Causal Reasoning Research Articles

The feasibility of climate action: Bridging the inside and the outside view through feasibility spaces

AbstractThe feasibility of different options to reduce the risks of climate change has engaged scholars for decades. Yet there is no agreement on how to define and assess feasibility. We define feasible as “do‐able under realistic assumptions.” A sound feasibility assessment is based on causal reasoning; enables comparison of feasibility across climate options, contexts, and implementation levels; and reflexively considers the agency of its audience. Global climate scenarios are a good starting point for assessing the feasibility of climate options since they represent causal pathways, quantify implementation levels, and consider policy choices. Yet, scenario developers face difficulties to represent all relevant causalities, assess the realism of assumptions, assign likelihood to potential outcomes, and evaluate the agency of their users, which calls for external feasibility assessments. Existing approaches to feasibility assessment mirror the “inside” and the “outside” view coined by Kahneman and co‐authors. The inside view considers climate change as a unique challenge and seeks to identify barriers that should be overcome by political choice, commitment, and skill. The outside view assesses feasibility through examining historical analogies (reference cases) to the given climate option. Recent studies seek to bridge the inside and the outside views through “feasibility spaces,” by identifying reference cases for a climate option, measuring their outcomes and relevant characteristics, and mapping them together with the expected outcomes and characteristics of the climate option. Feasibility spaces are a promising method to prioritize climate options, realistically assess the achievability of climate goals, and construct scenarios with empirically‐grounded assumptions.This article is categorized under: Climate, History, Society, Culture > Disciplinary Perspectives Assessing Impacts of Climate Change > Representing Uncertainty The Carbon Economy and Climate Mitigation > Decarbonizing Energy and/or Reducing Demand

Leveraging Causal Reasoning in Educational Data Mining: An Analysis of Brazilian Secondary Education

This study presents an approach to investigating the main interventions related to gains on performance using a combination of educational data mining (EDM) techniques and traditional theory-driven models. The goal is to overcome the limitation of previous EDM studies that lack of causal reasoning, which is a critical concern for educational specialists. We use large-scale assessment data from Brazil and map the main sources of unobserved confounders using causal graphs. We then use a two-way logistic regression fixed effects to account for these confounding factors. The model is evaluated for its predictive ability and further investigated through classification rules and decision trees, resulting in the proposition of new insights into the data. The findings of the study underline the importance of socio-economic factors and showcase the significant impact of faculty education policies as well as the vital role of Brazilian states in these policies.

Benchmarking causal reasoning algorithms for gene expression-based compound mechanism of action analysis

BackgroundElucidating compound mechanism of action (MoA) is beneficial to drug discovery, but in practice often represents a significant challenge. Causal Reasoning approaches aim to address this situation by inferring dysregulated signalling proteins using transcriptomics data and biological networks; however, a comprehensive benchmarking of such approaches has not yet been reported. Here we benchmarked four causal reasoning algorithms (SigNet, CausalR, CausalR ScanR and CARNIVAL) with four networks (the smaller Omnipath network vs. 3 larger MetaBase™ networks), using LINCS L1000 and CMap microarray data, and assessed to what extent each factor dictated the successful recovery of direct targets and compound-associated signalling pathways in a benchmark dataset comprising 269 compounds. We additionally examined impact on performance in terms of the functions and roles of protein targets and their connectivity bias in the prior knowledge networks.ResultsAccording to statistical analysis (negative binomial model), the combination of algorithm and network most significantly dictated the performance of causal reasoning algorithms, with the SigNet recovering the greatest number of direct targets. With respect to the recovery of signalling pathways, CARNIVAL with the Omnipath network was able to recover the most informative pathways containing compound targets, based on the Reactome pathway hierarchy. Additionally, CARNIVAL, SigNet and CausalR ScanR all outperformed baseline gene expression pathway enrichment results. We found no significant difference in performance between L1000 data or microarray data, even when limited to just 978 ‘landmark’ genes. Notably, all causal reasoning algorithms also outperformed pathway recovery based on input DEGs, despite these often being used for pathway enrichment. Causal reasoning methods performance was somewhat correlated with connectivity and biological role of the targets.ConclusionsOverall, we conclude that causal reasoning performs well at recovering signalling proteins related to compound MoA upstream from gene expression changes by leveraging prior knowledge networks, and that the choice of network and algorithm has a profound impact on the performance of causal reasoning algorithms. Based on the analyses presented here this is true for both microarray-based gene expression data as well as those based on the L1000 platform.

