Collaborative Proposal: Model Knowledge and Scientific Judgment

科学研究中的一个中心问题是证据和假设之间的关系。但科学研究并不是研究这种关系的唯一学科。科学研究太长时间忽视了半个世纪以来引人入胜的重要实证文献，我们称之为“改善心理学”。改善心理学的一个中心发现是，当涉及到对实际重要事项的预测时，统计预测规则（或SPRs）通常比人类专家更可靠（而且通常明显更可靠）（Meehl 1954, Sawyer 1966, Dawes 1994）。本文献提供了有效的推理策略，已经过反复和成功的经验检验。这一建议是发展和阐明人类知识和推理的观点，将改善心理学作为推理卓越的典范。它打算从文献中提取科学对良好推理的了解，然后将这些经验教训应用于个人推理者和社会机构，包括科学机构。新的立场被定义为战略可靠性：推理卓越涉及到认知资源对稳健可靠推理策略的有效分配。战略可靠性是一种明确的基于成本效益的认识论，适用于真实的、有限的知者。认识论是关于知识和良好推理的哲学分支。该提案将把认识论从分析哲学的学术辩论中解放出来，并将其视为科学哲学的一个分支：规范认识论应该作为认知科学的一个基本（尽管通常是隐含的）特征得到适当的理解和研究。pi将通过挖掘和澄清特定科学分支（在这种情况下是心理学）的哲学（在这种情况下是认识论）前提，将科学哲学家的技能和兴趣带入认识论。一个关于如何最好地对经验问题进行推理的新理论，尤其是一个有半个世纪的经验支持的理论，必然会产生广泛而重大的影响。以下是四个影响：(1)我们从改善心理学中提取的卓越推理观点自然提出了一些具体的有用的经验教训，关于人们如何提高他们对因果关系、诊断和成本效益分析（以及其他问题）的推理。该提案的目的是开始利用未被充分利用的科学力量来提高人们对重要问题的推理能力。(2)目前，一些机构正在利用社会责任报告做出一些非常重要的决策，这些决策会对人们的生活产生深远的影响（从医疗诊断到信贷决策）。社会机构对战略战略文件的使用是科学影响社会的一个重要（但在很大程度上被忽视的）方式。该项目将为改善心理学的基本原理和假设提供一个清晰的表述。(3)我们从改善心理学中提取的推理卓越性观点可以为社会机构提供替代决策程序。然而，仍有许多社会机构尚未听取SPR文献的建议，包括精神疾病诊断，预测暴力行为，以及研究生和专业学校的录取。(4)改良心理学发展的预测模型在对有许多不同证据的复杂现象进行推理时特别有价值。这是对那些负责在各个层面上评估科学的人所面临的任务的完美准确的描述。科学哲学家、历史学家和社会学家利用科学史上的大量数据，试图从科学的方法和理论中得出一般的教训。那些在拨款审查委员会的科学家也通过考虑和权衡一些不同的证据来评估各种项目的前景。具有讽刺意味的是，那些评估科学的人并没有使用具有最强有力的科学支持的方法。对于从科学史中汲取教训的个人（Faust & Meehl 1992），以及对科学拨款提案进行评级的科学机构（NIH Panel Report, 1996，尤其是第5部分），都是如此。

A central issue in science studies is the relationship between evidence and hypothesis. But science studies is not the only discipline that investigates this relationship. Science studies has too long neglected the half-century's worth of fascinating and important empirical literature we call Ameliorative Psychology. A central finding of Ameliorative Psychology is that statistical prediction rules (or SPRs) are generally more reliable (and usually significantly more reliable) than human experts when it comes to making predictions about practically important matters (Meehl 1954, Sawyer 1966, Dawes 1994). This literature offers effective reasoning strategies that have undergone repeated and successful empirical tests. This proposal is to develop and articulate a view of human knowledge and reasoning that takes Ameliorative Psychology seriously as an exemplification of reasoning excellence. It intends to extract from the literature what science has learned about good reasoning, and then apply those lessons back to individual reasoners and social institutions, including the institutions of science. The new position is defined as strategic reliabilism: Reasoning excellence involves the efficient allocation of cognitive resources to robustly reliable reasoning strategies. Strategic reliabilism is an explicitly cost-benefit-based epistemology for real, bounded knowers. Epistemology is the branch of philosophy concerned with knowledge and good reasoning. The proposal will liberate epistemology from the scholastic debates of analytic philosophy and treat it as a branch of the philosophy of science: Normative epistemology is to be properly understood and investigated as an essential, though often implicit, feature of cognitive science. The PIs will bring the skills and interests of the philosopher of science to epistemology by unearthing and clarifying the philosophical (in this case, epistemological) presuppositions of a particular branch of science (in this case, psychology). A novel theory about how best to reason about empirical matters, especially one that comes with a half-century's worth of empirical support, is bound to have broad and significant impact. Here are four such impacts: (1) The view of reasoning excellence we extract from Ameliorative Psychology naturally suggests a number of concrete useable lessons about how people can improve their reasoning about causation, diagnosis, and cost-benefit analysis (among other issues). The proposal's aim is to begin to employ the underutilized power of science to improve people's reasoning about important issues. (2) SPRs are currently being used by institutions to make some very important decisions that deeply affect people's lives (from medical diagnoses to credit decisions). The use of SPRs by social institutions is a significant (and largely ignored) way in which science is affecting society. The project will offer a clear articulation of the principles and assumptions underlying Ameliorative Psychology. (3) The view of reasoning excellence we extract from the Ameliorative Psychology can suggest alternative decision-procedures for social institutions. However, there are still many social institutions that have yet to heed the advice of the SPR literature, including diagnostics for mental illness, predicting violent behavior, and admissions to graduate and professional school. (4) The predictive models developed by Ameliorative Psychology are especially valuable for reasoning about complex phenomena about which one has many different lines of evidence. This is a perfectly accurate description of the task facing those who are charged with assessing science at various levels. Philosophers, historians and sociologists of science take the huge mass of data that is the history of science and try to draw general lessons about the methods and theories of science. Those scientists on grant review boards also make assessments about the promise of various projects by considering and weighing a number of different lines of evidence. It is ironic that those who assess science have not used methods that have the strongest scientific support. This is true for individuals extracting lessons from the history of science (Faust & Meehl 1992), and for scientific institutions rating scientific grant proposals (NIH Panel Report, 1996, especially section 5).