This paper presents an attempt to rationaly rebuild the discovery process in medicine. Our aim is to reconstruct Claude Bernard’s intellectual pathway leading to important discoveries, in particular his understanding of the effects of curare. Achieved in collaboration with epistemologists, we refer to the notebooks where Bernard recorded his experiments. Based on this material, this paper presents a computational model of Bernard’s activity. Our study shows that Bernard did not only used inductions, but also deductions and abductions. The deductions anticipate the consequences of working hypothesis; experiments attempt to confirm or to infirm those hypothesis; then abductions generate new hypothesis that explained unexpected observations. We focus here on the deductive part of this process, with a virtual laboratory that allows the construction of virtual experiments associated with different working hypotheses. Then, we show how this deductive part takes place in the discovery process and how it is related to the abductive and the inductive steps.