How Do People Generalize Causal Relations over Objects? A Non-parametric Bayesian Account

How do people decide how general a causal relationship is, in terms of the entities or situations it applies to? What features do people use to decide whether a new situation is governed by a new causal law or an old one? How can people make these difficult judgments in a fast, efficient way? We address these questions in two experiments that ask participants to generalize from one (Experiment 1) or several (Experiment 2) causal interactions between pairs of objects. In each case, participants see an agent object act on a recipient object, causing some changes to the recipient. In line with the human capacity for few-shot concept learning, we find systematic patterns of causal generalizations favoring simpler causal laws that extend over categories of similar objects. In Experiment 1, we find that participants’ inferences are shaped by the order of the generalization questions they are asked. In both experiments, we find an asymmetry in the formation of causal categories: participants preferentially identify causal laws with features of the agent objects rather than recipients. To explain this, we develop a computational model that combines program induction (about the hidden causal laws) with non-parametric category inference (about their domains of influence). We demonstrate that our modeling approach can both explain the order effect in Experiment 1 and the causal asymmetry, and outperforms a naïve Bayesian account while providing a computationally plausible mechanism for real-world causal generalization.

How do people generalize causal relations over objects? A non-parametric Bayesian account

How do people decide how general a causal relationship is, in terms of the entities or situations it applies to? What features do people use to decide whether a new situation is governed by a new causal law or an old one? How can people make these difficult judgments in a fast, efficient way? We address these questions in two experiments that ask participants to generalize from one (Experiment 1) or several (Experiment 2) causal interactions between pairs of objects. In each case, participants see an agent object act on a recipient object, causing some changes to the recipient. In line with the human capacity for few-shot concept learning, we find systematic patterns of causal generalizations favoring simpler causal laws that extend over categories of similar objects. In Experiment 1, we find that participants’ inferences are shaped by the order of the generalization questions they are asked. In both experiments, we find an asymmetry in the formation of causal categories: participants preferentially identify causal laws with features of the agent objects rather than recipients. To explain this, we develop a computational model that combines program induction (about the hidden causal laws) with non-parametric category inference (about their domains of influence). We demonstrate that our modeling approach can both explain the order effect in Experiment 1 and the causal asymmetry, and outperforms a naive Bayesian account while providing a computationally plausible mechanism for real world causal generalization.

Geography and the theory of uneven and combined development: Theorizing uniqueness and the return of China

As societal interaction and combination play a vital role in shaping spatio-temporal development paths, meta-theories of uneven development should give way to a relational meta-theory of uneven and combined development (U&CD). U&CD examines at multiple levels of abstraction not just internal causal mechanisms governing the trajectories of individual societies but also causal mechanisms deriving from societal interaction in a world of multiple unevenly developed societies and multiple development pathways. As a consequence it helps explain geographical differentiation and the multiplicity, hybridity and multilinearity of processes of development. As U&CD examines the roles of ideal and efficient causes and causal laws, it also entails revitalized social science and political economy approaches in geography and urban and regional development studies. To demonstrate the indispensability of analyses of the role of societal interaction and the explanatory significance of the theory of U&CD, a series of longue-durée vignettes explore their role in explanations of the spatio-temporality of the decline and subsequent return of China in a changing and interdependent world.

A short history of theories of intuive theories

Intuitive theories are sets of integrated concepts and causal laws that people adopt to comprehend, explain, and predict certain phenomena they encounter in the world. These theories are ‘intuitive’ because they are thought to drive our intuitions about how the physical and biological world, the mental life of people, and the society we live in work, without meeting the standards of explicit scientific theorizing. The proposal that people adopt such theories has been around at least since the 1970s. However, how psychologists think about intuitive theories has been changing since they have been first proposed. In this chapter, we provide a short overview of the approaches to the function of intuitive theories and belief-forming systems more generally. While early characterization of intuitive theories emphasized their epistemic function, later attempts took an evolutionary view, claiming that they serve adaptive functions that are not always aligned with the goal of accurately tracking environmental states. A recent twist in this story is the proposal that shared intuitive theories may also serve social functions by providing a ‘theoretical common ground’ on which people interpret unobservable entities, such as memories, character traits, entitlements, and obligations. Such shared theories might be essential for social coordination via communication.

