Ethics has Evidence Too

A tenet of tra­di­tional ra­tio­nal­ity is that you can’t learn much about the world from arm­chair the­o­riz­ing. The­ory must be epiphe­nom­e­nal to ob­ser­va­tion—our the­o­ries are func­tions that tell us what ex­pe­riences we should an­ti­ci­pate, but we gen­er­ate the the­o­ries from *past* ex­pe­riences. And of course we up­date our the­o­ries on the ba­sis of new ex­pe­riences. Our the­o­ries re­spond to our ev­i­dence, usu­ally not the other way around. We do it this way be­cause it works bet­ter then try­ing to make pre­dic­tions on the ba­sis of con­cepts or ab­stract rea­son­ing. Philos­o­phy from Plato through Descartes and to Kant is re­plete with failed ex­am­ples of the­o­riz­ing about the nat­u­ral world on the ba­sis of some­thing other than em­piri­cal ob­ser­va­tion. Socrates thinks he has de­duced that souls are im­mor­tal, Descartes thinks he has de­duced that he is an im­ma­te­rial mind, that he is im­mor­tal, that God ex­ists and that he can have se­cure knowl­edge of the ex­ter­nal world, Kant thinks he has proven by pure rea­son the ne­ces­sity of New­ton’s laws of mo­tion.

Th­ese mis­takes aren’t just found in philos­o­phy cur­ricula. There is a long list of peo­ple who thought they could de­duce Eu­clid’s the­o­rems as an­a­lytic or a pri­ori knowl­edge. Epicy­cles were a re­sponse to new ev­i­dence but they weren’t a re­sponse that truly priv­ileged the ev­i­dence. Geo­cen­tric as­tronomers changed their the­ory *just enough* so that it would yield the right pre­dic­tions in­stead of let­ting a new the­ory flow from the ev­i­dence. Same goes for pre-Ein­stei­nian the­o­ries of light. Same goes for quan­tum me­chan­ics. A kludge is a sign some­one is priv­ileg­ing the hy­poth­e­sis. It’s the same way many of us think the Ital­ian po­lice changed their hy­poth­e­sis ex­plain­ing the mur­der of Mered­ith Kercher once it be­came clear Lu­mumba had an al­ibi and Rudy Guede’s DNA and hand prints were found all over the crime scene. They just re­placed Lu­mumba with Guede and left the rest of their the­ory un­changed even though there was no longer rea­son to in­clude Knox and Sol­lecito in the ex­pla­na­tion of the mur­der. Th­ese the­o­ries may make it over the bar of tra­di­tional ra­tio­nal­ity but they sail right un­der what Bayes the­o­rem re­quires.

Most peo­ple here get this already and many prob­a­bly un­der­stand it bet­ter than I do. But I think it needs to be brought up in the con­text of our on­go­ing dis­cus­sion of nor­ma­tive ethics.

Un­less we have rea­son to think about ethics differ­ently, our nor­ma­tive the­o­ries should re­spond to ev­i­dence in the same way we ex­pect our the­o­ries in other do­mains to re­spond to ev­i­dence. What are the ex­pe­riences that we are try­ing to ex­plain with our eth­i­cal the­o­ries? Why bother with ethics at all? What is the mys­tery we are try­ing to solve? The only an­swer I can think of is our eth­i­cal in­tu­itions. When faced with cer­tain situ­a­tions in real life or in fic­tion we get strong im­pulses to re­act in cer­tain ways, to praise some par­ties and con­demn oth­ers. We feel guilt and some­times pay amends. There are some ac­tions which we have a visceral ab­hor­rence of.

Th­ese re­ac­tions are for ethics what mea­sure­ments of time and dis­tance are for physics—the ev­i­dence.

The rea­son ethi­cists use hy­po­thet­i­cals like the run­away trol­ley and the un­will­ing or­gan donor is that differ­ent nor­ma­tive the­o­ries pre­dict differ­ent in­tu­itions in re­sponse to such sce­nar­ios. Short of ac­tu­ally set­ting up these sce­nar­ios for real, this is as close as ethics gets to con­trol­led ex­per­i­ments. Now there are prob­lems with this method. Our in­tu­itions in fic­tional cases might be differ­ent from real life in­tu­itions. The sce­nario could be poorly de­scribed. It might not be as con­trol­led an ex­per­i­ment as we think. Or some fea­tures could be cloud­ing the is­sue such that our in­tu­itions about a par­tic­u­lar case might not ac­tu­ally falsify a par­tic­u­lar eth­i­cal prin­ci­ple. Just as there are op­ti­cal illu­sions there might be eth­i­cal illu­sions such that we can oc­ca­sion­ally be wrong about an eth­i­cal judg­ment in the same way that we can some­times be wrong about the size or ve­loc­ity of a phys­i­cal ob­ject.

The big point is that the way we should be rea­son­ing about ethics is not from first prin­ci­ples, a pri­ori truths, defi­ni­tions or psy­cholog­i­cal con­cepts. Kant’s Cat­e­gor­i­cal Im­per­a­tive is a paradigm ex­am­ple of screw­ing this up, but he is hardly the only one. We should be look­ing at our eth­i­cal in­tu­itions and try­ing to come up with the­o­ries that pre­dict fu­ture eth­i­cal in­tu­itions. And if your the­ory is out­putting re­sults that are sys­tem­at­i­cally or rad­i­cally differ­ent from ac­tual eth­i­cal in­tu­itions then you need to have a damn good ex­pla­na­tion for the dis­crep­ancy or be ready to change your the­ory (and not just by adding a kludge).