Identifying Chemistry Students' Baseline Systems Thinking Skills When Constructing System Maps for a Topic on Climate Change.

New resources have recently been emerging for educators to implement systems thinking (ST) in chemistry education, including a proposed set of ST skills. While these efforts aim to make ST implementation easier, little is known about how to assess these skills in a chemistry context. In this study, we investigated ST skills employed by students who constructed system maps of a topic related to climate change. Eighteen undergraduate chemistry students from first- to third-year participated in this study. We designed and implemented a ST intervention to capture how students engaged with three ST tasks, performed individually and collaboratively. In our analysis, we focused on 11 ST skills that aligned with five characteristics proposed in a recent study. We found that participants demonstrated most of these ST skills when engaging with the ST tasks, with nuances. Participants' system maps: (1) lacked concepts and connections at the submicroscopic level, (2) included multiple types of connections but few circular loops and causal connections, (3) lacked causal reasoning, although participants did predict how their system maps changed over time, (4) demonstrated the breadth of connections but did not describe human connections to the underlying chemistry of climate change topics. These findings identify aspects of ST where chemistry educators need to place emphasis when teaching ST skills to chemistry students and when guiding learning activities and other assessments. Using our findings, we created an adaptable ST rubric for the chemistry community as a tool for assessing ST skills.

THE THINKING OF STUDENTS AGED 15-18 IN EXPLAINING THE DISSOLUTION PHENOMENON

Explaining natural phenomena by determining causal relationships is conducive to understanding scientific concepts. In science education, numerous studies examine students' causal reasoning. Given the importance of core ideas for students’ understanding of how and why a phenomenon occurs, the study focused on the relationship between students' understanding of atoms/molecules interactions and the nature of reasoning. This study drew on a framework that identifies essential components of students' reasoning, which was used to analyze the dissolution phenomenon in the example of salt in water. Students in grades 9-12 (N=147) explained the dissolution of salt. The results showed that there were five types of reasoning: simple descriptive, fuzzy causal, linear causal, interactive causal, and mechanistic. More students in higher than lower grades exhibited non-causal reasoning. Based on the students' drawings of atoms/molecules interactions, the study summarized performance in the association category. Students’ performance in drawing indicated that their understanding of particle interactions was limited. The results showed that there was a large correlation between understanding of the core ideas and reasoning types. Keywords: core ideas understanding, causal reasoning, dissolution phenomenon, small-sample qualitative study

Reasoning about causality in games

Causal reasoning and game-theoretic reasoning are fundamental topics in artificial intelligence, among many other disciplines: this paper is concerned with their intersection. Despite their importance, a formal framework that supports both these forms of reasoning has, until now, been lacking. We offer a solution in the form of (structural) causal games, which can be seen as extending Pearl's causal hierarchy to the game-theoretic domain, or as extending Koller and Milch's multi-agent influence diagrams to the causal domain. We then consider three key questions:i)How can the (causal) dependencies in games – either between variables, or between strategies – be modelled in a uniform, principled manner?ii)How may causal queries be computed in causal games, and what assumptions does this require?iii)How do causal games compare to existing formalisms? To address question i), we introduce mechanised games, which encode dependencies between agents' decision rules and the distributions governing the game. In response to question ii), we present definitions of predictions, interventions, and counterfactuals, and discuss the assumptions required for each. Regarding question iii), we describe correspondences between causal games and other formalisms, and explain how causal games can be used to answer queries that other causal or game-theoretic models do not support. Finally, we highlight possible applications of causal games, aided by an extensive open-source Python library.

