A Gradualist Metaphysics of
Agency
Jesús H. Aguilar and Andrei Buckareff
Suppose that you just ate a meal. Having satisfied your hunger, you find
yourself craving a beer and realize that there is no beer left in the fridge.
You know that the corner store has a good selection of beers, so you walk
to the store to get some. At the same time that you finish your meal,
your cat finishes eating. Being in the habit of cleaning himself after
eating, he looks for a comfortable place and gets to work cleaning his
face. Both you and your cat are exercising agency. In your case, you do
so when you take a walk to the store to get some beer. In the case of your
cat, he exercises agency when he looks for a place to groom and sets to
grooming himself. It is clear that even if what you two are doing can be
done in automatic mode there is a difference between the walking and
the grooming on the one hand, and the process of digesting your respec-
tive meals on the other. And yet in all these cases you and your cat are
always doing something. In this respect, the walking, the grooming, and
the digesting, are cases of agency. In the case of walking and grooming,
you and your cat exhibit intentional agency; and in the case of digesting
you each exhibit a very modest species of (quasi-) agency. This generous
understanding of agency is controversial, to say the least. Some action
theorists will reject the idea that non-human animals like cats can be
true agents while some other action theorists will take the further step
of insisting that in cases like digestion there is simply no agency at all
since no one is really doing anything.
In this essay we reject such an exceptionalist view of agency and
argue in favor of a gradualist understanding of agency that accepts that
non-human animals like cats are agents, and goal-oriented processes
like digestion involve some species of agency. We endorse a bottom-up
approach to understanding agency and argue that, contrary to a widely
accepted view about agency, being an agent is not an all-or-nothing
30
A Gradualist Metaphysics of Agency 31
1 Quasi-agents
they are missing? In other words, what keeps us from treating chemical
compounds and similar substances as more robust agents? As we have
seen, it cannot simply be their lack of complexity or some kind of inten-
tionality, for they possess both.
2 Minimal agents
3 Rational agents
The sorts of agents with which we are typically concerned in the philos-
ophy of action are rational agents like normal adult humans. However,
36 Jesús H. Aguilar and Andrei Buckareff
does Lydia. And yet, if you are Lydia’s partner, while you may be upset
with Lydia and blame her for breaking your Matryoshka doll, it will be
inappropriate to have the same emotional response with Fred.8 But,
again, what exactly justifies this different response? The answer brings
us back to rational agency. The difference between these two agents is
the sort of internal life that allows Lydia but not Fred to be responsive
to certain types of normative reasons. For instance, Lydia can appreciate
that the standard rules of etiquette would proscribe knocking items off
shelves. She would also understand the value of the Matryoshka doll
and realize how inappropriate it is to destroy it as a way to get her
partner’s attention about her desire to eat. And so on. But clearly Fred
has no such understanding of the impropriety of his behavior or of
the value of an object like a Matryoshka doll. And Fred lacks a general
capacity that would enable him to be aware and assess all the different
normative reasons involved in his action. In this sense, Fred is essen-
tially non-responsive to a set of crucial normative reasons that iden-
tify true rational agents and justify appropriate responses from other
agents. And yet, despite this striking difference between a non-rational
agent like Fred and a rational agent like Lydia, the difference between
them as agents in general is one of complexity and not a difference
of kind. They both are part of an agential continuum, in which some
agents are less complex than others, and where the more sophisticated
ones attain their sophistication by being capable of acting for a wider
range of reasons.
The foregoing may have convinced some readers about the reality of
gradual agency. But there are others who will not be taken in by the story
we have been telling thus far. Such naysayers will insist that agency and
being an agent are not best understood in gradualist terms, with agency
and being an agent admitting of degrees. Such non-gradualists will
insist that being an agent and exercising intentional agency is an all-or-
nothing matter. One reason they may offer is that language is a funda-
mental capacity necessary for true agency; but language itself is not a
phenomenon that can be correctly understood as coming in degrees.
Thus, they would argue true agency in turn cannot be a phenomenon
that comes in degrees.
