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.

computation. In preparation. Address for correspondence:

Associative learning theory postulates two main determinants for human causal learning: contingency and contiguity. In line with such an account, participants in Shanks, Pearson, and Dickinson (1989) failed to discover causal relations involving delays of more than two seconds. More recent research has shown that the impact of contiguity and delay is mediated by prior knowledge about the timeframe of the causal relation in question. Buehner and May (2002, 2003) demonstrated that the detrimental effect of delay can be significantly reduced if reasoners are aware of potential delays. Here we demonstrate for the first time that the negative influence of delay can be abolished completely by a subtle change in the experimental instructions. Temporal contiguity is thus not essential for human causal learning. An associative learning analysis of human causal learning postulates two main determinants of judged causal strength: the contingency and the contiguity between the potential cause (cue) and the effect (outcome) (e.g., see Shanks & Dickinson, 1987). Empirical research in the last decades has largely focused on the congruency between cue–outcome contingency and judged causal strength. Early reports suggested that human causal judgements closely track variations in cue-outcome contingency (e.g., Jenkins & Ward, 1965), while more recent studies revealed a more complex picture (e.g., Chapman & Robbins, 1990). In fact, the theoretical and empirical relations between contingency and judged causality are still the subject of a hot debate

models provided the best account of data generated in tasks that require human observers to judge the relationship between binary events. In the intervening years, new data have been reported that provide evidence for higherorder processes. Some have argued that these new data pose a serious threat to the viability of the associative account. The purpose of the present paper is to review this evidence and to assess the severity of this threat. In 1978, Brooks described the interaction between analytic and nonanalytic processes, and argued that “there are many factors that push a person’s strategy toward one end of the scale or another – that is, toward learning individuals by codings that are designed to retain the item’s individuality, or toward tracking the validity of characteristics of the stimulus

Two experiments are reported that examine the role of temporal contiguity on judgments of contingency in a human analogue of the Pavlovian task. The data show that the effect of the actual delay on contingency judgment depends on the observers expectation regarding the delay. For a fixed contingency between the cue and the outcome, ratings of the contingency are higher when the actual delay is congruent with the observers expectation than when it is incongruent. We argue that our data can be understood within the context of the temporal coding hypothesis. There is considerable evidence of similarities between the operations that modulate the strength of conditioning in nonhuman animals and those that modulate the rating of the contingency between events by humans (see Allan, 1993). One of these similarities is the effect of temporal contiguity. It is well established in the animal literature that temporal contiguity is an important variable in both instrumental and Pavlovian conditioning (see Allan, Balsam, Church, & Terrace, 2002; Allan & Church, 2002). For example, increasing the delay between a response and reinforcement in an instrumental task decreases the rate of responding. Similarly, increasing the delay between a conditioned stimulus and an unconditioned stimulus in a Pavlovian task retards the acquisition of the conditioned response. The studies that have examined the effect of temporal contiguity on ratings of contingency have used human analogues of the animal instrumental procedure (e.g., Buehner & May,

Three experiments investigated the impact of delay on human causal learning. We present a new paradigm based on the presentation of continuous event streams, and use it to test two hypotheses drawn from associative learning theories of causal inference. Unlike free-operant procedures traditionally used to study temporal aspects of causal learning (Shanks, Pearson, & Dickinson, 1989; Shanks & Dickinson, 1987; Buehner & May, 2002, 2003, 2004), the procedure employed here allows full control over all aspects of stimulus delivery while at the same time overcoming the ecologically invalid notion of discrete learning trials. Results show that delays generally impair causal learning, but prior knowledge and experience mediate this detrimental effect. In accordance with associative learning theory, pre-exposure to an unreinforced background context facilitates the discovery of delayed causal relationships. However, contrary to associative learning theory, increasing the amount of experience with a delayed causal relationship does not improve discovery. Implications for associative learning and causal model theories are discussed.

In multiple‐cue learning people acquire information about cue‐outcome relations and combine these into predictions or judgments. Previous studies claim that people can achieve high levels of performance without explicit knowledge of the task structure or insight into their own judgment policies. It has also been argued that people use a variety of suboptimal strategies to solve such tasks. In two experiments we re‐examined these conclusions by introducing novel measures of task knowledge and self‐insight, and using ‘rolling regression ’ methods to analyze individual learning. Participants successfully learned a four‐cue probabilistic environment and showed accurate knowledge of both the task structure and their own judgment processes. Learning analyses suggested that the apparent use of suboptimal strategies emerges from the incremental tracking of statistical contingencies in the environment. These findings have wide repercussions for the study of multicue learning in both normal and patient populations. Insight 3