ICT Skills in the Deployment of 21st Century Skills: A (Cognitive) Developmental Perspective through Early Childhood

ICT technologies are an integral part of today’s digitized society. Therefore, it is important that children acquire ICT skills as part of 21st century skills education to prepare them for later life. Drawing on the literature, seven 21st century skills can profit from the addition of ICT skills, i.e., technical, information, communication, collaboration, critical thinking, creative, and problem-solving skills. While many efforts have been made to integrate ICT skills as part of 21st century skills education into primary and secondary school curricula, the implementation of these skills in early childhood education and care remains a challenge due to developmental concerns. This paper aims to uncover developmental antecedents for ICT 21st century skills in early childhood, mainly addressing children’s cognitive development, and propose ways to implement these skills in child-friendly ways. Drawing on the literature on developmental psychology, seven cognitive developmental antecedents were identified: inductive, deductive, abductive, causal, and scientific reasoning, executive functions, and computational thinking. Moreover, five additional developmental antecedents were identified: fine motor skills, language development, self-regulation, social-emotional development, and creativity. On the backdrop of these antecedents, ways of implementing ICT skills as part of 21st century skills education in early childhood classrooms are proposed that include digital games and learning apps, collaborative play or problem-solving activities with toy robots.

Association between Subjective Hearing Impairment and Cognitive Status among Korean Elderly.

The goal of this study was to investigate the impact of subjectively assessed hearing impairment on cognitive function in elderly Koreans living in the community. In the 2020 Survey of Living Conditions and Welfare Needs of Korean Older Persons Survey, 9920 subjects (5949 females; 60%) aged 65 or more years were examined. Using the Korean version of the Mini-Mental Status Examination (MMSE-KC), cognitive function was evaluated. To investigate the relationship between hearing impairment and cognitive status, multiple logistic regression analysis was performed with adjustment for multiple confounding variables (socioeconomic, health behavior, psychological factors, and functional status). There were 2297 participants in the hearing impairment group (23.2%) and 7623 subjects in the no-hearing impairment group. Cognitive impairment was significantly higher in the hearing impairment group (37.2%) compared to the no-hearing impairment group (27.5%). After adjusting for confounders, hearing impairment was significantly associated with an increased risk of cognitive decline (odds ratio [OR] 1.21; 95% confidence interval [CI] 1.08-1.35) compared to no-hearing impairment group. Although a cross-cectional design of this study does not allow a causal reasoning, our findings show a significant association between the hearing loss of older adults and their cognitive impairment. Hearing impairment should be regarded as a risk factor for cognitive disorders.

Modeling COVID-19 disease processes by remote elicitation of causal Bayesian networks from medical experts

BackgroundCOVID-19 is a new multi-organ disease causing considerable worldwide morbidity and mortality. While many recognized pathophysiological mechanisms are involved, their exact causal relationships remain opaque. Better understanding is needed for predicting their progression, targeting therapeutic approaches, and improving patient outcomes. While many mathematical causal models describe COVID-19 epidemiology, none have described its pathophysiology.MethodsIn early 2020, we began developing such causal models. The SARS-CoV-2 virus’s rapid and extensive spread made this particularly difficult: no large patient datasets were publicly available; the medical literature was flooded with sometimes conflicting pre-review reports; and clinicians in many countries had little time for academic consultations. We used Bayesian network (BN) models, which provide powerful calculation tools and directed acyclic graphs (DAGs) as comprehensible causal maps. Hence, they can incorporate both expert opinion and numerical data, and produce explainable, updatable results. To obtain the DAGs, we used extensive expert elicitation (exploiting Australia’s exceptionally low COVID-19 burden) in structured online sessions. Groups of clinical and other specialists were enlisted to filter, interpret and discuss the literature and develop a current consensus. We encouraged inclusion of theoretically salient latent (unobservable) variables, likely mechanisms by extrapolation from other diseases, and documented supporting literature while noting controversies. Our method was iterative and incremental: systematically refining and validating the group output using one-on-one follow-up meetings with original and new experts. 35 experts contributed 126 hours face-to-face, and could review our products.ResultsWe present two key models, for the initial infection of the respiratory tract and the possible progression to complications, as causal DAGs and BNs with corresponding verbal descriptions, dictionaries and sources. These are the first published causal models of COVID-19 pathophysiology.ConclusionsOur method demonstrates an improved procedure for developing BNs via expert elicitation, which other teams can implement to model emergent complex phenomena. Our results have three anticipated applications: (i) freely disseminating updatable expert knowledge; (ii) guiding design and analysis of observational and clinical studies; (iii) developing and validating automated tools for causal reasoning and decision support. We are developing such tools for the initial diagnosis, resource management, and prognosis of COVID-19, parameterized using the ISARIC and LEOSS databases.

