In her book Self-Constitution: Agency, Identity, and Integrity, Christine Korsgaard presents a Neo-Kantian approach to agency and morality. She combines elements from Kantian and Aristotelian thought to develop a theory of agency which characterizes action as something that is normative in constitution as well as in realization. She proposes that there are constitutive standards for any activity and that for all agency the constitutive standard is self-constitution, perseverance of the self, so to speak. This applies to animals and humans in the same way. The goal of this article is to examine how exactly Korsgaard aims to include animals in the realm of agents, despite working within a Kantian framework.
According to Korsgaard, rational agency, agency which is guided by principles chosen by the agent, differs in the fact that the purpose of self-constitution is accessible to a rational being, the animal acts on the same purpose but need not be aware of it. Therefore, the function of rational agency, self-constitution, is to choose the principles by which one’s action are guided and in doing so unifying oneself as an agent, as the author of these actions. These principles are Kant’s hypothetical and categorical imperatives and they are constitutive for action. The normativity in the principles of action links them directly to morality. Acting well means choosing those actions that constitute oneself well, which in turn means to become a functional rational agent, which in turn means to become a good person, according to Korsgaard.
To assess how convincing Korsgaard’s inclusion of animals in the realm of agents is, I will first give a summary of Korsgaard’s account of animal agency coupled with some remarks on rational agency in Kant’s framework. In a second step, I will discuss strengths and weaknesses of her argument and will introduce a hopefully insightful parallel to Tyler Burge’s account of agency.
I. Korsgaard’s notion of action
As Korsgaard, a neo-Kantian, notes, Kant’s view has difficulties accommodating animal action. This is because it sets the Kantian imperatives as a necessary condition for agency. The application of and deliberation about the imperatives in turn requires rationality, which Kant does not ascribe to non-human animals. The notions of autonomy and efficacy, central to the Kantian notion of agency, are closely related to action and the imperatives. For Kant, an entity must, at least try, to render itself efficacious and autonomous for it to be an agent. While Korsgaard does want to operate in a Kantian framework the exclusion of animals from the set of agents does not sit well with her. Her thesis is that efficacy and autonomy can be constitutive standards for action, even if animals do not knowingly try to conform to them. (Korsgaard, 92)
Before discussing this thesis, I will now provide Korsgaard’s notion of (animal) action. The first condition she mentions is that action is intelligent movement. This means that the animal’s movement is in some way responsive to a representation or conception it has of the world. Intelligence in this sense is given in cases where an animal moves towards its prey perceiving it as a source of food (a bat that chases an insect for example). A further specification is required here. Action or intelligent movement is to effect a change in the world by effecting a change in oneself. To effect change in oneself like this is prime in the sense that there are no means to it. If I raise my arm, I just raise my arm. In every action I’m disposing myself somehow. This corresponds to the idea of self-determination as a necessary condition for action. (Korsgaard, 94-95)
The second necessary condition for agency is already implied. Intelligence in this sense entails some form of perception or representation, or, more specifically, intentional content. This could be a goal, for example. Herein, lies the difference between animals and plants or artefacts. While all three have some teleological organization, only animal’s movements are intentional, in the sense, that the animal is guided by a representation or conception that it forms of the world.
While human agents can weigh and choose different reasons for action, animals are more restricted in that respect. Reason is the dividing factor. Intelligence Korsgaard describes as a capacity to forge new connections through learning. This applies to animals. For example, an animal can choose between different hiding spots. What the animal has no control over, however, is what makes these options choice worthy. It can choose means but not ends. In Kantian moral psychology, every action starts with an incentive. Incentives can be understood as motivationally loaded representations. Correspondingly, reason is the capacity to choose between different grounds for beliefs or actions and therefore different responses to an incentive. According to Korsgaard, even in the face of grave danger, a human animal can decide to resist the temptation to hide, fight or run and not take these incentives as a reason for action, while non-human animals cannot. (Korsgaard, 104, 116) For animals to act, on a Kantian view, incentives would have to work causally and indirectly on them because, if incentives directly caused animals to act, it would then not be the animal as a whole that acts or causes the action but only its perceptions or representations, rendering it a mere puppet of incentives that behaves accordingly. Correspondingly, every action on this account consists of an incentive and a principle. For humans the principles are those of practical reason, the Kantian imperatives, for a non-human animal they are its instincts. (Korsgaard, 106) Instincts are established connections between representations (incentives) and a primitively normative response to those representations. An example is that for birds of prey the representation of a smaller rodent as prey is connected to the movements necessary for chasing it. (Korsgaard, 110-111) Navigating the world with such loaded representations structured by instinct means that the animal perceives certain things as positive or negative, as for eating or for running away from. This is what Korsgaard calls the primitively normative response. The important thing is that the animal need not know about this normativity, it is built right into its perception of the world, but that it is guided by it and that her movements are evaluated against these standards. (Korsgaard, 113)
With these points in mind, we can now piece together what animal action in a Kantian sense looks like. There are at least two forms or degrees of autonomy or self-determination. For animals, autonomy is to be governed by the principles of their own causality given by their teleological form. These principles are definitive of their will. ‘Will’ translates to ‘instincts’ in the context of non-human animals. For humans on the other hand, it means the ability to choose the principles that are definitive to one’s will. (Korsgaard, 108) Now it also becomes clear in what way non-human animals can be efficacious. Since the very idea of action, also for non-human animals, is normative in the sense that it is associated with success or failure regarding certain constitutive standards and animals are in a primitive way autonomous, their movements or actions are subject to these standards, therefore they can be more or less efficacious. (Korsgaard, 109-110) In this sense animals can be autonomous and efficacious, making them agents.
