Responsibility Under Uncertainty: Which Climate Decisions Matter Most?

We propose a new method for estimating how much decisions under monadic uncertainty matter. The method is generic and suitable for measuring responsibility in finite horizon sequential decision processes. It fulfills “fairness” requirements and three natural conditions for responsibility measures: agency, avoidance and causal relevance. We apply the method to study how much decisions matter in a stylized greenhouse gas emissions process in which a decision maker repeatedly faces two options: start a “green” transition to a decarbonized society or further delay such a transition. We account for the fact that climate decisions are rarely implemented with certainty and that their consequences on the climate and on the global economy are uncertain. We discover that a “moral” approach towards decision making – doing the right thing even though the probability of success becomes increasingly small – is rational over a wide range of uncertainties.

286Collider-stratification bias when estimating variable importance using Random Forests

Background Advances in causal inference have helped explain the longstanding birthweight and obesity paradoxes: selection bias due to conditioning on a collider variable i.e. collider-stratification bias (CSB). The lessons learned have critical implications for the interpretation of machine learning (ML), including decision trees and random forests (RFs), that implicitly condition on input variables. RFs are a popular approach for identifying important “predictors” from large data through variable importance, defined by the average decrease in prediction accuracy. While CSB has become a recognized concern when estimating exposure-outcome effects, knowledge of its impact on ML’s variable importance measures (VIMs) is limited. Applying the causal inference framework, we investigated the accuracy of RFs’ VIMs in data-mechanisms prone to CSB. Methods A Monte Carlo simulation study was conducted, with binary outcome and collider variables generated from logistic models. Two exposure variables stochastically determined the outcome and a collider variable, independent of the outcome. VIMs from RFs were compared to the known causal relevance of the input variables on the outcome. Results While variable importance of true exposure variables was not systematically affected by CSB, validity of VIMs can be affected, leading to erroneous selection of collider variables, causally independent of the outcome, as outcome predictors. Conclusions In presence of CSB, VIMs are not valid measures of the causal relevance of variables and may mislead selection of truly important factors that affect the outcome. Key messages ML must consider causal data-generating mechanisms otherwise it may lead to erroneous assessment of variable importance regarding outcome prediction.

Assessment of the causal relevance of ECG parameters for risk of atrial fibrillation: A mendelian randomisation study

Background Atrial electrical and structural remodelling in older individuals with cardiovascular risk factors has been associated with changes in surface electrocardiographic (ECG) parameters (e.g., prolongation of the PR interval) and higher risks of atrial fibrillation (AF). However, it has been difficult to establish whether altered ECG parameters are the cause or a consequence of the myocardial substrate leading to AF. This study aimed to examine the potential causal relevance of ECG parameters on risk of AF using mendelian randomisation (MR). Methods and findings Weighted genetic scores explaining lifelong differences in P-wave duration, PR interval, and QT interval were constructed, and associations between these ECG scores and risk of AF were estimated among 278,792 UK Biobank participants (mean age: 57 years at recruitment; 19,132 AF cases). The independent genetic variants contributing to each of the separate ECG scores, and their corresponding weights, were based on published genome-wide association studies. In UK Biobank, genetic scores representing a 5 ms longer P-wave duration or PR interval were significantly associated with lower risks of AF (odds ratio [OR] 0.91; 95% confidence interval [CI]: 0.87–0.96, P = 2 × 10−4 and OR 0.94; 95% CI: 0.93–0.96, P = 2 × 10−19, respectively), while longer QT interval was not significantly associated with AF. These effects were independently replicated among a further 17,931 AF cases from the AFGen Consortium. Investigation of potential mechanistic pathways showed that differences in ECG parameters associated with specific ion channel genes had effects on risk of AF consistent with the overall scores, while the overall scores were not associated with changes in left atrial size. Limitations of the study included the inherent assumptions of MR, restriction to individuals of European ancestry, and possible restriction of results to the normal ECG ranges represented in UK Biobank. Conclusions In UK Biobank, we observed evidence suggesting a causal relationship between lifelong differences in ECG parameters (particularly PR interval) that reflect longer atrial conduction times and a lower risk of AF. These findings, which appear to be independent of atrial size and concomitant cardiovascular comorbidity, support the relevance of varying mechanisms underpinning AF and indicate that more individualised treatment strategies warrant consideration.

Modeling configurational explanations

How can Qualitative Comparative Analysis contribute to causal knowledge? The article's answer builds on the shift from design to models that the Structural Causal Model framework has compelled in the probabilistic analysis of causation. From this viewpoint, models refine the claim that a ‘treatment’ has causal relevance as they specify the ‘covariates’ that make some units responsive. The article shows how QCA can establish configurational models of plausible ‘covariates’. It explicates the rationale, operations, and criteria that confer explanatory import to configurational models, then illustrates how the basic structures of the SCM can widen the interpretability of configurational solutions and deepen the dialogue among techniques.