The Broad Reach of Multivariable Thinking

Simple explanations are very often inadequate and can encourage faulty inferences. We examined college students’ explanations regarding illegal immigration to determine the prevalence of single-factor explanations. The form of students’ explanations was predicted by their responses on a simple three-item forced-choice multivariable causal reasoning task in which they selected the strongest evidence against a causal claim. In a further qualitative investigation of explanations by a sample of community adults, we identified positive features among those who scored high on this multivariable causal reasoning task. We consider limitations of single-factor reasoning and means of encouraging more comprehensive explanations to support claims.

Mechanism of action deconvolution of the small-molecule pathological tau aggregation inhibitor Anle138b

BackgroundA key histopathological hallmark of Alzheimer’s disease (AD) is the presence of neurofibrillary tangles of aggregated microtubule-associated protein tau in neurons. Anle138b is a small molecule which has previously shown efficacy in mice in reducing tau aggregates and rescuing AD disease phenotypes.MethodsIn this work, we employed bioinformatics analysis—including pathway enrichment and causal reasoning—of an in vitro tauopathy model. The model consisted of cultured rat cortical neurons either unseeded or seeded with tau aggregates derived from human AD patients, both of which were treated with Anle138b to generate hypotheses for its mode of action. In parallel, we used a collection of human target prediction models to predict direct targets of Anle138b based on its chemical structure.ResultsCombining the different approaches, we found evidence supporting the hypothesis that the action of Anle138b involves several processes which are key to AD progression, including cholesterol homeostasis and neuroinflammation. On the pathway level, we found significantly enriched pathways related to these two processes including those entitled “Superpathway of cholesterol biosynthesis” and “Granulocyte adhesion and diapedesis”. With causal reasoning, we inferred differential activity of SREBF1/2 (involved in cholesterol regulation) and mediators of the inflammatory response such as NFKB1 and RELA. Notably, our findings were also observed in Anle138b-treated unseeded neurons, meaning that the inferred processes are independent of tau pathology and thus represent the direct action of the compound in the cellular system. Through structure-based ligand-target prediction, we predicted the intracellular cholesterol carrier NPC1 as well as NF-κB subunits as potential targets of Anle138b, with structurally similar compounds in the model training set known to target the same proteins.ConclusionsThis study has generated feasible hypotheses for the potential mechanism of action of Anle138b, which will enable the development of future molecular interventions aiming to reduce tau pathology in AD patients.

