Does agency always imply choice?

Question

I am a biologist but have strong interests in the most fundamental questions of Life, so I had to educate quite a bit on philosophy to be able to delve into them. One of the concepts I had to become familiar with was that of agency. I now think there is simply no way to be familiar with theoretical biology if you haven't heard of the concept of agency. It is often implied that what distinguishes animate from inanimate matter is the agency of the first over the second. Basically, it means living organisms can act on their own behalf, while inanimate objects always follow a predetermined path. Am I correct?

Then, the question comes-can an agent choose not to use its ability to counter the effects of the environment through its agency and follow a predetermined path just like any other physical object. But then is it really an agent?

To translate the question in a little but more "biological"(and "scientific" for that matter)way consider this. If a molecule "chooses" to evolve into a more fit version of itself for a certain process does it really has this "choice" and the agency for this matter or is it predetermined to do so. There are several papers which I think pertain very well to theoretical biology here, here and here among many others which discuss the role of the notion of agency/autonomy for a proper definition of Life. As far as I know agency always(I want to pay attention here to this key point)implies its bearer to have the free will to take one course of action or the other. I can very well see how this concept pertains to biology as a molecule in a sense has "the choice"(under the circumstances used to defined the word)to transform its "progeny" into different versions of itself. This is how chemical evolution works and chemical evolution is an important step on the road to Life. But ultimately it is driven by natural laws and as such it is under the constraint of those laws and not an independent process itself. The molecule can evolove different forms of itself but it can not not choose to evolve. Then is its choice of the particular form it evolves into an act of agency or not? And more precisely when does agency starts for such molecules-when they start to change under the pressure of the environment or when they self-generate this change going into a cycle of change where the change of the previous step is the cause of the change in the next step and so on until a maximum fitness is achieved. Does the concept of agency implies this process is the "choice" of the molecules or is it an extension of the changes started by the environment to them(therefore, implying agency isn't the reason of the particular "choice" made by the molecules)which are determined by the laws of nature and therefore the molecules aren't themselves responsible for them?

Does the agency starts when the molecule can choose its new form and generate change by itself or when it starts to change no matter the causes? Is agency equal to the ability to change or is it equal to changes under only special circumstances driven by the changes in the molecule itself(e.g. its "choice" to evolve)?

Answer 1

There are two positions on the question of freewill:

• Incompatibilism is the position that freewill and determinism are incompatible: Either the universe is deterministic, or we have freewill, but not both. In this world view we have freewill only if we are able to choose among multiple possible outcomes, or as some put it "we could do otherwise". Incompatibilism then breaks down to two positions: hard determinism (we don't have freewill), and libertarianism (we have metaphysical freewill - not to be confused with political libertarianism). One can then speak of "agent causation": Agent-Causality is the idea that agents can start new causal chains that are not pre-determined by the events of the immediate or distant past and the physical laws of nature.

• Compatibilism is the position that determinism and freewill are compatible, but only because we have to change the definition of freewill. Freewill is not the ability to choose among multiple possible courses of action, it is instead the ability to act freely according to one's own motivations. Once we have refined our definition of freewill, then causal determinism and freewill are compatible, hence the name of this position. Agents are not able to start new causal chains, but they still have freewill because they have agency.

So now we have two definitions of freewill: Metaphysical freewill and compatibilist freewill.

In the case of compatibilism, your question is dissolved: agency is the same thing as freewill,.

In the case of incompatibilsim: freewill is a metaphysical concept, and agency implies freewill, but it is not the same thing as freewill.

Daniell Dennett writes and speaks on this topic, see his book "Freedom Evolves", and his presentations on youtube.

Answer 2

The short answer is no, as used today it is a much weaker concept. The use of "agency" evolved quite a bit from the original meaning in the somewhat cryptic passage from Aristotle's De Anima about "agent intellect". What is meant by "agent" today is any kind of entity that can initiate actions for a purpose (real or perceived). While "initiating actions" might suggest ability to start new causal chains, as in "agent causation" and "free will", the common use is much more permissive. For instance, an animal acting on instinct is an actor with a purpose, and therefore an agent, but, assuming that instinctive actions are automatic, no free will or self-causation is involved here.

In fact, the language of agency is often used as a way of getting on with practical research while avoiding the perennial controversy over the empirically moot questions, like the "nature" of causality, consciousness, free will, etc. In other words, whenever it is practically useful to employ teleological and/or intentional explanations, the language of agency can be applied. "Actors with a purpose" can be plants (sunflowers move their leaves to face the sun), animals (dog is looking for a bone), animal colonies (ants are trying to surround the intruder), ecosystems, organizations, machines (my car is acting up), robots, etc. It is an admission that there is similarity enough with "truly intentional" human action for the language to be useful, but in and of itself its use presupposes nothing about ontology. Indeed, it may indicate nothing more than antropomorphic stereotypes about animals, or "delegation" of human intentionality to machines.

Allison Adam surveys the recent uses of "agency" in Artificial Knowing:

"Hanging onto intentionality points to one of the last refuges of enlightenment thinking, the uniqueness of the human animal... Dennett’s intentional stance offers a get-out clause, a way of acting as if certain objects have intentionality without worrying about whether they actually do have it. Taking an intentional stance towards something is a way of granting it some level of agency... Both in the popular sub-domain of distributed AI (DAI), where knowledge is distributed through several knowledge bases, or where intelligent agents act in concert to solve a problem, and also in robotics, it is curious to see that the language of agency and intentionality abounds, possibly much more so than in other areas of symbolic AI.

...using the language of agency permits a use of intentional language, almost by sleight of hand. If you call something an ‘agent’ then you can use intentional terms without examining them, without justifying them and indeed without grounding them. Such terms can be used in a purely operational way and then the metaphor of their functionalism can be allowed to slip into a reality... A recent robotics paper gives a computational definition of agent:"Embedded agents are computer systems that sense and act on their environments, monitoring complex dynamic conditions and affecting the environment in goal-directed ways". The definition is full of intentional terms - ‘sense’, ‘act’, ‘monitor’, ‘affect’, ‘goal’. Yet at the same time the definition is purely operational or functional; it says nothing about what it means to have a computer system sense, act, monitor and so on."

Among others, late Quine in Pursuit of Truth came to accept pragmatic use of such "mentalistic" or "intensional" language despite his original physicalism:

"The residual oddity of these mentalistic predicates de dicto is purely semantic: they do not interlock productively with the self-sufficient concepts and causal laws of natural science. Still the mentalistic predicates. for all their vagueness (§27) have long interacted with one another, engendering age-old strategies for predicting and explaining human action. They complement natural science in their incommensurable way... Read Dennett and Davidson."

Of course, one could still wonder when there is more to the use of "agency" than linguistic convenience, and people keep arguing about the borders between life and non-life (are viruses alive?), sentience and non-sentience (are dolphins sentient?), philosophical zombies, etc. But the situation seems to be reminiscent of Hume's analysis of causality. There are no individual facts of the matter about causation, we can not tell from a single event, or even a sequence of them, if X caused Y. We infer causation holistically, if the event falls into a template that our theories use causal laws to predict. It is similar with "agency": there can be no individual "Turing tests" for it (hence the failure of Turing tests for AI). The agency should be assigned holistically on pragmatic grounds, when the use of mentalistic predicates in our theories leads to better predictions overall.