Causal Reasoning Research Articles

Speaking of Kinds: How Correcting Generic Statements can Shape Children's Concepts.

Generic language (e.g., "tigers have stripes") leads children to assume that the referenced category (e.g., tigers) is inductively informative and provides a causal explanation for the behavior of individual members. In two preregistered studies with 4- to 7-year-old children (N = 497), we considered the mechanisms underlying these effects by testing how correcting generics might affect the development of these beliefs about novel social and animal kinds (Study 1) and about gender (Study 2). Correcting generics by narrowing their scope to a single individual limited beliefs that the referenced categories could explain what their members would be like while broadening the scope to a superordinate category (Study 2) uniquely limited endorsement of gender norms. Across both studies, correcting generics did not alter beliefs about feature heritability and had mixed effects on inductive inferences, suggesting that additional mechanisms (e.g., causal reasoning about shared features) contribute to the development of full-blown essentialist beliefs. These results help illuminate the mechanisms by which generics lead children to view categories as having rich inductive and causal potential; in particular, they suggest that children interpret generics as signals that speakers in their community view the referenced categories as meaningful kinds that support generalization. The findings also point the way to concrete suggestions for how adults can effectively correct problematic generics (e.g., gender stereotypes) that children may hear in daily life.

SA16 Enhancing Clinical Trial Diversity and Representation Through Data-Adaptive Targeted Learning: A Tale of Lurking Berkson Bias

There is a ubiquitous need for enhanced diversity in clinical trials for generalizability. Diverse trial populations allow for more accurate predictions of the safety and efficacy of potential medicines with implications for equitable treatments. Further, the real-world data (RWD) collection process may result in biased estimates because of structural differences in access to healthcare, socioeconomic status, and cultural environments. Additionally, many traditional analytic methods can restrict cohorts and disproportionately exclude patient groups. Applying causal structural reasoning and data-adaptive targeted learning (TL) provides an efficient opportunity to bolster diversity and inclusion in trial analysis.

Rationality Now!

Rationality Now!

Hybrid Bayesian network discovery with latent variables by scoring multiple interventions

In Bayesian Networks (BNs), the direction of edges is crucial for causal reasoning and inference. However, Markov equivalence class considerations mean it is not always possible to establish edge orientations, which is why many BN structure learning algorithms cannot orientate all edges from purely observational data. Moreover, latent confounders can lead to false positive edges. Relatively few methods have been proposed to address these issues. In this work, we present the hybrid mFGS-BS (majority rule and Fast Greedy equivalence Search with Bayesian Scoring) algorithm for structure learning from discrete data that involves an observational data set and one or more interventional data sets. The algorithm assumes causal insufficiency in the presence of latent variables and produces a Partial Ancestral Graph (PAG). Structure learning relies on a hybrid approach and a novel Bayesian scoring paradigm that calculates the posterior probability of each directed edge being added to the learnt graph. Experimental results based on well-known networks of up to 109 variables and 10 k sample size show that mFGS-BS improves structure learning accuracy relative to the state-of-the-art and it is computationally efficient.

Students navigating shifting literacy expectations in grade 6 history teaching: a qualitative text-based study

This qualitative study focuses on Grade 6 students’ possibilities to engage in disciplinary writing practices in history teaching in a school located in a socially disadvantaged and linguistically diverse area. The aim is to contribute knowledge about how students negotiate different literacy expectations in the teaching. The researcher used ethnographic methods to follow one teacher and two groups of Grade 6 students for 12 weeks’ teaching about the Vasa era in Swedish history. The material used consists of samples of students writing, parts of the text material they studied, and written questions they answered in tests and pre-writing discussion. In the analysis, systemic-functional linguistics (SFL) and thematic content analysis were employed to develop categories of literacy expectations based on the written questions. SFL was also used to analyze samples of the students’ writing and relevant textbook material. The findings show that the expectations increased in a final test because most questions required causal reasoning. In contrast, the questions used in text discussions and homework tests conveyed a variety of literacy expectations, with an emphasis on chronological reconstructions of events. While some students adjusted their writing to causal reasoning, others relied on a reconstruct approach that adhered closely to the simple text material that was chronologically structured. Outlier examples showed students using multimodal representations, a substitute teacher’s oral elaboration on a text, and resources of learner language in their answers. Implications for teaching that support all students’ engagement in disciplinary practices of history are discussed.

