# The Disappearing Agent as an Exclusion Problem **Authors:** Johannes Himmelreich **Published in:** Inquiry 67(6), 2024 **DOI:** [10.1080/0020174X.2021.1935709](https://doi.org/10.1080/0020174X.2021.1935709) *This is the author's manuscript. Please cite the published version.* --- # Human agency and the exclusion principle This paper puts forward a new (or renewed) view on a central problem in the philosophy of action.[^1] I argue that “the problem of human agency” (Mayr 2011) is to a large extent the problem of exclusion. This view on the problem of human agency enables us to look for a new solution to it. I proceed in three steps. I will describe first the problem of human agency, and second the problem of exclusion. Finally, I argue that the problem of human agency is a version of the exclusion problem.[^2] ## I. The problem of human agency Assume you want to reach for a cup that sits in front of you on the table. Some think that science tells us a causal story of what happens when you reach for it. The cup is in the visual field whose image your eyes receive. The information is processed in the somasensory cortex. The appropriate motor units in the adjacent primary motor cortex are activated, motor neurons cause the muscles in your arm to contract and your arm moves. This is a very rough neurophysiological story of what happens when you reach for the cup. It might be called the scientific story of agency.[^3] We can tell a different story. You see the cup of coffee in front of you on the table. You have a desire to have a sip of it. Your belief that there is coffee in the cup and that you could reach for it, and your desire to drink some of it cause you to move your arm. You reach for the cup, and have some coffee (Davidson 1963; Smith 1998). The problem of human agency is how the scientific story of agency relates to the manifest story of agency (cf. Sellars 1962).[^4] The scientific story is taken to suggest that agency is fully accounted for by causal processes in your brain, body, and perhaps your environment. The manifest story is taken to suggest that agents have causal powers and are active in a way that does not reduce to causal processes.[^5] The problem is one of “finding a place for agents in the explanatory order of the world” (Velleman 1992, 465). As Nagel (1986, 111) puts it **“**\[t\]here seems no room for agency in a world of neural impulses, chemical reac­tions, and bone and muscle movements”. This problem is widely recognized in philosophy of action (Alvarez and Hyman 1998, 227–8; Bunnin and Yu 2004, sec. Agent; Clarke 2011).[^6] It gives rise to a dichotomous taxonomy of conceptions of agency. It is assumed that this tension between the scientific story and the manifest story of agency maps to conceptions of agency, which are either event-causalist or agent-causalist.[^7] Roughly, event-causalist conceptions hold that all fundamental causal facts concern events. They are taken to privilege the scientific story. In contrast, agent-causalist conceptions hold that there are fundamental causal facts that concern agents. For example, agents may cause events qua their substance as agents.[^8] They often privilege the manifest story of agency. Ideally, however, we would like to have a conception of agency that instead of privileging one story over the other finds a way to unite them.[^9] This suggests two desiderata on conceptions of agency, which seem to pull in different directions.[^10] Mayr (2011, chap. 1) formulates these two desiderata along the following lines. Conceptions of agency should be such that (i) “actions are instances of activity: the agent is active with regard to what he is doing and not a merely passive sufferer”; (ii) it “fits within the ‘emerging scientific image of the world’”. These theses have some intuitive import. However, both are unsatisfactory. Clause (i) lacks a clear notion of activity (and passivity). It is unclear what exactly an action would need to be like to meet or violate this condition. Clause (ii) remains rather dark with respect to the notions of “fit” and the “emerging scientific image of the world”. The following formulation hopes to make the above ideas more precise without jeopardizing substantive parts. Let agential properties either be mental properties of the agent, such as desires, beliefs, etc., or distinct properties that are modally dependent on mental properties.[^11] Agential properties can be seen as a species of mental properties. ***Desiderata*** Any conception of agency should be such that > **(*Activity*)** for any action *a*, there is an agential property *m* such that (i) it causes the results of *a* and (ii) *m* is not token-identical and not reducible to a physical property; > > **(*Physicalism*)** a version of physicalism is true. This might again not seem satisfying. The *Activity* condition seems too specific in assuming causation by properties. The *Physicalism* condition in contrast is not specific at all, given the varieties of physicalism. For the purposes of this paper this needs to suffice. Lacking a clearer account of activity, a causing by distinct and non-reducible properties seems a sensible first pass. This is because, presumably, activities would be identified by their properties regardless of the specific ontology of an activity. *Activity* is thereby permissive enough to accommodate conceptions of agency that take processes or activities as an ontological category (such as Hornsby 2012).[^12] *Physicalism* lacks specificity, to say the least. There are many versions of physicalism. The desideratum wants to refrain from imposing any specific version as necessary for any conception of agency. Instead, it hopes to capture the original desideratum since it is an intrinsic requirement of physicalism that it fit, in some sense, within the scientific image of the world. For illustration, consider what is known as the standard theory of action due to Davidson (2001).