"An agent who flips a switch, thereby turning on the light, illuminating the room, and alerting a burglar acts only once, but his act has numerous descriptions, intentional under some but not others. Being intentional cannot, therefore, be a property of an act because if it were, an act would both have and not have the property of being intentional. [Davidson's] claim that "there is an irreducible difference between psychological explanations that involve the propositional attitudes and explanations in sciences like physics and chemistry" is an echo of Anscombe's position."
p. 9. Frederick Stoutland, 'Introduction: Anscombe's Intention in context,' in Essays on Anscombe's Intention. Edited by Anton Ford, Jennifer Hornsby and Frederick Stoutland. 2011. Cambridge, Massachusetts: Harvard University Press.
The quotation from Davidson highlights one of the most attractive features that Anscombe's account of intentional action proposes: to understand how we classify and interpret actions, how we explain them, without having to say that all the elements used in such explanations have to be coupled up with real, reducible or emergent, mental states. If we would accept such a strategy then we would have to treat philosophical theories of action and mind as offering a theoretical model that the empirical sciences can use when looking at how the type of animals or organisms that we humans are work.
Of course one might be drawn to a way of understanding action-explanation as non-reducible for many reasons. Many of the researchers working on Anscombe nowadays do so because they hold that there is some way in which agents act for reasons, and this way cannot be adequately captured by causalist, naturalist accounts. Thus we should look for alternatives and maybe Anscombe's Aristotelianism is such a way.
Another motivation might be to take her theory as a good outline of the model we employ to explain and understand actions. We might add that this model is not at all adequate to what is going on in the world. Thus we might be antirealists or fictionalists about the mental and agents, while trying to make understand what pattern people employ when they talk and think about actions. We employ many models that do not correspond to reality and aren't correct. But that doesn't mean that they cannot serve as if our interests are superficial in actions, i.e. if we are not scientists doing work on motivation, social interaction, neural coding, etc.
Just consider the following two examples of information lacking any correlate in reality influencing our actions: one might plausibly hold that the value or art is not a natural property artworks possess. It is rather something we attribute to them. Nevertheless, these values can influence our behavior in significant ways. For example if I believe that the Mona Lisa is an especially valuable painting, I would like to own it, or at least see it and I may even exert a significant amount of energy to travel and have a quick look at it.
Also, we might think, rules of politeness are not grounded in reality in a way that anything in nature would decisively determine what these rules are. But this does not mean that we shouldn't adhere to such rules, or that they cannot make our lives easier and more pleasant (as well as they can be frustrating and form all sorts of obstacles, but nevertheless they only do so because people take them to be real and act accordingly).
