Random Sample Consensus: A Paradigm for Model Fitting with Applications to Image Analysis and Automated Cartography


Fischler, M. A., & Bolles, R. C. (1981). Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Communications of the ACM, 24(6), 381-395.


It may seem odd for two cognitive scientists, each with little specific expertise in social psychology, to present a chapter that focuses on social cognition. Indeed, our past work may seem much more in the realm of scientific reasoning than in that of social reasoning. But one question that we have been asking, both of ourselves and of our colleagues, is, “What is the difference between ‘scientific reasoning’ and plain old ‘reasoning’?” Generally, people hem and haw when confronted with this question, then speak of the latter as if it were social reasoning—and quite often, they mention socially based ruminations that involve suboptimal decisions, faulty heuristics, and inappropriately biased values, goals, and the like (see Gigerenzer, 1991; Tversky & Kahneman, 1974, and many others). Useful follow-up questions to such respondents include, “Well, is the difference between these two sorts of reasoning qualitative or quantitative?” Put another way (as many—including Einstein, 1950,—seem to have occasionally wondered), “Is scientific reasoning
just (a) more likely to employ formal tools (like deduction or mathematics) and/or (b) more likely to involve the vigilant search for disconfirmation—something that just plain folks (Lave, 1988, p. 4) do, but less frequently?

Put rather bluntly, we have not been able to reject the hypothesis that the word “scientific” in “scientific reasoning” is superfluous. In an era of specialization, we realize that it is a bit out of fashion to undifferentiate reasoning (although one can argue that interdisciplinary cognitive science itself similarly bucks the trend); still, we are more struck by how much of the everyday is found in scientific reasoning (and vice versa) than by how unique scientific reasoning is. Thus, we believe that the principles of reasoning that have been seen primarily as characterizing scientific reasoning can equally well be viewed as central to social reasoning .1
Bifurcating the set of reasoning processes into the social and the scientific is a bit like bifurcating a deity and still considering the encompassing religion to be monotheistic. In essence, adding either the modifier “social” or “scientific” seems unnecessary, unless one speaks about the domain being reasoned about (discussed later).

Read more from SRI