Social Network Experiments

What is the most effective way of determining causal processes involving social networks? Many, if not most, network researchers would argue that networks have a causal impact on individual and collective outcomes, but identifying causal relationships in networks is notoriously challenging. This chapter provides an overview of experimental designs and gives several examples of how experimental methods can be employed in social network research. Experiments provide unsurpassed certainty in causal identification by allowing the researcher to control all aspects of the process while in many cases substantially reducing data collection costs. Moreover, with the aid of computers, the internet, and a little creativity, it is often possible to collect highly reliable data from online samples with the aid of crowdsourcing services, such as Amazon Mechanical Turk. With the aid of these tools, the opportunities have never been better for showing the causal relevance of network structures.

Entraining neurons via noninvasive electric stimulation improves cognition

Transcranial Alternating Current Stimulation (tACS) is a method that injects rhythmic currents into the human brain via electrodes attached to the scalp of a participant. This technique allows researchers to control naturally occurring brain rhythms and study their causal relevance for cognition. Recent findings, however, cast doubts on the effectiveness of tACS to stimulate the brain and its mode of action. Two new studies by Vieira and colleagues and Marchesotti and colleagues reported in the current issue report promising new results in showing that tACS can entrain single neuron activity and improve reading abilities in dyslexic individuals.

Genetic associations of adult height with risk of cardioembolic and other subtypes of ischaemic stroke: a Mendelian randomisation study in multiple ancestries

Background: Taller adult height is associated with lower risks of ischaemic heart disease in both observational and Mendelian randomisation studies, but little is known about the causal relevance of height for different subtypes of ischaemic stroke and the mechanisms involved. Methods: Height-associated genetic variants (up to 2,931) from previous genome-wide association studies were used to construct genetic instruments in different populations. Two-sample Mendelian randomisation approaches were used to examine the associations of genetically-determined height with ischaemic stroke and its subtypes in multiple ancestries (MEGASTROKE: 60,341 ischaemic stroke cases) supported by additional cases in Europeans (UK Biobank: 4,055 cases) and in Chinese (China Kadoorie Biobank: 10,297 cases). The associations of genetically-determined height with established cardiovascular and other risk factors were also examined in Europeans (UK Biobank: 336,750 participants) and Chinese (China Kadoorie Biobank: 58,277 participants). Results: Genetically-determined height was inversely associated with ischaemic stroke (4% [95% CI: 1-7] lower risk per 1 standard deviation taller height in MEGASTROKE). This masked much stronger opposing associations of height with different subtypes, with a 12% (95% CI: 6-17) higher risk of cardioembolic stroke, 11% (6-16) lower risk of large-artery stroke, and 14% (9-18) lower risk of small-vessel stroke. Genetically-determined height was strongly positively associated with atrial fibrillation, lean body mass and lung function, and inversely associated with levels of LDL cholesterol and blood pressure in both Europeans and Chinese. Conclusions: In multiple ancestries, genetic associations support the causal relevance of taller adult height for higher risk of cardioembolic stroke (in addition to atrial fibrillation) and lower risk of other ischaemic strokes, highlighting the need to properly differentiate subtypes of ischaemic stroke in both clinical practice and research.

Criticism of D.B Volkov’s Teleofunctional Solution to the Problem of Mental Causation

The articles examines the teleofunctional solution to the problem of mental causation, presented by Dmitry Volkov in his recently published book Free Will. An Illusion or an Opportunity. D.B. Volkov proposes solutions to three big metaphysical problems – mental causation, personal identity, and free will. Solving the first problem, Volkov creatively combines the advantages of Dennett’s teleofunctional model and Vasilyev’s local interactionism. Volkov’s teleofunctional model of mental causation seeks to prove the causal relevance of mental properties as non-local higher order properties. In my view, its substantiation is based on three points: (a) critics of the exclusion problem and Kim’s model of mental causation, (b) “Library of first editions” argument, (c) reduction of the causal trajectories argument (CTA 1) by Vasilyev to the counterpart argument (CTA 2) by Volkov. Each of these points faces objections. Kim’s criticism is based on an implicit confusion of two types of reduction – reduction from supervenience and from multiple realizability. The latter type does not threaten Kim’s ideas, but Volkov uses this very type in his criticism. The “Library of first editions” argument does not achieve its goal due to compositional features and because non-local relational properties are a type of external properties that cannot be causally relevant. The reduction of CTA 1 to CTA 2 is unsuccessful since, in the case of this reduction, important features of CTA 1 are lost – these are local mental properties, due to which the influence of non-local physical factors occurs. My main objection is that the concept of causally relevant non-local properties is incompatible with the very concept of cause. The set of causally relevant properties of cause can only be local.

Necessitarianism and Dispositions

In this paper, I argue in favor of necessitarianism, the view that dispositions, when stimulated, necessitate their manifestations. After introducing and clarifying what necessitarianism does and does not amount to, I provide reasons to support the view that dispositions once stimulated necessitate their manifestations according to the stimulating conditions and the relevant properties at stake. In this framework, I will propose a principle of causal relevance and some conditions for the possibility of interference that allow us to avoid the use of ceteris paribus clauses. I then defend necessitarianism from recent attacks raised by, among others, Mumford and Anjum, noting that the antecedent strengthening test is a test for causal relevance that raises no difficulties for necessitarianism.