Causal reasoning for algorithmic fairness in voice controlled cyber-physical systems

Automated speaker recognition is enabling personalized interactions with the voice-based interfaces and assistants part of the modern cyber-physical-social systems. Prior studies have unfortunately uncovered disparate impacts across demographic groups on the outcomes of speaker recognition systems and consequently proposed a range of countermeasures. Understanding why a speaker recognition system may lead to this disparate performance for different (groups of) individuals, going beyond mere data imbalance reasons and black-box countermeasures, is an essential yet under-explored perspective. In this paper, we propose an explanatory framework that aims to provide a better understanding of how speaker recognition models perform as the underlying voice characteristics on which they are tested change. With our framework, we evaluate two state-of-the-art speaker recognition models, comparing their fairness in terms of security, through a systematic analysis of the impact of more than twenty voice characteristics. Our findings include important takeaways to enable voice controlled cyber-physical-social systems for everyone. Source code and data are available at https://bit.ly/EA-PRLETTERS.

Causal reasoning about epidemiological associations in conversational AI

We present a Socratic dialogue with ChatGPT, a large language model (LLM), on the causal interpretation of epidemiological associations between fine particulate matter (PM2.5) and human mortality risks. ChatGPT, reflecting probable patterns of human reasoning and argumentation in the sources on which it has been trained, initially holds that “It is well-established that exposure to ambient levels of PM2.5 does increase mortality risk” and adds the unsolicited remark that “Reducing exposure to PM2.5 is an important public health priority.” After patient questioning, however, it concludes that “It is not known with certainty that current ambient levels of PM2.5 increase mortality risk. While there is strong evidence of an association between PM2.5 and mortality risk, the causal nature of this association remains uncertain due to the possibility of omitted confounders.” This revised evaluation of the evidence suggests the potential value of sustained questioning in refining and improving both the types of human reasoning and argumentation imitated by current LLMs and the reliability of the initial conclusions expressed by current LLMs.

Using cognitive psychology to understand GPT-3

We study GPT-3, a recent large language model, using tools from cognitive psychology. More specifically, we assess GPT-3's decision-making, information search, deliberation, and causal reasoning abilities on a battery of canonical experiments from the literature. We find that much of GPT-3's behavior is impressive: It solves vignette-based tasks similarly or better than human subjects, is able to make decent decisions from descriptions, outperforms humans in a multiarmed bandit task, and shows signatures of model-based reinforcement learning. Yet, we also find that small perturbations to vignette-based tasks can lead GPT-3 vastly astray, that it shows no signatures of directed exploration, and that it fails miserably in a causal reasoning task. Taken together, these results enrich our understanding of current large language models and pave the way for future investigations using tools from cognitive psychology to study increasingly capable and opaque artificial agents.

Editor’s Introduction to JPE Micro

Editor’s Introduction to <i>JPE Micro</i>

The contribution to policies of an exposome-based approach to childhood obesity.

Childhood obesity is an increasingly severe public health problem, with a prospective impact on health. We propose an exposome approach to identify actionable risk factors for this condition. Our assumption is that relationships between external exposures and outcomes such as rapid growth, overweight, or obesity in children can be better understood through a "meet-in-the-middle" model. This is based on a combination of external and internal exposome-based approaches, that is, the study of multiple exposures (in our case, dietary patterns) and molecular pathways (metabolomics and epigenetics). This may strengthen causal reasoning by identifying intermediate markers that are associated with both exposures and outcomes. Our biomarker-based studies in the STOP consortium suggest (in several ways, including mediation analysis) that branched-chain amino acids (BCAAs) could be mediators of the effect of dietary risk factors on childhood overweight/obesity. This is consistent with intervention and animal studies showing that higher intake of BCAAs has a positive impact on body composition, glycemia, and satiety. Concerning food, of particular concern is the trend of increasing intake of ultra-processed food (UPF), including among children. Several mechanisms have been proposed to explain the impact of UPF on obesity and overweight, including nutrient intake (particularly proteins), changes in appetite, or the role of additives. Research from the Avon Longitudinal Study of Parents and Children cohort has shown a relationship between UPF intake and trajectories in childhood adiposity, while UPF was related to lower blood levels of BCAAs. We suggest that an exposome-based approach can help strengthening causal reasoning and support policies. Intake of UPF in children should be restricted to prevent obesity.

