“Inferring from the behaviour of a whole to a quality of its parts expressed only when they are acting as parts of that whole is just the opposite of reduction, understood as explaining the behaviour of a whole in terms of the properties of its parts when these parts are considered in isolation (McMullin, 1972). The former inference might be better described as explaining the parts in terms of the whole. The whole in such a case might still be said to be nothing more in ontological terms than a collection of its parts, but only on condition that the ‘parts’ are defined in equally ontological terms by the role they play in the whole. Reductionism and holism point in different ways.”