Relinquishing experimental control to improve translation

Abstract Here I draw on a recent movement in preclinical biomedical research – use of what are often called “dirty mice” – to challenge the standard view that less controlled experiments have diminished epistemic power. Dirty mice are research mice with diverse microbial exposures, and dirty mouse methods, I argue, are strategies for loosening the control that researchers have over their experiments. I use recent experimental results to show why, as compared to conventional mouse models, dirty mice may better support the translation of findings from laboratories to human clinical contexts. I discuss how to evaluate the claim that dirty mice, as compared to conventional mice, both (a) more closely match relevant human features and (b) more accurately predict human responses (to drugs and interventions). Thus, contra the control ideal, deliberately relinquishing some experimental control improves an epistemic desideratum (in this case, translatability). This paper contributes to literature on the philosophy of experiment, and the final upshot will be a suggestion for when scientists in other fields might consider purposely designing less controlled experiments.