Understanding pretense as causal inference

There is a long-standing interest in the role that children’s understanding of pretense plays in their more general theory of mind development. Some argue that children understand pretense as a mental state, and the capacity to pretend is indicative of children possessing the capacity for mental representations. Others argue that children understand pretense in terms of actions and appearances, and an understanding of the mental states involved in pretending has a prolonged developmental trajectory. The goal of this paper is to integrate these ideas by positing that children understand pretense as a form of causal inference, which is based on both their general causal reasoning capacities and specific knowledge of mental states. I will first review literature on children’s understanding of pretense, and how such understanding can be conceptualized as integrating with children’s causal reasoning ability. I will then consider how children’s causal knowledge affects the ways they make inferences about others’ pretense. Next, I will consider the role of causal knowledge more broadly in children’s reasoning about pretense worlds, judgments of possibility, and counterfactual reasoning. Taken together the goal of this review is to synthesize how children understand pretending into a rational constructivist framework for understanding social cognitive development in a more integrative manner.

Water table prediction through causal reasoning modelling

This research is mainly aimed to analyze and model the relationship of the binomial Rainfall-Piezometry. In this sense, the inherent causality contained in temporal hourly Rainfall and Groundwater levels (piezometry) data records has been taken. This has been done through Bayesian Causal Reasoning (BCR) which is technique belonging to Artificial Intelligence (AI) based on Bayesian Theorem. The methodology comprises two main stages, first an analytical method from classic regression analysis, and second, a Bayesian Causal Modelling Translation (BCMT) that itself comprises several iterative steps. This research ultimately becomes a tool for aquifers management that comprises a bivariate function made of two variables Rainfall and Piezometry (Temporal Groundwater level evolution). This innovative methodology has been successfully applied in the Quaternary aquifer of the Campo de Cartagena groundwater body, which is an aquifer system that directly is connected to Mar Menor coastal lagoon (Murcia region, SE Spain).

Bringing cumulative technological culture beyond copying versus reasoning.

The dominant view of cumulative technological culture suggests that high-fidelity transmission rests upon a high-fidelity copying ability, which allows individuals to reproduce the tool-use actions performed by others without needing to understand them (i.e., without causal reasoning). The opposition between copying versus reasoning is well accepted but with little supporting evidence. In this article, we investigate this distinction by examining the cognitive science literature on tool use. Evidence indicates that the ability to reproduce others' tool-use actions requires causal understanding, which questions the copying versus reasoning distinction and the cognitive reality of the so-called copying ability. We conclude that new insights might be gained by considering causal understanding as a key driver of cumulative technological culture.

Police Recognition of Gender Issues in Relation to Intimate Partner Domestic Violence and Abuse in Greece

AbstractThe aim of the research was to understand the perceptions and beliefs of police officers who work and support victims of intimate partner domestic violence and abuse (DVA) and evaluate the way it is recorded. A sample was collected from eight police departments in Greece. The head of each police department was approached by the researcher and asked to distribute a structured questionnaire to front-line police officers who work with DVA victims. In total 358 police officers completed the questionnaire. Overall, nearly twice as many police officers referred to female victims (n = 170) compared to male victims (n = 83) of DVA. The police officers indicated that female victims usually suffered from severe physical violence/GBH (43%) and psychological abuse (27%), whereas male victims were most likely to disclose minor physical injuries and verbal abuse (49%) and less psychological violence (10%). An arrest was the most common response for both female and male victims (51% and 60%, respectively). Police participants indicated that offenders against males were more likely to be charged (28%) but less likely to be convicted (4%), whereas the opposite was reported for offenders of female victims (14% and 13%, respectively). If the offender was known to the police, the opinion was they were less likely to be prosecuted. Police respondents reported no significant differences for female and male victims as to causal factors and reasons for NOT reporting DVA. These were discrimination (31%), few DVA services (23%), intimidation (15%), shame and stigma of media reports (14%), lack of community support (6%), and personal reasons (9%).