This site uses cookies, so that our service can work better. I understand

Marcin Miłkowski

Session:
Special Lecture 

Institute:
Institute of Philosophy and Sociology PAS, Warsaw, Poland
 
Website:
https://marcinmilkowski.pl/pl/o-mnie

ResearchGate:
https://www.researchgate.net/profile/Marcin_Mitkowski
 
Biography: 

Associate Professor in the Section for Logic and Cognitive Science at the Institute of Philosophy and Sociology, Polish Academy of Sciences. He published Explaining the Computational Mind (MIT Press 2013), awarded with the Tadeusz Kotarbiński Prize of the Section I of the Polish Academy of Sciences and the National Science Center Award for outstanding young scholars in social sciences and humanities in 2014. He was presented with Herbert A. Simon by Association for Computers in Philosophy (IACAP) for his signi cant con- tributions in the foundations of computational neuroscience (2015). He is now Principal Investigator of NCN SONATA BIS 5 grant “Co- gnitive Science in Search of Unity” (2015-2020).

Abstract:

Replicability and reproducibility of computational models has been somewhat understudied by “the replication movement”

In this talk, I draw on methodological studies into the replicability of psychological experiments and on the mechanistic account of explanation to analyze the functions of model replications and model reproductions in computational neuroscience. Model replicability, or independent researchers’ ability to obtain the same output using original code and data, and model reproducibility, or independent researchers’ ability to recreate a model without original code, serve different functions and fail for different reasons. This means that measures designed to improve model replicability may not enhance (and, in some cases, may actually damage) model reproducibility. Although both are undesirable, low model reproducibility poses more of a threat to long-term scienti c progress than low model replicability. Low model reproducibility stems mostly from authors’ omitting to provide crucial information in scientic papers and we stress that sharing all computer code and data is not a solution. Reports of computational studies should remain selective and include all and only relevant bits of code.