often focus on inferences about a single object or feature: accounts of causal reasoning often focus on a single object with one or more unobserved features, and accounts of property induction often focus on a single feature that is unobserved for one or more objects. We explore problems where people must make inferences about multiple objects and features, and propose that people solve these problems by integrating knowledge about features with knowledge about objects. We evaluate three computational methods for integrating multiple systems of knowledge: the output combination approach combines the outputs produced by these systems, the distribution combination approach combines the probability distributions captured by these systems, and the structure combination approach combines a graph structure over features with a graph structure over objects. Three experiments explore problems where participants make inferences that draw on causal relationships between features and taxonomic relationships between animals, and we find that the structure combination approach provides the best account of our data.

Taking the rationality out of probabilistic models

Rational models vary in their goals and sources of justification. While the assumptions of some are grounded in the environment, those of others – which I label probabilistic models – are induced and so require more traditional sources of justification, such as generalizability to dissimilar tasks and making novel predictions. Their contribution to scientific understanding will remain uncertain until standards of evidence are clarified.

Category transfer in sequential causal learning: the unbroken mechanism hypothesis

The goal of the present set of studies is to explore the boundary conditions of category transfer in causal learning. Previous research has shown that people are capable of inducing categories based on causal learning input, and they often transfer these categories to new causal learning tasks. However, occasionally learners abandon the learned categories and induce new ones. Whereas previously it has been argued that transfer is only observed with essentialist categories in which the hidden properties are causally relevant for the target effect in the transfer relation, we here propose an alternative explanation, the unbroken mechanism hypothesis. This hypothesis claims that categories are transferred from a previously learned causal relation to a new causal relation when learners assume a causal mechanism linking the two relations that is continuous and unbroken. The findings of two causal learning experiments support the unbroken mechanism hypothesis.

Rich in vitamin C or just a convenient snack? Multiple-category reasoning with cross-classified foods

Two studies examined multiple category reasoning in property induction with cross-classified foods. Pilot tests identified foods that were more typical of a taxonomic category (e.g., "fruit"; termed 'taxonomic primary') or a script based category (e.g., "snack foods"; termed 'script primary'). They also confirmed that taxonomic categories were perceived as more coherent than script categories. In Experiment 1 participants completed an induction task in which information from multiple categories could be searched and combined to generate a property prediction about a target food. Multiple categories were more often consulted and used in prediction for script primary than for taxonomic primary foods. Experiment 2 replicated this finding across a range of property types but found that multiple category reasoning was reduced in the presence of a concurrent cognitive load. Property type affected which categories were consulted first and how information from multiple categories was weighted. The results show that multiple categories are more likely to be used for property predictions about cross-classified objects when an object is primarily associated with a category that has low coherence.

Inductive reasoning

Inductive reasoning entails using existing knowledge or observations to make predictions about novel cases. We review recent findings in research on category-based induction as well as theoretical models of these results, including similarity-based models, connectionist networks, an account based on relevance theory, Bayesian models, and other mathematical models. A number of touchstone empirical phenomena that involve taxonomic similarity are described. We also examine phenomena involving more complex background knowledge about premises and conclusions of inductive arguments and the properties referenced. Earlier models are shown to give a good account of similarity-based phenomena but not knowledge-based phenomena. Recent models that aim to account for both similarity-based and knowledge-based phenomena are reviewed and evaluated. Among the most important new directions in induction research are a focus on induction with uncertain premise categories, the modeling of the relationship between inductive and deductive reasoning, and examination of the neural substrates of induction. A common theme in both the well-established and emerging lines of induction research is the need to develop well-articulated and empirically testable formal models of induction. Copyright © 2010 John Wiley & Sons, Ltd. For further resources related to this article, please visit the WIREs website.