Kinesthetic Unity as Motivated Association

SummaryWithin Husserl’s theory of perception, the role attributed to kinesthetic sensations determines a phase of the perceptive constitution that marks the boundary between pure receptivity and a first form of self-determination of consciousness. Kinesthetic experiences are, in fact, characterized not just as acts that are performed but rather that can be performed, albeit according to predetermined paths.This primitive form of ‘instinctive’ spontaneity of the Ego (linked to primal impulses) as realization of pre-established potentialities, characterizes what Husserl defines the ‘ idiopsychic’ dimension of consciousness (Husserl, 1952, p. 135). However, although this level of consciousness unity presupposes a spontaneous activity, it can be investigated according to the ‘causal’ laws of motivation.The phenomenon of motivation was notoriously introduced by Husserl in §56 of Ideen II, as a specific law of spiritual life. However, there are two possible forms of motivation, one in which the Ego is actively involved, and a second one, called “associative motivation.” The latter basically indicates the passive tendency of creating associations between unities of the immanent sphere. In other terms, Husserl acknowledges the existence of “motivated relations” within the immanent sphere of mental acts which do not necessarily call for an active participation of the Ego. In this sense, the relation between motivating factors and motivated elements could be considered a kind of conditioning of the form “because-therefore,” in which the two elements arrange themselves in a succession of experiences. This work aims to show that this very kind of association is the same that pre-determines the unfolding possibility of kinesthetic chains.

Statistical Law and Human Freedom

This chapter evaluates the criticism of statistics. Already in the early nineteenth century, the statistical approach was attacked on the ground that mere statistical tables cannot demonstrate causality, or that mathematical probability presupposes the occurrence of events wholly by chance. The intent of these early critics was not to suggest the inadequacy of causal laws in social science, but to reject the scientific validity of statistics. The new interpretation of statistics that emerged during the 1860s and 1870s was tied to a view of society in which variation was seen as much more vital. Statistical determinism became untenable precisely when social thinkers who used numbers became unwilling to overlook the diversity of the component individuals in society, and hence denied that regularities in the collective society could justify any particular conclusions about its members. These social discussions on natural science and philosophy bore fruit in the growing interest in the analysis of variation evinced by the late-century mathematical statisticians. To be sure, Francis Galton gave little attention to the debates on human freedom, but Francis Edgeworth was closely familiar with them, and Wilhelm Lexis's important work on dispersion can only be understood in the context of this tradition.

Emergent Causal Laws and Physical Laws

Contrasting accounts of physicalism and strong emergentism face two problems. According to the neutrality problem, contrasting supervenience-based formulations of these positions cannot be neutral with respect to certain unrelated metaphysical commitments. According to the collapse problem, emergent properties can be accounted for using an appropriately expansive physical ontology, rendering strong emergentism metaphysically suspect. I argue that both these problems can be solved with a principled distinction between emergent causal laws and physical laws. I propose such a distinction based on a finite discontinuity in the behavior of fundamental physical constituents as a function of complexity.

Sticking to the evidence? A behavioral and computational case study of micro-theory change in the domain of magnetism.

Constructing an intuitive theory from data confronts learners with a “chicken-and-egg” problem: the laws can only be expressed in terms of the theory’s core concepts, but these concepts are only meaningful in terms of the role they play in the theory’s laws; how can a learner discover appropriate concepts and laws simultaneously, knowing neither to begin with? We explore how children can solve this chicken-and-egg problem in the domain of magnetism, drawing on perspectives from computational modeling and behavioral experiments. We present four- and five-year-olds with two different simplified magnet-learning tasks. Children appropriately constrain their beliefs to two hypotheses following ambiguous but informative evidence. Following a critical intervention they learn the correct theory. In the second study, children infer the correct number of categories given no information about the possible causal laws. Children’s hypotheses in these tasks are explained as rational inferences within a Bayesian computational framework.

Under Ideal Conditions

In this chapter, we discuss how reality is messy. Most events are not causally connected. Those that are, though, are important because they are a basis for prediction. It is useful, then, to have causal laws. The problem is that they all seem to have exceptions, and thus need to be ceteris paribus qualified. We might still think that if we could put a cause in ideal conditions, such as in a laboratory, its effect would follow as a matter of necessity. We find, however, that even this cannot be empirically proven. Any perfect regularity thus remains a philosophical postulate.

Same Cause, Same Effect

Hume gave the classic formulation of the regularity theory. The causal laws are fixed by the pattern of regularity in events. We have a conceptual constraint on the notion of causal law: the effects that laws describe must be repeatable and robust, applying at every time and place. From the same cause, therefore, we can infer the same effect. And by modus tollens, from a different effect, we infer a different cause. Apparent counterexamples can be marginalized, treated as exceptions, outliers, non-respondents, or effects of background noise or interference. This suggests that scientists accept the broadly Humean notion of causation as regularity.