An influential account of language and agency along these excep-
tionalist lines is ascribable to Donald Davidson.9 Agents, Davidson
argues, must be language users since ‘in order to be a thinking,
38 Jesús H. Aguilar and Andrei Buckareff
like the one we defend here would be seen as misguided. For an excep-
tionalist like Davidson, agency is a whole package that cannot be teased
apart in the metaphysically reductive manner underlying the gradualist
approach that we favor.
If Davidson and those who share similar views are right, then many of
the creatures who would count as agents on our account are not, in fact,
agents at all. This is so because they cannot be truthfully described as
thinkers. And they are not thinkers because they are not language-users.
However, contra Davidson, we can say that, while having the capacity
for speech and knowledge of a language can make action-explanation
easier and allow for very complex actions to be performed, language is
not necessary for every thought, and, a fortiori, is not necessary for the
type of agency that uses this type of non-linguistic thought. For some-
thing can be an agent and exercise the capacity for thought even if it
lacks the capacity for language.
As we noted before, we do not think of the correct application of
the concepts of ‘agent’ or ‘agency’ as an all or nothing matter. That is,
we think it is better to understand agency and counting as an agent
as coming in degrees.11 Returning to the example of Fred the cat: if
Davidson is right, then Fred should not be described as an agent and
he exercises no agency when he attacks a mouse. That is to say, Fred
does not intentionally attack the mouse. What he does cannot be accu-
rately described as an ‘intentional action’ and Fred cannot be truthfully
referred to as an ‘agent’. What about the complexity of Fred’s behavior?
Isn’t the complexity exhibited by Fred in how he behaves sufficient
to truthfully describe his behavior as intentional and, hence, for us
to truthfully say of Fred that he is an agent who exercises intentional
agency when he attacks a mouse and, hence, has thoughts? Davidson
has a response:
A creature may react with the world in complex ways without enter-
taining any propositions. It may discriminate among colors, tastes,
sounds and shapes. It may ’learn’, that is, change its behavior in ways
that preserve its life or increase its food intake. It may ’generalize’, in
the sense of reacting to new stimuli as it has come to react to similar
stimuli. Yet none of this, no matter how successful by my standards,
shows that the creature commands the subjective-objective contrast,
as required by belief. (1982a: 326)
40 Jesús H. Aguilar and Andrei Buckareff
So, while Fred the cat may exhibit complex behavior that suggests the
presence of thought, including beliefs, strictly speaking Fred is not a
thinking organism and, hence, not a true agent.
One obvious reply to those who endorse the sort of reasoning
defended by Davidson is to point out that the concept of belief is not the
same thing as having a belief (Heil 2012: 256). To take an analogy: for
centuries, humans did not have the concept of pathogenic bacteria. But
they were no less vulnerable to such bacteria. A human infected with
cholera or syphilis did not fail to have either infectious disease just on
account of not possessing the concept of a bacterium (or the concept of
cholera or syphilis, for that matter). This reply may satisfy some persons,
but for many it will seem that there is an important difference between
thoughts and bacteria. For one, thoughts can be reflected on in the first-
person and interpersonally. We can think about our own thoughts and
the thoughts of others. This is part of what Davidson seems to have in
mind in making his case for the interdependence of the capacity for
thought on the capacity for language. In response, it might be argued,
you can think about the probiotic bacteria in your body and think about
the probiotic bacteria in the body of a friend. But notice that a bacte-
rium, whether in your body or the body of a friend, is quite different
from a first-person representation of your mental state. A first-order
representation can be represented in higher-order thought. You can
believe that you believe things and believe that your first-order belief is
false upon reflection. The Davidsonian will insist that the representative
content of your higher-order belief in such a case requires that you have
the capacity for language.
There is another response to Davidson that is a bit more promising
(see Heil 2012: 265–73). Some thought requires the capacity to use
language (for instance, Davidson’s example of a higher-order belief
about a false first-order belief, articulating abstract ideas in your head,
and thinking about a grocery list you wrote out but forgot at home).
