. An attempt is made to identify a concept of ‘downward causation’ that will fit the claims of some recent writers and apply to interesting cases in biology and cognitive theory, but not to trivial cases. After noting some difficulties in achieving this task, it is proposed that in interesting cases commonly used to illustrate ‘downward causation’, (a) regularities hold between multiply realizable properties and (b) the explanation of the parallel regularity at the level of the realizing properties is non-trivial. (...) It is argued that the relation between a realizable property and the property that realizes its effect in a particular case is not usefully regarded as a species of causation and that use of the concept of downward causation deflects our attention from our central explanatory tasks. (shrink)