Teleonomy as a problem of self-causation

AbstractA theoretical framework is provided to explore teleonomy as a problem of self-causation, distinct from upward, downward and reticulate causation. Causality theories in biology are often formulated within hierarchy theories, where causation is conceptualized as running up or down the rungs of a ladder-like hierarchy or, more recently, as moving between multiple hierarchies. Research on the genealogy of cosmologies demonstrates that in addition to hierarchy theories, causality theories also depend upon ideas of time. This paper explores the roots and impact of both time and hierarchy thinking on causal reasoning in the evolutionary sciences. Within evolutionary biology, the Neodarwinian synthesis adheres to a linear notion of time associated with linear hierarchies that portray upward causation. Eco-evo-devo schools recognize the importance of downward causation and consequently receive resistance from the standard view because downward causation is sometimes understood as backward causation, considered impossible by adherents of a linear time model. In contrast, downward causation works with a spatial or presential time notion. Hybridization, lateral gene transfer, infective heredity, symbiosis and symbiogenesis require recognition of reticulate causation occurring in both space and time, or spacetime, between distinct and interacting ontological hierarchies. Teleonomy is distinct from these types of causation because it invokes the problem of self-causation. By asking how the focal level in a hierarchy can persist through time, self-causation raises philosophical concerns on the nature of duration, identity and individuality.

D’ya Like DAGs? A Survey on Structure Learning and Causal Discovery

Causal reasoning is a crucial part of science and human intelligence. In order to discover causal relationships from data, we need structure discovery methods. We provide a review of background theory and a survey of methods for structure discovery. We primarily focus on modern, continuous optimization methods, and provide reference to further resources such as benchmark datasets and software packages. Finally, we discuss the assumptive leap required to take us from structure to causality.

Causal models versus reason models in Bayesian networks for legal evidence

In this paper we compare causal models with reason models in the construction of Bayesian networks for legal evidence. In causal models, arrows in the network are drawn from causes to effects. In a reason model, the arrows are instead drawn towards the evidence, from factum probandum to factum probans. We explore the differences between causal models and reason models and observe several distinct advantages with reason models. Reason models are better aligned with the philosophy of Bayesian inference, as they model reasons for up-dating beliefs. Reason models are better suited for measuring the combined support of the evidence, and a prior probability of guilt that reflects the number of possible perpetrators is accommodated more easily with reason models.

A Hybrid Model Integrating HFACS and BN for Analyzing Human Factors in CFIT Accidents

Controlled flight into terrain (CFIT) is considered a typical accident category of “low-probability-high consequence”. Human factors play an important role in CFIT accidents in such a complex and high-risk system. This study aims to explore the causal relationship and inherent correlation of CFIT accidents by the Human Factors Analysis and Classification System (HFACS) and Bayesian network (BN). A total of 74 global CFIT accident investigation reports from 2001 to 2020 were collected, and the main contributing factors were classified and analyzed based on the Human Factors Analysis and Classification System. Then, the model was transformed into a Bayesian network topology structure. To ensure accuracy, the prior probability of each root node was computed by the fuzzy number theory. Afterward, using the bidirectional reasoning ability of the Bayesian network under uncertainty, this study performed a systematic quantitative analysis of the controlled flight into terrain accidents, including causal reasoning analysis, diagnostic analysis, sensitivity analysis, most probable explanation, and scenario analysis. The results demonstrate that the precondition for unsafe acts (30.5%) has the greatest impact on the controlled flight into terrain accidents among the four levels of contributing factors. Inadequate supervision, intentional noncompliance with SOPs/cross-check, GPWS not installed or failure, adverse meteorological environment, and ground-based navigation aid malfunction or not being available are recognized as the top significant contributing factors. The contributing factors of the high sensitivity and most likely failure are identified, and the coupling effect between the different contributing factors is verified. This study can provide guidance for CFIT accident analysis and prevention.

