Non-bayesian inference: causal structure trumps correlation

The study tests the hypothesis that conditional probability judgments can be influenced by causal links between the target event and the evidence even when the statistical relations among variables are held constant. Three experiments varied the causal structure relating three variables and found that (a) the target event was perceived as more probable when it was linked to evidence by a causal chain than when both variables shared a common cause; (b) predictive chains in which evidence is a cause of the hypothesis gave rise to higher judgments than diagnostic chains in which evidence is an effect of the hypothesis; and (c) direct chains gave rise to higher judgments than indirect chains. A Bayesian learning model was applied to our data but failed to explain them. An explanation-based hypothesis stating that statistical information will affect judgments only to the extent that it changes beliefs about causal structure is consistent with the results.     

Beyond the Information Given: Causal Models in Learning and Reasoning

ABSTRACT— The philosopher David Hume's conclusion that causal induction is solely based on observed associations still presents a puzzle to psychology. If we only acquired knowledge about statistical covariations between observed events without accessing deeper information about causality, we would be unable to understand the differences between causal and spurious relations, between prediction and diagnosis, and between observational and interventional inferences. All these distinctions require a deep understanding of causality that goes beyond the information given. We report a number of recent studies that demonstrate that people and rats do not stick to the superficial level of event covariations but reason and learn on the basis of deeper causal representations. Causal-model theory provides a unified account of this remarkable competence.     

The tight coupling between category and causal learning

The main goal of the present research was to demonstrate the interaction between category and causal induction in causal model learning. We used a two-phase learning procedure in which learners were presented with learning input referring to two interconnected causal relations forming a causal chain (Experiment 1) or a common-cause model (Experiments 2a, b). One of the three events (i.e., the intermediate event of the chain, or the common cause) was presented as a set of uncategorized exemplars. Although participants were not provided with any feedback about category labels, they tended to induce categories in the first phase that maximized the predictability of their causes or effects. In the second causal learning phase, participants had the choice between transferring the newly learned categories from the first phase at the cost of suboptimal predictions, or they could induce a new set of optimally predictive categories for the second causal relation, but at the cost of proliferating different category schemes for the same set of events. It turned out that in all three experiments learners tended to transfer the categories entailed by the first causal relation to the second causal relation.     

On the origin of personal causal theories

Detecting the causal relations among environmental events is an important facet of learning. Certain variables have been identified which influence both human causal attribution and animal learning: temporal priority, temporal and spatial contiguity, covariation and contingency, and prior experience. Recent research has continued to find distinct commonalities between the influence these variables have in the two domains, supporting a neo-Humean analysis of the origins of personal causal theories. The cues to causality determine which event relationships will be judged as causal; personal causal theories emerge as a result of these judgments and in turn affect future attributions. An examination of animal learning research motivates further extensions of the analogy. Researchers are encouraged to study real-time causal attributions, to study additional methodological analogies to conditioning paradigms, and to develop rich learning accounts of the acquisition of causal theories.     

Sufficiency and Necessity Assumptions in Causal Structure Induction

Research on human causal induction has shown that people have general prior assumptions about causal strength and about how causes interact with the background. We propose that these prior assumptions about the parameters of causal systems do not only manifest themselves in estimations of causal strength or the selection of causes but also when deciding between alternative causal structures. In three experiments, we requested subjects to choose which of two observable variables was the cause and which the effect. We found strong evidence that learners have interindividually variable but intraindividually stable priors about causal parameters that express a preference for causal determinism (sufficiency or necessity; Experiment 1). These priors predict which structure subjects preferentially select. The priors can be manipulated experimentally (Experiment 2) and appear to be domain‐general (Experiment 3). Heuristic strategies of structure induction are suggested that can be viewed as simplified implementations of the priors.     

Inferences about unobserved causes in human contingency learning

Estimates of the causal efficacy of an event need to take into account the possible presence and influence of other unobserved causes that might have contributed to the occurrence of the effect. Current theoretical approaches deal differently with this problem. Associative theories assume that at least one unobserved cause is always present. In contrast, causal Bayes net theories (including Power PC theory) hypothesize that unobserved causes may be present or absent. These theories generally assume independence of different causes of the same event, which greatly simplifies modelling learning and inference. In two experiments participants were requested to learn about the causal relation between a single cause and an effect by observing their co-occurrence (Experiment 1) or by actively intervening in the cause (Experiment 2). Participants' assumptions about the presence of an unobserved cause were assessed either after each learning trial or at the end of the learning phase. The results show an interesting dissociation. Whereas there was a tendency to assume interdependence of the causes in the online judgements during learning, the final judgements tended to be more in the direction of an independence assumption. Possible explanations and implications of these findings are discussed.     

