The authors of the article, “Causal Inference in Generalizable Environments: Systematic Representative Design”, boldly announce their core point in the opening line of the abstract stating that, “Causal inference and generalizability both matter.” While a surface glance might suggest this to be a simple notion, a closer examination reveals the complexity of what they are proposing. This complexity is apparent when one considers that the bulk of human experimental research has always been challenged in its inability to concurrently deliver on both of these aims. This is no slight on the tens of 1000’s of human researchers and behavioral scientists who have devoted long careers to highly controlled human psychological and social science laboratory research. Rather, it reflects the sheer enormity of the challenges for conducting human studies designed to specify human function with physics-informed lab methods, while at the same time producing results that lead to enhanced understanding and prediction of how people will operate in the complex and ever-changing contexts that make up everyday life. At the core of this issue is a methodological and philosophical challenge that is relevant to all areas of human subjects’ research, beyond the social science focus of the Miller et al. (this issue) article. It is our aim to discuss the central topics in their article through the lens of our own work using Virtual/Augmented Reality and Virtual Human simulation technologies for clinical and training applications. The Miller et al. (this issue) piece is dense with diverse theoretical viewpoints in support of their analysis of the causal inference vs. generalization dilemma and proposes a remedy that integrates these aims by leveraging mixed reality technologies. The authors engage the reader in a tour de force of classical (and current) theory and research aimed at dissecting this timeless challenge in the field of experimental psychology in the social sciences. This challenge can be stated simply in terms of goals that are synergistic in the ideal, but competing in the pragmatic: How does one conduct solid scientific human research under controlled laboratory conditions, which at the same time creates generalizable knowledge about thinking, feeling, behaving, and interacting in everyday life? To address this challenge, the authors propose the application of simulation technology as a new approach that optimizes representativeness to the organism-in-situation, to which they aim to generalize. This is not a newly recognized challenge. Over the last century, psychology has strived to establish its place in the “hard” sciences via the rigorous application of the scientific method aiming to measure, understand, and modify (or treat) human cognition, emotion, behavior, and social interaction under highly controlled experimental conditions. However, such laboratory-based research conditions are often lacking in context and bears little resemblance to the dynamic stimulus complexity of the everyday world. Not only does reality richly vary from moment-to-moment and place-to-place, but since humans are not clones, they also vary dramatically in genetically endowed capabilities, limitations, and predispositions. Thus, while the knowledge generated under very constrained laboratory-based conditions, where the rich diversity of human response and function is reduced to mean values, can describe phenomenon with strong internal consistency (to the highly controlled and reality-constrained/limited experimental setting), it often produces limited generalizability for understanding and/or predicting human function in everyday life. The Behaviorist movement strived to apply rigor to this experimental challenge by maintaining a militant focus on observable behavior as the only variable of interest in any attempt to study humans in a scientific fashion. And in some sense that view has endured, whether explicitly stated or implicit in the value that researchers place on behavioral measures in counterpoint to what some view as the fuzzy and variably-biased nature of introspective self-report data. However, the relevance of verbal declarations of perspective, intent, cognitive appraisal/analysis, and emotional expression in everyday life cannot be denied or discounted, and thus the various waves of cognitive-behavior approaches have succeeded in “sneaking the mind back into psychology”, to the dismay of old school behaviorists. However, the good news as Miller et al. (this issue) clearly state, is that recent advances in modern simulation technologies (Virtual Reality (VR), Augmented Reality (AR), Virtual Human (VH) intelligent agents, etc.), have now provided new opportunities for creating research tools that aim to support a more predictive and ecologically relevant analysis of human function in everyday life. VR simulations can now present research participants with highly controlled, systematically deliverable stimulus presentations/challenges while they are immersed within the context of a functionally

中文翻译：

---

系统代表性设计与临床虚拟现实

文章“可推广环境中的因果推理：系统代表性设计”的作者在摘要的开头部分大胆地宣布了他们的核心观点，指出“因果推理和可概括性都很重要”。虽然表面上的一瞥可能表明这是一个简单的概念，但仔细检查会发现他们所提出的内容的复杂性。当人们考虑到大部分人类实验研究一直面临着无法同时实现这两个目标的挑战时，这种复杂性就显而易见了。对于数以千计的人类研究人员和行为科学家来说，这可不是什么小事，他们长期致力于高度控制的人类心理和社会科学实验室研究。相当，它反映了进行人类研究的巨大挑战，这些研究旨在通过物理知识的实验室方法来指定人体功能，同时产生的结果可以增强对人们在复杂和不断变化的环境中将如何运作的理解和预测构成日常生活的语境。这个问题的核心是一个方法论和哲学挑战，它与人类学科研究的所有领域相关，超出了 Miller 等人的社会科学重点。（本期）文章。我们的目标是通过我们自己的工作镜头来讨论他们文章中的中心主题，将虚拟/增强现实和虚拟人模拟技术用于临床和培训应用。米勒等人。（本期）文章中包含各种理论观点，以支持他们对因果推理与泛化困境的分析，并提出了一种通过利用混合现实技术整合这些目标的补救措施。作者让读者参与了经典（和当前）理论和研究的精髓，旨在剖析社会科学实验心理学领域的这一永恒挑战。这一挑战可以简单地用在理想中协同但在务实中竞争的目标来表述：一个人如何在受控的实验室条件下进行扎实的科学研究，同时创造关于思维、感觉、行为的普遍知识，并在日常生活中互动？为了应对这一挑战，作者提出应用模拟技术作为一种新方法，可以优化对原地生物的代表性，他们旨在推广这一点。这不是一个新近认识到的挑战。在过去的一个世纪里，心理学通过严格应用旨在测量、理解和修改（或治疗）高度环境下人类认知、情感、行为和社会互动的科学方法，努力在“硬”科学中占据一席之地。控制实验条件。然而，这种基于实验室的研究条件往往缺乏背景，与日常世界的动态刺激复杂性几乎没有相似之处。不仅现实在每时每刻和地点之间变化很大，而且由于人类不是克隆人，他们在遗传能力、局限性和倾向方面也有很大差异。因此，虽然在非常受限的实验室条件下产生的知识，其中人类反应和功能的丰富多样性被减少到平均值，但可以描述具有很强内部一致性的现象（对于高度控制和现实约束/有限的实验设置） ，它通常对理解和/或预测日常生活中的人类功能产生有限的普遍性。行为主义运动通过将可观察的行为作为唯一感兴趣的变量来以科学方式研究人类，从而努力将严密性应用于这一实验挑战。从某种意义上说，这种观点一直存在，无论是明确说明还是隐含在研究人员对行为测量的价值中，与某些人认为的内省自我报告数据的模糊和可变偏见性质相对立。然而，日常生活中的观点、意图、认知评估/分析和情感表达的口头声明的相关性不可否认或打折扣，因此各种认知行为方法的浪潮已经成功地“将思想潜入心理学” ，让老派行为主义者感到沮丧。然而，米勒等人的好消息。（本期）明确指出，现代模拟技术（虚拟现实（VR）、增强现实（AR）、虚拟人（VH）智能代理等）的最新进展，现在为创建旨在支持对日常生活中的人类功能进行更具预测性和生态相关性的分析的研究工具提供了新的机会。VR 模拟现在可以向研究参与者呈现高度可控的、可系统化的刺激演示/挑战，同时他们沉浸在功能性环境中