Discovering Common Hidden Causes in Sequences of Events

Human cognition is marked by its ability to explain patterns in the world in terms of variables and regularities that are not directly observable, e.g., mental states, natural laws, and causal relationships. Previous research has demonstrated a capacity for inferring hidden causes from covariational evidence, as well as the use of temporal information to identify causal relationships among observed variables. Here we explore the human ability to use temporal information to make inferences about hidden causes, causal cycles, and other causal relationships, without relying on interventions. We examine two behavioral experiments and compare participants’ judgments to those of Bayesian computational-level models that use temporal order and delay information to infer the causal structure behind observed event sequences. Our results indicate that participants are able to use order and timing information to discover hidden causes, and make inferences about causal structures relating hidden and observable variables. Computational modeling indicates that most participants are best described by normative delay model predictions, but also reveals several clusters of participants who made unexpected inferences, suggesting opportunities to enrich future models of human causal reasoning.

KEPT: Knowledge Enhanced Prompt Tuning for event causality identification

Event causality identification (ECI) aims to identify causal relations of event mention pairs in text. Despite achieving certain accomplishments, existing methods are still not effective due to the following two issues: (1) the lack of causal reasoning ability, imposing restrictions on recognizing implicit causal relations; (2) the significant gap between fine-tuning and pre-training, which hinders the utilization of pre-trained language models (PLMs). In this paper, we propose a novel Knowledge Enhanced Prompt Tuning (KEPT) framework for ECI to address the issues mentioned above. Specifically, this method leverages prompt tuning to incorporate two kinds of knowledge obtained from external knowledge bases (KBs), including background information and relational information, for causal reasoning. To introduce external knowledge into our model, we first convert it to textual descriptions, then design an interactive attention mechanism and a selective attention mechanism to fuse background information and relational information, respectively. In addition, to further capture implicit relations between events, we adopt the objective from knowledge representation learning to jointly optimize the representations of causal relations and events. Experiment results on two widely-used benchmarks demonstrate that the proposed method outperforms the state-of-the-art models.

Invited Perspective: The Mysterious Case of Social Determinants of Health.

Invited Perspective: The Mysterious Case of Social Determinants of Health.

Diagnose Like a Radiologist: Hybrid Neuro-Probabilistic Reasoning for Attribute-Based Medical Image Diagnosis.

During clinical practice, radiologists often use attributes, e.g., morphological and appearance characteristics of a lesion, to aid disease diagnosis. Effectively modeling attributes as well as all relationships involving attributes could boost the generalization ability and verifiability of medical image diagnosis algorithms. In this paper, we introduce a hybrid neuro-probabilistic reasoning algorithm for verifiable attribute-based medical image diagnosis. There are two parallel branches in our hybrid algorithm, a Bayesian network branch performing probabilistic causal relationship reasoning and a graph convolutional network branch performing more generic relational modeling and reasoning using a feature representation. Tight coupling between these two branches is achieved via a cross-network attention mechanism and the fusion of their classification results. We have successfully applied our hybrid reasoning algorithm to two challenging medical image diagnosis tasks. On the LIDC-IDRI benchmark dataset for benign-malignant classification of pulmonary nodules in CT images, our method achieves a new state-of-the-art accuracy of 95.36% and an AUC of 96.54%. Our method also achieves a 3.24% accuracy improvement on an in-house chest X-ray image dataset for tuberculosis diagnosis. Our ablation study indicates that our hybrid algorithm achieves a much better generalization performance than a pure neural network architecture under very limited training data.

Counterfactual curiosity: motivated thinking about what might have been.

Counterfactual information, information about what might have been, forms the content of counterfactual thoughts and emotions like regret and relief. Recent research suggests that human adults and children, as well as rhesus monkeys, demonstrate 'counterfactual curiosity': they are motivated to seek out counterfactual information after making decisions. Based on contemporary theories of curiosity and information seeking and a broad range of empirical literature, we suggest multiple heterogeneous psychological processes that contribute to people's motivation for counterfactual information. This includes processes that are identified in the curiosity literature more generally-the potential use of counterfactual information for adaptive decision making (its long-term instrumental value) and the drive to reduce uncertainty. Additionally, we suggest that counterfactual information may be particularly alluring because of its role in causal reasoning; its relationship with prediction and decision making; and its potential to fulfil emotion regulation and self-serving goals. Some future directions have been suggested, including investigating the role of individual differences in counterfactual curiosity on learning and wellbeing. This article is part of the theme issue 'Thinking about possibilities: mechanisms, ontogeny, functions and phylogeny'.