Accessing causal relations in semantic memory

Most studies investigating semantic memory have focused on taxonomic or associative relations. Little is known about how other relations, such as causal relations, are represented and accessed. In three experiments, we presented participants with pairs of words one after another, describing events that referred to either a cause (e.g., spark) or an effect (e.g., fire). We manipulated the temporal order of word presentation and the question participants had to respond to. The results revealed that questions referring to the existence of a causal relation are answered faster when the first word refers to a cause and the second word refers to its effect than vice versa. However, no such asymmetry was observed with questions referring to the associative relation. People appear to distinguish the roles of cause and effect when queried specifically about a causal relation, but not when the same information is evaluated for the presence of an associative relation.     

Serial causation: Occasion setting in a causal induction task

The temporal relations among candidate causes were studied in a causal induction task using a design that is known to produce occasion setting in animal learning preparations. For some subset of the observations, one event, the occasion setter, was accompanied by another event, the conditional cause; for another subset of the observations, the conditional cause occurred alone. The efficacy of the conditional cause depended on whether it was or was not accompanied by the occasion setter. Participants used the occasion setter to modulate their effect expectancy to the conditional cause when the events were presented serially, but not simultaneously. Current causal induction models are unable to account for the full range of effects that we observed; the relative roles of time, attention, and cue distinctiveness are discussed.     

Statistical word learning at scale: the baby's view is better

A key question in early word learning is how children cope with the uncertainty in natural naming events. One potential mechanism for uncertainty reduction is cross‐situational word learning – tracking word/object co‐occurrence statistics across naming events. But empirical and computational analyses of cross‐situational learning have made strong assumptions about the nature of naming event ambiguity, assumptions that have been challenged by recent analyses of natural naming events. This paper shows that learning from ambiguous natural naming events depends on perspective. Natural naming events from parent–child interactions were recorded from both a third‐person tripod‐mounted camera and from a head‐mounted camera that produced a ‘child's‐eye’ view. Following the human simulation paradigm, adults were asked to learn artificial language labels by integrating across the most ambiguous of these naming events. Significant learning was found only from the child's perspective, pointing to the importance of considering statistical learning from an embodied perspective.     

Individual differences in causal learning and decision making

In judgment and decision making tasks, people tend to neglect the overall frequency of base-rates when they estimate the probability of an event; this is known as the base-rate fallacy. In causal learning, despite people's accuracy at judging causal strength according to one or other normative model (i.e., Power PC, DeltaP), they tend to misperceive base-rate information (e.g., the cause density effect). The present study investigates the relationship between causal learning and decision making by asking whether people weight base-rate information in the same way when estimating causal strength and when making judgments or inferences about the likelihood of an event. The results suggest that people differ according to the weight they place on base-rate information, but the way individuals do this is consistent across causal and decision making tasks. We interpret the results as reflecting a tendency to differentially weight base-rate information which generalizes to a variety of tasks. Additionally, this study provides evidence that causal learning and decision making share some component processes.     

Time and Singular Causation—A Computational Model

Causal queries about singular cases, which inquire whether specific events were causally connected, are prevalent in daily life and important in professional disciplines such as the law, medicine, or engineering. Because causal links cannot be directly observed, singular causation judgments require an assessment of whether a co-occurrence of two events c and e was causal or simply coincidental. How can this decision be made? Building on previous work by Cheng and Novick (2005) and Stephan and Waldmann (2018), we propose a computational model that combines information about the causal strengths of the potential causes with information about their temporal relations to derive answers to singular causation queries. The relative causal strengths of the potential cause factors are relevant because weak causes are more likely to fail to generate effects than strong causes. But even a strong cause factor does not necessarily need to be causal in a singular case because it could have been preempted by an alternative cause. We here show how information about causal strength and about two different temporal parameters, the potential causes' onset times and their causal latencies, can be formalized and integrated into a computational account of singular causation. Four experiments are presented in which we tested the validity of the model. The results showed that people integrate the different types of information as predicted by the new model.     

Determining whether causal order affects cue selection in human contingency learning: Comments on Shanks and Lopez (1996)

Shanks and Lopez (1996) reported three experiments in which they attempted to test whether causal order affects cue selection, and concluded that it does not. Their study provides an opportunity to highlight some basic methodological criteria that must be met in order to test whether and how causal order influences learning. In particular, it is necessary to (1) ensure that participants consistently interpret the learning situation in terms of directed cause-effect relations; (2) measure the causal knowledge they acquire; (3) manipulate causal order; and (4) control the statistical relations between cause and effect. With respect to these criteria, each experiment reported by Shanks and Lopez fails on multiple counts. Moreover, several aspects of the results reported by Shanks and Lopez are explained by causal-model theory, but not by associative accounts. Their study thus adds to a growing body of evidence from different laboratories indicating that human contingency learning can be guided by causal interpretation.     