II. Can animals act?
Korsgaard’s argument for animal agency in a Kantian framework is convincing. She must however make concessions, which bend the boundaries of the Kantian framework to the point where I am not convinced that many Kantians would agree. Most notably her move to discard the dichotomy between rational agency and mere non-rational behaviour in favour of a gradual distinction is what separates her account from the Kantian original. She does so, for example, by characterizing instinct in non-human animals as analogous to the will in human animals. At the same time, this is the step that makes her view convincing, for the graduality of capacities like perception, and with it self-consciousness in nature is commonplace in current psychology. Phenomena like scaffolding and the observations of primates and many other animals strongly suggest that the transition from the pre-perceptual to the perceptual and from the non-rational to the rational domain is a matter of degree and comes down to steps on a continuum rather than hard switches between different categories. In this context, it seems only natural to discard the hard lines in the Kantian framework for in unison with Korsgaard I say that there are no hard lines in nature between action and non-action since there are no hard lines between perception and mere sensory registration. (Korsgaard, 97)
A difficulty I find with Korsgaard’s notion of agency is that I think working out the details of what role perception plays in action and opening up a continuum of agency leads to finding further soft lines between action and non-action which suggest extending the class of agents even further than Korsgaard does. So much so, that the application of a Kantian notion of action would not suffice anymore. What Korsgaard does not have the time to discuss in detail, I argue, Tyler Burge provides in his article on primitive agency. Where Korsgaard merely refers to a continuum of agency and the workings of instincts and the kind of primitive normativity that follows, Burge elaborates on how exactly mere behaviour may be discriminated from agency and which standards and norms apply to non-human animal agency. While, sadly, I cannot discuss his position in detail here, it suffices to note that, while the two accounts overlap where movement in animals is guided by perception, Burge’s notion of action extends even further on the spectrum than Korsgaard’s. Finding Burge’s position convincing, I suggest using the label of action even for some instances of pre-perceptual movement. The term Burge coins here is ‘primitive agency’. Because Korsgaard does not discuss perception in detail, it is hard to tell which animals and which intelligent movements she would consider as actions. This would require a study of examples, including fringe cases. All things considered, Korsgaard’s account is convincing because it matches findings in modern psychology and provides a Kantian argument for animal rights. She manages to include animals in the realms of agents even in a Kantian framework and the weakness I found with her approach is more a matter of labelling than a theoretical problem for her position since she takes the most important step which is opening a continuum of agency.
 For general information about the debate on agency, you’ll find an extensive article on agency here: https://plato.stanford.edu/entries/agency/#AgeDisHumAct
 For more information on Kant’s imperatives see: https://plato.stanford.edu/entries/kant-moral/#CatHypImp
 This corresponds with Lucy o’ Brien’s idea that actions are prime. See: O'Brien, Lucy (2017). Actions as Prime. Royal Institute of Philosophy Supplement, 80, 265-285.
 See: Burge, Tyler, (2014). Perception: Where Mind Begins. Philosophy, 89(03), 385–403.
Korsgaard, Christine M. (2009): Self-Constitution: Agency, Identity, and Integrity. Oxford: Oxford University Press.
Burge, Tyler (2009). Primitive Agency and Natural Norms. Philosophy and Phenomenological Research 79 (2):251-278.