What's wrong with investing the sub-personal with intentionality?

Gallagher, in his survey of ways of relating cognitive science to phenomenology suggests that we need to be vigilant against investing sub-personal processes with intentional contents.

Now, to be sure, there is definitely something wrong with arbitrarily or unreflectively investing, for example, beliefs and desires into brain processes. Dennett has done a wonderful job of articulating this in his discussion of "What the frog's eye tells the frog's brain".

However, what empirical grounds are there for denying that these processes (just calling them processes rather than behaviours seems to prejudge the question) are directed toward anything at all? It seems to me that ruling out any and all talk of intentionality at this level is just as metaphysical (in the bad sense) as ruling it in.

It rests on the presupposition that the personal and the subpersonal are discrete levels of description. There are however good reasons for undermining their discreteness, not least because though treating them as discrete allows us to describe pathological behaviours, it makes normal behaviour seem mysterious.

Ron McClamrock's argument for a task-centred version of the information processing hypothesis, which he claims will make it compatible with externalism, will provide an opportunity for us to propose a conception of the subpersonal which exhibits a kind of intentionality that is scientifically bearable. Our conception involves expanding on McClamrock's suggestion that a task-based account of cognition might even be extended to the design level.

Our conception attempts to take account of the way in which scientific taxonomies are developed, or rather, the responsibility to the organism that prompts scientific taxonomies to change. Recognizing this responsibility does not allow us to shortcircuit the development of these taxonomies, but it does capture the sense in which these taxonomies are always a scientific attempt to do justice to the otherness of its objects. In this case, the move to enactivism and DST is an attempt to bear witness to the self-conceptuality of the organism (Morris), i.e. to the fact that the dimension of its description are and should be relative to its strategies of inherence.

So we revise the relation of embeddedness, from something akin to the relations of immersion or component-to-system, to something more like a inalienable strategic relation. In doing so, we are able to inject a little intentionality - an intentionality that needn't be equated with the intentionality of projects - into our third person descriptions.