[^13] It is usually understood to be an event-causalist conception of agency. This is because it claims that the activities of an agent are identical to events. Accordingly, this conception meets *Physicalism* but not *Activity*. On the standard theory, actions are caused by primary reasons of the agent. Having a reason is identical with an event within the agent. Furthermore, actions are also events. Hence, on the standard theory, on event (the reason) causes another (the action). *Physicalism* is thereby met. *Activity* is *not* met because the reasons that cause actions are identical to events in the agent. There are no distinct agential properties. This violates clause (ii) of the *Activity* condition. The standard theory interpreted as a token-identity theory thereby does not solve the problem of human agency. It satisfies one desideratum but not the other. The problem of human agency remains. The challenge is to devise a conception of agency that satisfies both.[^14] ## II. The Exclusion Problem The exclusion principle, as understood here, is a restricting condition on causal relations. Roughly put, in a property version, it says that no effect can be sufficiently caused by distinct properties on two different levels. ***Exclusion*** For any effect *e* that has a sufficient cause *c* at some time *t*, there does not exist a *c*\* such that (i) *c*\* is a cause of *e* at *t*, (ii) *c* supervenes *c*\*.[^15] *Exclusion* is a restricted principle of no-overdetermination. Clause (i) rules out overdetermination of *e*. Clause (ii) restricts the principle to apply only to causes *c* and *c*\* that are related by supervenience. The two causes are on two different levels in the following sense. The metaphor of different levels is often construed as supervenience relations. In this sense, *Exclusion* is a no-overdetermination principle restricted to overdetermination across levels. Versions of the Exclusion principle similar to the one above have been developed by Jaegwon Kim and put forward as arguments against non-reductive physicalism (1989b; 1992; 1998; 2008). The challenge is how a non-reductive physicalist conception of the mind can accommodate mental causation.[^16] Consider the following example. You sit on your balcony with your glass of lemonade. A wasp circles around aiming for the straw. You are annoyed by this and wave your hand towards the wasp to make it fly away. A non-reductive physicalist maintains, roughly, that you being annoyed is a mental state *M*. She also concedes that this mental state supervenes on a physical state *P*. Furthermore she holds that *P* causes you to wave your hand. Let your hand waiving be the physical state *P*\*. The non-reductive physicalist faces a dilemma. Either *M* is merely epiphenomenal, or *M* causes *P*\*. In the latter case, *M* is a causing but *Exclusion* is violated. Call this the problem of exclusion. ## III. The problem of human agency is a problem of exclusion Given the formulation of the *Desiderata* above, it seems clear that the problem of human agency gives rise to the problem of exclusion. More precisely, the exclusion problem arises given the *Desiderata* if *Physicalism* includes a causal completeness principle. If *Physicalism* furthermore entails *Exclusion*, the *Desiderata* are inconsistent. I need to introduce the completeness principle. It says, roughly, that all physical effects have an immediate physical cause. More precisely, it is the following. ***Completeness*** For any effect *e* at some time *t*, if *e* instantiates a physical property, then there exists a *c*, such that (i) *c* sufficiently causes *e* at *t,* and (ii) *c* instantiates a physical property. *Argument.* Assume there is an action *a*. It follows from clause (i) of *Activity* that there is an agential property *m*, which causes the results of *a*. Let the property *m* be instantiated by the mental state *M*. It follows from *Physicalism* that *M* is realized by a physical state *P*.[^17] Also by *Physicalism* the results of *a* involve another physical state *P*\*. We have now the situation that *m*, realized by *P*, causes *P*\*. Assume that *Physicalism* entails *Completeness*, which requires that *P*\* has a cause. Let *P* be the sufficient cause of *P*\*. Now we have the situation that *m* and *P* each cause *P*\*. To meet clause (i) of *Activity*, we need to assume that *m* is distinct and not reducible to *P*. This gives us the exclusion problem. If we assume that *Exclusion* is entailed by *Physicalism*, we have a contradiction.[^18] Some parts of the argument should seem surprising others should not. It should not be surprising that the problem of exclusion arises in the problem of human agency. The exclusion problem arises in situations of mental causation for non-reductive physicalist conceptions of the mind. The problem of human agency assumes by *Activity* and *Physicalism* a position similar to non-reductive physicalism,[^19] and human agency involves mental causation. Hence, it should not be a surprise that the problem of exclusion arises in the problem of human agency as defined above. It should be surprising, however, that we were able to derive a possible contradiction from two desiderata. Desiderata, which lead to a contradiction, can hardly be reasonable desiderata. Before discussing this finding, let me summarize the argument so far. I have presented the problem of human agency, as it is conceived in philosophy of action and gave a more precise formulation of this problem as the *Desiderata*. I have briefly presented a version of the *Exclusion* principle and have explained how it poses a problem for non-reductive physicalist conceptions