But not all thoughts are like this. Of course, all thinking is representa-
tional. But insofar as it is representational, it is best to think of thinking
as ‘a matter of using imagery’ (Heil 2012: 266).12 And not all conscious
thought involves the use of verbal imagery. Following John Heil, we will
take all imagery to be pictorial (2012: 267). So, some imagery involves
verbal imagery, but not all does. For instance, we may think about lines
from a poem or a remark someone made, but we can also think about a
meal we are preparing: how it smells, and how we expect it will taste. In
the case of Fred the cat, even if he is incapable of using verbal imagery
he can make use of non-verbal imagery in his thinking. What matters
A Gradualist Metaphysics of Agency 41
is that he can use imagery in the way appropriate for him to count as
thinking, even if the imagery used is nonlinguistic.13
While not all thought is linguistic, involving verbal imagery, at least
some nonlinguistic thinking resembles linguistic thought. Davidson
seems to recognize this, but he resists labeling such activity ‘thought’.
Philosophers and cognitive scientists have suggested different ways in
which the thought of non-language users can exhibit features similar to
linguistic thought. For instance, C.B. Martin contends that, ‘nonlinguistic
activity at its most sophisticated and structured levels has a remarkable
pattern of parallels to linguistic activity’ (2007: 93). Martin refers to the
richer variants of such nonlinguistic activity, whether overt or covert,
as ‘protolanguage’. Protolanguage is not itself a kind of language. But it
has features in common with language. According to Martin ‘protolan-
guage is a structured, rule-governed network of semantic, procedural
activity prior and basic to linguistic activity, having an almost totally
unnoticed and surprising pattern of parallels to language itself’ (Martin
2007: 94). If you will, protolinguistic activity is a ‘procedural analogue’
to linguistic activity (Heil 2012: 270). This relates to the previous point
about imagery. There are protolinguistic procedures that Fred the cat
may employ involving the use of non-verbal imagery that parallel the
way in which a language user may use verbal imagery. For instance,
we can identify a protolinguistic procedure whereby Fred can discover
some truth about human behavior using a nonlinguistic analog to induc-
tive generalization. Fred has seen that when he is about to be fed one
of his human companions grabs a can and takes the lid off and scoops
food into a bowl. After a while, once he witnesses this occur enough, he
adjusts his behavior and simply waits patiently in the location where he
is fed once he notices this pattern of behavior in his human companions.
Fred makes connections between behavior types – even if the behaviors
are not identified under a particular linguistic description – and infers
that food is about to be served. In this case, there is a procedure aimed
at an outcome. The procedure and outcome could be expressed linguis-
tically via a simple inductive inference and the outcome would be the
conclusion (Heil 2012: 270–1).
If we are right, then it seems reasonable to hold that Fred the cat is an
agent. Fred is an agent in a way that a compass, a chemical compound,
and a tree are not. Fred engages in purposeful actions in response to how
the world is represented to him in his thoughts. Fred’s agency is not as
robust as what we find in normal human agents. While his agency is
more robust than the agency of a spider, owing in part to the richness of
Fred’s mental life vis-à-vis that of the spider, it is not as complex as what
42 Jesús H. Aguilar and Andrei Buckareff
we find in other agents, such as normal adult humans. Thus, we take the
degree to which something counts as an agent and the sort of agency
they can exercise to be a function of the complexity of their mental
lives. In this respect, Fred’s mental life is surely complex enough for him
to legitimately count as an agent.
6 Conclusion
Notes
The order of names should not be taken to imply priority of authorship.
1. Many readers will likely be familiar with the recent work of Fred Dretske
(1999), Don Ross (2012), and Helen Steward (2012) on agency. Those readers
will no doubt recognize that, while Dretske, Ross, Steward, and the authors of
this essay have significant differences between one another, we stand together
in favoring a bottom-up approach to thinking about the metaphysics of agents
and agency.
A Gradualist Metaphysics of Agency 43