Behavior analysts assume that private events like thinking and feeling have the same kinds of physical dimensions as other events in the world. They still claim, however, that private events can never be initiating causes of behavior. I point out that this position seems theoretically inconsistent, though exactly what qualifies as an initiating cause is unclear. I go on to show that focusing almost exclusively on external causes may impede realization of the behavior analytic goals of prediction and control. I argue that assuming private events to sometimes be full and real causes of behavior is fully compatible with the pragmatic stance of behavior analysis, which assumes propositions to be true when they work, and not because people agree. I end by recommending that the division between initiating and non-initiating causes be discarded.