I finished the first volume of Parfit's On What Matters some time ago. The aim of the book is to reconcile consequentialism and Kantian deontic moral theory. The argument is largely constructed by analysis of various examples with an eye to deontic explanations, and then the deontic explanations by consequentialist theory. (Not quite that order, but I think that is one way to read the overall structure.) That volume has elicited a good deal of discussion of the nature and role of reasons, and I think volume 2 takes up the topic in some detail. (Have not read volume 2 as yet.) The book sort of went off the tracks for me relatively early on. The argument formed by examples or cases, and then some analysis. The examples are familiar variations of the Trolley Problems. The Trolley Problems are all variant cases asking about one's intuitions re whether it is morally permissible (or obligatory or prohibited) to cause one death in order to avoid killing a larger number. The cases are, e.g., may one throw a switch causing a trolley to switch tracks killing one person but thereby saving five others. Then you push the one person off a bridge, or open a trap-door, and so on. They are found in discussion of the doctrine of double effect, for example. Here is the problem from volume 1: Parfit introduces examples and then moves to the explanation of how the outcome is made to fit with one or another interpretation of deontic rules. But what happens when the intuition about how the cases should be decided are not shared? Parfit does not argue for particular interpretations of the cases; instead, the argument is from his intuition about the case to its fit with, e.g., maxim of ends as he interprets the maxim. If one does not share the intuition, the rest becomes hypothetical (in the sense that it just seeing how the machine runs). Which means there is no argument about what matters, only a complicated exegesis of what matters to Parfit. Maybe this concern is addressed in volume 2. I will have to read it to find out.