Résumé : The measurement of birefringence variations related to nerve activity is a promising label-free technique for sensing compound neural action potentials (CNAPs). While widely applied in crustaceans, little is known about its efficiency on mammal peripheral nerves. In this work, birefringence recordings to detect CNAPs, and Stokes parameters measurements were performed in rat and lobster nerves. While single-trial detection of nerve activity in crustaceans was achieved successfully, no optical signal was detected in rats, even after extensive signal filtering and averaging. The Stokes parameters showed that a high degree of polarization of light is maintained in lobster sample, whereas an almost complete light depolarization occurs in rat nerve. Our results indicate that depolarization itself is not sufficient to explain the absence of birefringence signals in rats. We hypothesize that this absence comes from the myelin sheets, which constraint the birefringence changes to only take place at the nodes of Ranvier.