Expectations and interpretations during causal learning

In existing models of causal induction, 4 types of covariation information (i.e., presence/absence of an event followed by presence/absence of another event) always exert identical influences on causal strength judgments (e.g., joint presence of events always suggests a generative causal relationship). In contrast, we suggest that, due to expectations developed during causal learning, learners give varied interpretations to covariation information as it is encountered and that these interpretations influence the resulting causal beliefs. In Experiments 1A-1C, participants' interpretations of observations during a causal learning task were dynamic, expectation based, and, furthermore, strongly tied to subsequent causal judgments. Experiment 2 demonstrated that adding trials of joint absence or joint presence of events, whose roles have been traditionally interpreted as increasing causal strengths, could result in decreased overall causal judgments and that adding trials where one event occurs in the absence of another, whose roles have been traditionally interpreted as decreasing causal strengths, could result in increased overall causal judgments. We discuss implications for traditional models of causal learning and how a more top-down approach (e.g., Bayesian) would be more compatible with the current findings.     

How causal knowledge affects classification: A generative theory of categorization

Several theories have been proposed regarding how causal relations among features of objects affect how those objects are classified. The assumptions of these theories were tested in 3 experiments that manipulated the causal knowledge associated with novel categories. There were 3 results. The 1st was a multiple cause effect in which a feature's importance increases with its number of causes. The 2nd was a coherence effect in which good category members are those whose features jointly corroborate the category's causal knowledge. These 2 effects can be accounted for by assuming that good category members are those likely to be generated by a category's causal laws. The 3rd result was a primary cause effect, in which primary causes are more important to category membership. This effect can also be explained by a generative account with an additional assumption: that categories often are perceived to have hidden generative causes.     

Processing correlates of lexical semantic complexity

This paper explores how verb meanings that differ in semantic complexity are processed and represented. In particular, we compare eventive verbs, which denote causally structured events, with stative verbs, which denote facts without causal structure. We predicted that the conceptually more complex eventive verbs should take longer to process than stative verbs. Two experiments, a lexical decision task and a self-paced reading study, confirmed this prediction. The findings suggest that (a) semantic complexity is reflected in processing time, (b) processing verb meanings involves activating properties of the event structure beyond participants' roles, and (c) more generally, lexical event structures, which subsume thematic roles, may mediate between syntactic knowledge and semantic interpretation.     

On the representational systems underlying prospection: Evidence from the event-cueing paradigm

The ability to think about the future-prospection-is central to many aspects of human cognition and behavior, from planning and decision making, to self-control and the construction of a sense of identity. Yet, the exact nature of the representational systems underlying prospection is not fully understood. Recent findings point to the critical role of episodic memory in imagining specific future events, but it is unlikely that prospection depends solely on this system. Using an event-cueing paradigm in two studies, we here show that specific events that people imagine might happen in their personal future are commonly embedded in broader event sequences-termed event clusters-that link a set of envisioned events according to causal and thematic relations. These findings provide novel evidence that prospection relies on multiple representational systems, with general autobiographical knowledge structures providing a frame that organizes imagined events in overarching event sequences. The results further suggest that knowledge about personal goals plays an important role in structuring these event sequences, especially for the distant future.     

Just do it? Investigating the gap between prediction and action in toddlers’ causal inferences

Adults’ causal representations integrate information about predictive relations and the possibility of effective intervention; if one event reliably predicts another, adults can represent the possibility that acting to bring about the first event might generate the second. Here we show that although toddlers (mean age: 24 months) readily learn predictive relationships between physically connected events, they do not spontaneously initiate one event to try to generate the second (although older children, mean age: 47 months, do; Experiments 1 and 2). Toddlers succeed only when the events are initiated by a dispositional agent (Experiment 3), when the events involve direct contact between objects (Experiment 4), or when the events are described using causal language (Experiment 5). This suggests that causal language may help children extend their initial causal representations beyond agent-initiated and direct contact events.     

Acting knowingly: effects of the agent's awareness of an opportunity on causal attributions

According to difference-based (e.g. counterfactual/covariational) models of causal judgement, the epistemic state of the agent should not affect judgements of cause. Four experiments examined opportunity chains in which a physical event (distal cause) enabled a subsequent proximal cause to produce an outcome. All four experiments showed that when the proximal cause was a human action, it was judged as more causal if the agent was aware of his opportunity than if he was not or if the proximal cause was a physical event. The first two experiments showed that these preferences could not be explained in terms of differences in perceived conditional probability (whether from the observer's or the agent's point of view), social controllability or perceptions of the causal sequence as forming a single unit. The third experiment showed that awareness affected the perceived deliberateness with which the action brought the outcome about but not its perceived voluntariness. The fourth experiment showed that when the outcome was intended, the perceived deliberateness of the agent's action was a plausible mediator of the effect of awareness of opportunity on causal preference. We conclude that awareness of the opportunity allows inferences about the deliberate production of the outcome when the action is voluntary, which in turn influence causal judgements.