Models of human scientific discovery

Goldstone, R. L., Gopnik, A., Thagard, P., & Ullman, T. D. (2018).  Models of human scientific discovery.  Proceedings of the 40th Annual Conference of the Cognitive Science Society. (pp. 29-30). Madison, Wisconsin: Cognitive Science Society.

The scientific understanding of scientific understanding has┬ábeen a long-standing goal of cognitive science. A satisfying┬áformal model of human scientific discovery would be a┬ámajor intellectual achievement, requiring solutions to core┬áproblems in cognitive science: the creation and use of apt┬ámental models, the prediction of the behavior of complex┬ásystems involving interactions between multiple classes of┬áelements, high-level perception of noisy and multiply┬áinterpretable environments, and the active interrogation of a┬ásystem through strategic interventions on it ÔÇô namely, via┬áexperiments. Over the past decades there have been┬ánumerous attempts to build formal models that capture what┬áPerkins (1981) calls some of the ÔÇťmindÔÇÖs best workÔÇŁ ÔÇô┬áscientific explanations for how the natural world works by┬ásystematic observation, prediction, and testing. Early work┬áby Hebert Simon and his colleagues (Langley, Simon,┬áBradshaw, & Zytkow, 1987) developed production rule┬ásystems employing heuristics to tame extremely large┬áconjoint search spaces of experiments to run and hypotheses┬áto test. Qualitative physics approaches seek to understand┬áphysical phenomena by building non-numeric, relational┬ámodels of the phenomena (Forbus, 1984). Some early┬áconnectionist models interpreted scientific explanation in┬áterms of emerging patterns of strongly activated hypotheses┬áthat mutually support one another (Thagard, 1992).

Download PDF of paper here