Optimizing Scientific Reasoning

Abstract

B. Bahrami et al. (“Optimally interacting minds,” Reports, 27 August, p. [1081][1]) found that social interactions can enhance the quality of decision-making. The Perspective by M. O. Ernst (“Decisions made better,” 27 August, p. 1022) suggested that Bahrami et al. 's results could be applied to real-world issues such as soccer refereeing. I propose another application: the nature of scientific reasoning. Science is the collective selection of theories given available evidence and Bayesian probabilistic deduction ([ 1 ][2]). But scientists reason in ways that go beyond the requirements of Bayes. For example, they collectively work to ensure that scientific ideas are stated clearly, exposed to open criticism, and judged purely on their merits (blind review). Bahrami et al. 's research sheds light on why such norms are central to science. Bahrami et al. found that inferences made collectively by a group were superior to inferences made with only individual judgment. However, some social interactions were more effective than others in arriving at a correct decision. Science is the accumulated product of many centuries of experience in how best to organize research communication to optimize its capacity for collective Bayesian inference. Bahrami et al. 's experimental paradigm, although not directly a science reasoning task, provides a means by which these norms central to scientific reasoning can be researched and perhaps even improved. 1. [↵][3] 1. C. Howson, 2. P. Urbach , Scientific Reasoning: The Bayesian Approach (Open Court, La Salle, IL, 1989). [1]: /lookup/doi/10.1126/science.1185718 [2]: #ref-1 [3]: #xref-ref-1-1 View reference 1 in text