Answering Causal Queries about Singular Cases

因果推理是我们最核心的认知能力之一，在过去的几十年里吸引了大量的研究。这项研究的主要重点是对一般因果关系的建模判断。例如，吸烟和肺病之间一般因果关系的一个指标可能是发现在一个样本中，吸烟者往往比非吸烟者更有可能患上肺病。然而，还有第二类判断，关于单一因果关系的判断，迄今为止在心理学文献中基本上被忽视了。例如，当观察到一个特定的吸烟者患有肺部疾病时，可以提出这样的问题：这个人的吸烟是否是她肺部疾病的原因，或者这些事件的同时发生是否只是巧合。关于单一因果关系的争论在日常和专业背景中很普遍，比如法律或医学。到目前为止，奇异因果关系主要是由哲学家研究的，他们在大多数情况下都将注意力集中在确定性因果关系上。相比之下，我们的目标是研究概率因果关系的更普遍的情况下，这提出了一个问题，人们如何区分单一的因果关系从巧合和一般的因果关系。该项目被细分为三个部分：项目1侧重于一个单一的原因和一个单一的影响元素单一的因果判断。我们将测试一个新的贝叶斯计算模型，奇异因果关系的结构归纳模型（SISC），我们最近开发的模型响应奇异因果关系查询。该模型假设人们在评估奇异因果关系查询时利用他们的一般因果知识。此外，它形式化的假设，人们评估的可能性，原因已经产生了它的影响，因为这两个事件已被观察到存在。最后，该模型假设人们考虑到因果模型参数的大小和假设的因果结构的不确定性。项目1测试人们是否能以SISC预测的方式区分一般和单一因果关系判断。此外，SISC将与可能的竞争对手进行测试。在项目2中，我们研究了扩展的基本模型，占机制知识，这已被证明是重要的奇异因果关系的判断。最后，项目3侧重于可以灵活选择一般因果关系的参考类的情况。该项目解决了人们在判断一般和奇异因果关系时选择哪种参考类的问题。此外，我们将调查是否语用因素影响一般与奇异因果关系查询处理。

Causal reasoning is one of our most central cognitive competencies, which has attracted a substantial amount of research in the past decades. The main focus of this research has been on modelling judgments about general causal relations. For example, an indicator of a general causal relation between smoking and lung disease might be the finding that in a sample of people smokers tend to contract lung disease with a higher probability than non-smokers. However, there is a second class of judgments, judgments about singular causation, that have so far been largely neglected in the psychological literature. For example, when observing a specific smoker having contracted lung disease, the question can be raised whether this person's smoking is the cause of her lung disease, or whether the co-occurrence of the events is just a coincidence. Queries about singular causation are prevalent in everyday and professional contexts, such as the law or medicine. So far, singular causation has mainly been studied by philosophers who in most cases have restricted their focus on deterministic causal relations. By contrast, our goal is to study the more prevalent case of probabilistic causal relations, which raises the question how people distinguish singular causation from coincidence and from general causation. The project is subdivided into three parts: Project 1 focuses on elemental singular causal judgments with a single cause and a single effect. We will test a new Bayesian computational model, the structure induction model of singular causation (SISC), which we have recently developed to model responses to singular causation queries. The model assumes that people draw on their general causal knowledge when assessing singular causation queries. Moreover, it formalizes the assumption that people assess the likelihood that the cause has generated its effect given that both events have been observed to be present. Finally, the model assumes that people take into account uncertainty about the size of the causal model parameters and about the hypothesized causal structure. Project 1 tests whether people differentiate between general and singular causation judgments in a way predicted by SISC. Moreover, SISC will be tested against possible competitors. In Project 2 we study an extension of the basic model to account for mechanism knowledge, which has proven important in judgments about singular causation. Finally, Project 3 focuses on cases in which the reference class underlying the general causal relations can be flexibly chosen. The project addresses the question which reference class people choose when making judgments about general and singular causal relations. Additionally, we will investigate whether pragmatic factors influence how general versus singular causation queries are processed.