Nicholas Shea
Affiliation: University of London
Category: Philosophy
Keywords: agency, primitive agent, rational agent, goal-directedness, purposes, philosophy of cognitive science
Date: Tuesday 2nd of September
Time: 17:00
Location: Maria Skłodowska-Curie Hall (123)
View the full session: Agency
(Attachment includes the text below, plus references and formatting)
There is a striking parallel between Tyler Burge’s account of primitive agency and Christine Korsgaard’s account of rational agency. Burge’s account is designed to apply to simple animals that lack representational states, and to extend even to motile bacteria. Korsgaard’s account is designed to apply to the most sophisticated reflective reasoners. It seems as if they must be concerned with different phenomena. Nevertheless, they both hold that agency supports a distinction between behaviour caused by a mere part of the agent and behaviour produced by the agent itself.
‘to regard some movement of my mind or my body as my action, I must see it as an expression of my self as a whole, rather than as a product of some force that is at work on me or in me.’ (Korsgaard 2009, p. 18)
‘We distinguish firmly between an animal’s actions, on one hand, and both things that happen to the animal and processes that occur within the animal, on the other.’ (Burge 2009, p. 256)
This suggests that understanding a system to be an agent involves a common schema, albeit one that applies in different ways in the two kinds of case. There is some system, mechanism or organising principle whose operation constitutes causation by the agent. Effects that are produced outside of that system do not qualify as exercises of agency. The type of agency involved differs substantially between the two accounts, but because they conform to a common schema for thinking about agency, both accounts include a ‘not by a part’ condition.
This condition cannot be a matter of a simple part-whole distinction. Whatever mechanism or form of organisation counts as the operation of the whole-agent system, it will still be realized by parts and their relations. So the fact that a property of a part of the agent is a difference-maker does not on its own disqualify a movement from being an exercise of agency.
One influential account of the distinction traces back to Frankfurt (1978). Frankfurt appeals to the operation of a guidance or control mechanism. Some movements are guided by a mechanism ‘with which the agent can be identified’, others not. While the requirement that the agent – the non-part – exercises guidance or control over behaviour is doubtless important, that cannot be the whole story. Korsgaard’s account excludes behaviours like weeping for which a suite of behavioural responses are coordinated across the whole organism (breathing, movements of the whole body, producing tears, etc.). Korsgaard also wants to exclude cases where the agent simply gives in to desire without exercising their rational capacity to decide which principles to act on, and hence what to value. Burge’s account also implies that some cases where a whole-organism control system is operative are nevertheless excluded.
‘your movement will not be an action unless it is attributable to you—to you as a whole or a unified being—rather than merely to something in you. And the task of deliberation is to determine what you—you as a whole or a unified being—are going to do.’ (Korsgaard 2009, pp. 125-6)
‘the relevant notion of action is grounded in functioning, coordinated behavior by the whole organism, issuing from the individual’s central behavioral capacities, not purely from sub-systems.’ (Burge 2009, p. 260)
The parasitic lancet fluke Dicrocoelium dendriticum gets inside the brain of an ant and hijacks its behavioural control mechanism. The parasite makes the ant climb to the top a grass stalk and stay there (where it is prone to being grazed, completing the parasite’s life cycle). The ant retains its capacity for coordinating the movements of its body and behaving in goal-directed ways. Its behaviour when infected by the parasite does not meet Burge’s requirement that agency should be functional for the individual – function here being a matter of biological functions of the whole organism, like contributing to fitness or survival for mating. The ant’s coordinated, controlled, goal-directed behaviour on this occasion is serving a function for the parasite, not for the ant.
These examples point to a deeper insight shared by Korsgaard’s and Burge’s accounts. Agents display a characteristic unity, and both hold that this is a unity, not just of control or the coordination of behaviour, but of purpose. Primitive agents direct their behaviour at performing functions for the individual. Rational agents choose which acts to perform for which reasons. They achieve unity by deliberation, and that includes deciding what to value. For both, agency involves a commitment about the purpose or value for which behaviour is performed (in addition to a condition to do with the coordination and control of movement).
This means that the agency schema applies to humans at two levels (at least). We have mechanisms for the coordination of whole-organism behaviour serving biological functions for the individual, for example when we reflexively duck to avoid a projectile. These are mechanisms of primitive agency. The knee-jerk reflex, by contrast, is behaviour produced by a mere part. It is not agentive. We also have the capacity for rational, reflective agency. Relative to that capacity the reflexive ducking counts as the operation of a mere part – as non-agentive – even though it involves a mechanism for whole-organism control of behaviour.
What is interesting, for our question, is that at both levels the distinction is not just a matter of which system is producing coordination or exercising control. In both cases, for a system to count as agentive there has to be a connection to purpose or value, and to purposes or values of the whole agent itself.