Mechanistic Models and the Explanatory Limits of Machine Learning

We argue that mechanistic models elaborated by machine learning cannot be explanatory by discussing the relation between mechanistic models, explanation and the notion of intelligibility of models. We show that the ability of biologists to understand the model that they work with (i.e. intelligibility) severely constrains their capacity of turning the model into an explanatory model. The more a mechanistic model is complex (i.e. it includes an increasing number of components), the less explanatory it will be. Since machine learning increases its performances when more components are added, then it generates models which are not intelligible, and hence not explanatory.

[1]  P. Machamer,et al.  Thinking about Mechanisms , 2000, Philosophy of Science.

[2]  E. Ratti.  ‘Models of’ and ‘Models for’: On the Relation between Mechanistic Models and Experimental Strategies in Molecular Biology , 2020, The British Journal for the Philosophy of Science.

[3]  William Stafford Noble,et al.  Machine learning applications in genetics and genomics , 2015, Nature Reviews Genetics.

[4]  S. Glennan.  The New Mechanical Philosophy , 2017 .

[5]  T. Ideker,et al.  The cancer cell map initiative: defining the hallmark networks of cancer. , 2015, Molecular cell.

[6]  Nancy J. Nersessian,et al.  Creating Scientific Concepts , 2008 .

[7]  Carl F. Craver,et al.  When mechanistic models explain , 2006, Synthese.

[8]  W. Bechtel.  Using computational models to discover and understand mechanisms. , 2016, Studies in history and philosophy of science.

[9]  H. W. de Regt,et al.  Understanding Scientific Understanding , 2017 .

[10]  Helen E. Longino,et al.  Discovering Complexity: Decomposition and Localization as Strategies in Scientific Research , 1995 .

[11]  Margaret Morrison,et al.  Models as Mediating Instruments , 1999 .

[12]  C. Craver.  Explaining the Brain: Mechanisms and the Mosaic Unity of Neuroscience , 2007 .

[13]  David Haussler,et al.  Inference of patient-specific pathway activities from multi-dimensional cancer genomics data using PARADIGM , 2010, Bioinform..

[14]  M. Hegarty.  Mechanical reasoning by mental simulation , 2004, Trends in Cognitive Sciences.

[15]  L. Darden,et al.  In Search of Mechanisms: Discoveries across the Life Sciences , 2013 .

[16]  V. Nagaraja,et al.  Diverse Functions of Restriction-Modification Systems in Addition to Cellular Defense , 2013, Microbiology and Molecular Reviews.

[17]  Gary D. Bader,et al.  The mutational landscape of phosphorylation signaling in cancer , 2013, Scientific Reports.

[18]  Henk W. De Regt,et al.  Scientific understanding: truth or dare? , 2014, Synthese.

[19]  Ronald N. Giere,et al.  An agent-based conception of models and scientific representation , 2009, Synthese.

[20]  Arnon Levy,et al.  What was Hodgkin and Huxley’s Achievement? , 2014, The British Journal for the Philosophy of Science.

[21]  M. Hegarty.  Capacity Limits in Mechanical Reasoning , 2002 .