This project aims to acquire a deeper understanding of the process of scientific discovery.
One aspect of the project was an ESRC funded project on scientific discovery, which combined cognitive science and computational techniques to semi-automate the construction of computational models from data. Genetic programming was used to create computer programs producing cognitive models which simulated results from psychology experiments. A system including a language for theories, a class of process based models, and algorithms for evolving models was created and applied to model the delayed-match-to-sample task, Hick-Hyman task, and categorisation experiments. In doing this, a new way to evolve cognitive science theories using computers was developed, which opens up new possibilities for efficiently employing data to further theoretical knowledge. See Genetically Evolving Models in Science (GEMS) for a continuation of this line of research.
A second and related aspect of the project is the broader study of computational scientific discovery, including the refinement and assessment of theories. Creativity is at the heart of scientific research in the natural and social sciences as well as being the basis of innovation in business and industry. Psychology research suggests that the Darwinian mechanisms of variation and selection underpin human creativity and philosophers of science such as Popper have also emphasised the evolutionary properties of scientific knowledge. The project challenges current views about the generation of scientific theories being an exclusively human activity by altering the perception of social science theories from that of being abstract objects to computationally tractable concrete data structures.
The third aspect concerns the extent to which scientific discovery is possible in social science and encompasses a broad range of methodological issues.
Current research includes linking discovery to statistical questions about evidence and the generalisability of findings.
-
Addis M., Gobet F., Lane P.C., and Sozou, P.D (eds.) (2014). Proceedings of the 50th Anniversary Convention of the AISB: Computational Scientific Discovery Symposium. London: AISB.
-
Addis M., Gobet F., Lane P.C. and Sozou P.D. (2019). Semi-automatic generation of cognitive science theories. In Scientific Discovery in the Social Sciences. Springer: Heidelberg.
-
Addis M., Lane P.C., Sozou P.D. and Gobet F. (eds.) (2019). Scientific Discovery in the Social Sciences. Springer: Heidelberg.
-
Addis M., Sozou P.D., Lane P.C. and Gobet F. (2016). Computational scientific discovery and cognitive science theories. In Müller V. (ed.), Computing and Philosophy: Proceedings of IACAP 2014. Springer: Heidelberg.
-
Dimarogkona M., Addis M. and Stefaneas P. (2019). Syntax, semantics and the formalisation of social science theories. In Scientific Discovery in the Social Sciences. Springer: Heidelberg.
-
Gobet F., Addis M., Lane P.C. and Sozou P.D. (2019). Introduction: scientific discovery in the social sciences. In Scientific Discovery in the Social Sciences. Springer: Heidelberg.
-
Lane P.C., Sozou P.D., Gobet F. and Addis M. (2016.) Analysing psychological data by evolving computational models. In Wilhelm A. and Kestler H. (eds.), ECDA Conference 2014 Bremen: Analysis of Large and Complex Data. Springer: Heidelberg.
-
Sozou P.D., Lane P.C., Addis M. and Gobet F. (2017). Computational scientific discovery. In Magnani L. and Bertolotti T. (eds.), Springer Handbook of Model-Based Science. Springer: Heidelberg.
-
Symposium The Foundational Significance of Abstract Model Theory, 15th Congress on Logic, Methodology, and Philosophy of Science, Helsinki, 3-8 August 2015
-
Scientific Discovery in the Social Sciences, London School of Economics, 30-31 January 2015
-
Symposium Computational Scientific Discovery, 50th Anniversary Convention of the AISB, Goldsmiths College, 1-4 April 2014
-
Rethinking Theory Construction in Social Science, London School of Economics, 11 March 2014