Causal learning, counterfactual reasoning and pretend play: a cross-cultural comparison of Peruvian, mixed- and low-socioeconomic status U.S. children.

Pretend play universally emerges during early childhood and may support the development of causal inference and counterfactual reasoning. However, the amount of time spent pretending, the value that adults place on pretence and the scaffolding adults provide vary by both culture and socioeconomic status (SES). In middle class U.S. preschoolers, accuracy on a pretence-based causal reasoning task predicted performance on a similar causal counterfactual task. We explore the relationship between cultural environment, pretence and counterfactual reasoning in low-income Peruvian (N = 62) and low-income U.S. (N = 57) 3- to 4-year olds, and contrast findings against previous findings in an age-matched, mixed-SES U.S. sample (N = 60). Children learned a novel causal relationship, then answered comparable counterfactual and pretence-based questions about the relationship. Children's responses for counterfactual and pretence measures differed across populations, with Peruvian and lower-income U.S. children providing fewer causally consistent responses when compared with middle class U.S. children. Nevertheless, correlations between the two measures emerged in all populations. Across cohorts, children also provided more causally consistent answers during pretence than counterfactually. Our findings strengthen the hypothesis that causal pretend play is related to causal counterfactual reasoning across cultural contexts, while also suggesting a role for systematic environmental differences. This article is part of the theme issue 'Thinking about possibilities: mechanisms, ontogeny, functions and phylogeny'.

Causal Approach to Integrate Component Health Data into System Reliability Models

Two of the challenges of current plant reliability approaches are the ability to integrate plant health data, and to support decision making. Condition based data and diagnostic/prognostic information are in fact not considered into plant reliability models to inform system engineers on the most critical components. Currently, the propagation of quantitative health data from the component to the system level is a challenge given the diverse nature/structure of the data. On the other hand, plant reliability methods (which are typically based on fault-trees or reliability block diagrams) can effectively propagate data from the component to the system level, but values of failure rates or failure probabilities are an approximated integral representation of the past industry-wide operational experience, and it neglects the present component health status (e.g., diagnostic and condition-based data) and health projection (when available from prognostic data). Our first claim is that system reliability models should propagate health information from the component to the system/plant level in order to provide a quantitative snapshot of system/plant health and identify the most critical components. Our second claim is that component health should be informed solely by that specific component current and historical performance data and should not be an approximated integral representation of the past industry-wide operational experience. This paper is directly supporting these two claims by proposing a different approach to perform reliability modeling which relies on available component diagnostic, prognostic and condition-based data to measure component health, and it propagates this information through fault tree models. The propagation of health data from the component to the system level is performed not in terms of probability, but in terms of margins where margin is defined as the “distance” between the present actual status and an undesired event (e.g., failure or unacceptable performance). Through a cause-effect lens, while classical reliability models target the effect associated to a component performance, a margin-based approach focuses on the cause of an undesired component performance (i.e., component health). Hence, thinking of reliability in terms of margins implies decision making based on causal reasoning. We will show how fault tree models can be solved using a margin language and how this process can effectively assist system engineers to identify the most critical components.

Watch your language: An exploration of the use of causal wording in veterinary observational research.

Observational research may be conducted to predict an outcome or to identify associations between an intervention or risk factor (an "exposure") and an outcome. However, the end goal of observational research often is to identify exposures that can be manipulated to improve an outcome, meaning that the aim is identify causal relationships. Causal inference from observational studies may be appropriate when an exposure-outcome of interest is identified, causal reasoning is used to identify confounders, confounders are adequately controlled, and theoretical issues, such as temporality, are considered. If these conditions are not met, causal inference cannot be made in an observational study. The objective of our study was to explore the use of causal language in veterinary observational studies, and to compare the use of causal language between studies that appear to be predictive or associational in purpose vs. those that appear to be exploring causal relationships. The dataset comprised 200 observational studies in veterinary species published between 2020 and 2022. The majority (117 out of 200) were cross-sectional studies. There were 48 studies that described an exposure-outcome of interest, and we considered these studies to be exploring potential causal relationships; of note, this liberal categorization would be anticipated to overestimate the proportion of studies suitably designed for causal inference. Overall, 172 studies (86%) used causal wording in at least one section of the article. Causal language was used in 128/152 (84%) of studies exploring predictions or associations; this language implies causation when it is not appropriate to do so. In studies designed such that causal inference might be possible, 44/48 (92%) used causal language in one or more sections. There were no substantive differences in the use of causal wording between observational study designs, exposure types, or whether the first author's affiliation was a country in which English is an official language. There is a need for authors of veterinary observational studies to explicitly state the purpose of the study (associational, predictive, or causal), and to use causal wording appropriately based on the aim of the study.

