Why a descriptive theory of agency is inadequate
In the contemporary debates about which approach is the right way to understand agency, the descriptive theories of agency are typically pitted against “normative” theories of agency. I aim to lay bare why I think this is misguided.
The problem with aiming for a descriptive theory of agency is that this conflates two forms of agency as if they were the same thing.
Firstly, it conflates the ability to accomplish one's goals with the ability to choose one's goals. Let us call goal-accomplishment agentic agency or agenticness for short. Agenticness can be constrasted with goal-choosing which I call autonomous agency or autonomy. The problems posed by autonomous agency are not answerable through the empiricism alone. This is because self (auto) and rule (nomos) are not empirically discoverable. That is to say, selves and rules are not waiting out there in the world to be uncovered through scientific methods. Selves and rules are not natural kinds—i.e. intellectual categories that reflect the world independent of human goals. Selves and rules are constituted by the very goals that descriptivists seem to presuppose.
A descriptive theory of agency aims to accurately map the current territory to accomplish a given goal. But this is only true when one assumes the givenness of the goal i.e. the goal is fixed and cannot evolve according to changing circumstances. The ability to alter one's goals and to gain new ones as well as discarding old ones will depend on what values are held by such an autonomous agent capable of goal-choosing. Imagine a highly intelligent agent capable of predicting multiple possible outcomes which accomplish different goals. How ought it to decide which goals to achieve? By coming up with a perfect descriptive model of the current world and the goal world and all the causal steps it will take from one to the other? Suppose it already has this map not just for one of these goal worlds, but for many of these goal worlds. Now it has to choose which goals are worth pursuing. Deciding the choiceworthiness of a goal is not solvable through more descriptions of the world.
Some philosophers have suggested that choiceworthiness can be maximized by simply presupposing a theory-neutral scale of choiceworthiness, in the form of a universal scale (2020, 133). The problem with this account is that this so-called universal already presupposes there are a set of universal values that exist outside of theories. We are still left with the problem of which goal are we meant to pursue? The dilemma essentially boils down to the question of “What do we want?”
And that is a question that descriptive theories of agency cannot answer.