Scientific discovery is the emergence of novel insights about the natural world. These insights are characterized by a new explanatory power, greater elegance or simplicity than existing theories, and by the presence of new evidence or experimental results. These discoveries lead to new applications, such as improved methods for mining, agriculture, health care and communication. They also provide a basis for scientific predictions that allow for novel technological developments and advances in the natural sciences, engineering, technology and humanities.
Philosophers have disagreed about the possibility of a philosophy of scientific discovery. One prominent view is that it is not possible to prescribe a logical method for generating new ideas or to reconstruct logically the process of discovery. This view, known as the non-inferential way, reflects the view that scientific knowledge is produced through “black box” processes that are inherently opaque to rational analysis (see Popper 1961).
Others, however, hold that there is a logical path to new discoveries. One line of argument, developed most fully by Norwood R. Hanson, is that the act of devising a new hypothesis follows a distinctive logic that differs from deductive and inductive reasoning. This logic, called the logic of discovery, includes the criterion of judging whether a happy thought has sufficient predictive value and is sufficiently promising to warrant further investigation.
Other approaches to the methodology of discovery rely on problem-solving models in order to account for the emergence of new ideas. These model-based approaches resemble the AI-based theories of scientific discovery discussed in section 6. They treat the articulation and verification stages as iterative algorithms whereby heuristic rules help to filter available data. These models also involve a form of generative justification whereby new predictions are tested in order to verify the theory resulting from the articulation stage.