The effects of task complexity on L2 English rapport-building language use and its relationship with paired speaking test task performance

Abstract Although task complexity effects on L2 oral production have been widely studied in teaching contexts, their application to task-based language assessment and pragmatic language use remains underexplored. Since pragmatic competence is part of effective communication and interactional abilities and is context-dependent, an investigation of the relationships among task complexity, pragmatic language use, and paired speaking test performance is needed. This study examined the effect of the resource-directing and resource-dispersing dimensions (the number of elements, causal reasoning demand, and planning time) of Robinson’s task complexity construct on fifty-two intermediate-level English as a Second Language (ESL) learners’ rapport-building language use during two decision-making tasks as the achievement test in an EAP program, and the relationships between rapport language use and three dimensions of paired speaking test performance: collaboration, task completion, and style. The results showed that frequency and variety of rapport-building language use did not significantly differ between the two tasks. However, the study found that only in the simple task did different types of rapport-building language have statistically significant positive or negative relationships with different dimensions of paired speaking test scores. Specifically, greeting language use had a strong or close to strong positive relationship with collaboration and style scores, whereas agreeing language had a strong negative relationship with collaboration scores. Additionally, thanking language had a strong negative association with task completion scores. The findings further suggest that task complexity effects learners’ production of rapport-building language in terms of alignment with their peer interlocutors and formality of style, and also impacts raters’ perceptions of paired speaking task performance. The findings highlight the importance of task effects, paired oral assessment rubrics development, and rapport-building language instruction.

Developing prospective science teachers’ multivariable thinking capabilities

Contemporary research on scientific thinking has overlooked the need to develop multivariable thinking among adults and the importance of doing so. This study examines whether problem-based inquiry activities with graded feedback, in conjunction with scaffolding, help prospective elementary science teachers in developing multivariable thinking. Activities involving multivariable causal models along with metacognitive scaffolding were introduced to half (n = 21) of a general science class of first-year undergraduates in a teacher-education program. The other half of the class served as a control group. On a pretest assessing control-of-variables and multivariable causal reasoning skills, both groups performed equally poorly. Employing the same teacher and teaching style in both groups, the impact of the intervention in one group was assessed. Results showed significant development of multivariable thinking on the part of the experimental group.

Leveraging on causal knowledge for enhancing the root cause analysis of equipment spot inspection failures

Causal correlation data over the equipment spot-inspection operation and maintenance (O&M) records and fault investigation sheets potentially reflect the state related to the causal effect of equipment failures. Various factors influence equipment failures, making it difficult to effectively analyze the main cause of the problems. Mining and leveraging these causal data from the equipment spot inspection records will undoubtedly significantly improve the root cause analysis of the fault in the O&M system. Hence, this paper introduces causal knowledge in equipment fault O&M for the first time and proposes to exploit causal knowledge for enhancing root cause analysis of equipment spot inspection failures. Specifically, an equipment fault O&M knowledge graph with causal knowledge called CausalKG is constructed to provide knowledge support for the causal analysis of faults. That is, CausalKG consists of spot-inspection knowledge graph (SIKG) and causal relationship knowledge (CRK) in equipment fault O&M. Further, a CausalKG-ALBERT knowledge reasoning model is designed. The model transforms CausalKG into network embeddings based on relational graph convolutional networks. In turn, it combines the Q&A mechanism of the language model ALBERT to mine the root cause knowledge of equipment failures. The case study confirms that incorporating the CRK is more effective than directly using the SIKG for causality reasoning; The model can fully use causal relationship knowledge to enhance the reliability of root cause analysis. This method is valuable to help engineers strengthen their causal analysis capabilities in preventive equipment maintenance.