of the mind. Such conceptions conflict with *Exclusion* in situations of mental causation. I have assumed that physicalism entails a principle of causal *Completeness*. I have shown that the *Desiderata* lead to the position in which non-reductive physicalism finds itself: a conflict with *Exclusion*. I conclude that the problem of human agency is the exclusion problem. ## IV. Discussion The above argument may meet resistance in two objections. Both object to the formulation of the problem of human agency. The first objection focuses on the *Physicalism* condition. It argues that *Physicalism* does not entail *Exclusion* and, therefore, the contradiction does not arise. The second objection focuses on the *Activity* condition. It argues that *Activity* shows that I have misconstrued the problem of human activity as causality. Consider the first objection. It rejects the *Exclusion* principle or denies that it follows from *Physicalism*. This makes the *Desiderata* consistent and the problem of exclusion does not arise. The problem of human agency is hence not the problem of exclusion, argues the objection. I am very friendly towards the view that *Exclusion* is false; and I am open to the view that it is not entailed by physicalism. In this respect, I agree with the objection. I tend to think that the desiderata are consistent since either *Exclusion* may be false, or it may not be entailed by physicalism.[^20] My conclusion that the problem of human agency is a problem of exclusion stands despite the objection. My point is not that the *Desiderata* are conflicting. Instead, I argued that the *Desiderata* bring conceptions of agency into a similar situation, in which non-reductive physicalist positions find themselves. Both non-reductive physicalism and the *Desiderata* conflict with *Exclusion*. This was the point that I wanted to push. Whether or not *Exclusion* is true and whether or not it follows from physicalism is a separate matter. Consider the second objection. The second objection argues that *Activity* fails to satisfyingly capture what the desideratum really is. In particular, it objects to framing the notion of activity in causal terms. It rejects talk of agent-causation and demands that we require perhaps something like identification, or some other condition by which an agent makes the actions his own.[^21] I have three points to respond to this objection. First, it is true that I took causalist conceptions of agency as a basic assumption. Indeed, my argument only applies to those. For interesting reasons, they seem to be some sort of industry standard (D’Oro 2012). Whether or not *Activity* misconstrues our manifest and intuitive notion of activity is not an issue for this paper, it remains for the debate between non-causalist and causalist conceptions of agency to decide.[^22] Note however, that if we walk away from causalist conceptions of agency, we are likely to go get to a place, where the exclusion problem is awaiting us already.[^23] If we adopt a non-causalist conception of agency then we are likely to face an *explanatory* problem of exclusion instead of a causal version of the problem of exclusion. Roughly, the explanatory exclusion principle says that there cannot be two distinct explanations for the same explanandum. This seems to be precisely what a non-causalist conception of action does. It aims to give an explanation for an action that is distinct from the scientific explanation. Hence, it seems to the problem of exclusion again. Just in its version of explanatory exclusion as opposed to its version of causal exclusion. Second, I concede that I have not fully captured the problem of human agency. My discussion has left out, for example, the role of reasons and intentionality for action. How reasons relate to human agency can be seen as a further desideratum for conceptions of human agency next to *Activity* and *Physicalism* (Mayr 2011, chap. 1). I focused only on the relation between the latter two to investigate them in separation. The exclusion problem might not be all that there is to the problem of human agency, nevertheless it seems to be a crucial part of it. Finally, bear in mind that my formulation of *Activity* as causation from distinct and non-reducible properties was rather an ad-hoc fix for a vague intuition about what activities are. It seems to me that few satisfying accounts of what activity is have been given.[^24] Lacking a better account of activity, I want to maintain that *Activity* as a causing by a distinct and non-reducible property seems an acceptable construal. It might be more precise, but it is surely not the most enlightening possible proposal on the matter. ## V. Conclusion This paper has argued that the problem of human agency, or at least a crucial part of it, faces a problem of exclusion. I have defended the view that the *Desiderata* give rise to a situation that is similar to that in which non-reductive physicalism finds itself. The truth of *Exclusion* would be bad news for the *Desiderata* just as it might be for such positions. A review of positions in philosophy of action suggests that these positions struggle to meet the *Desiderata* while avoiding *Exclusion* at the same time. Most positions fail to meet both *Desiderata* simultaneously and collapse either into event-causalist or agent-causalist conceptions of agency (see Mayr 2011). The argument here may explain this struggle: If *Exclusion* is supposed to be true, then the *Desiderata* cannot be met simultaneously. By pointing out the analogy between the problem of human agency and the exclusion problem, this paper hopes to open up a new, or renewed, view on an old problem, which might enable us to look for new solutions. The exclusion problem has been widely discussed in the philosophy of mind. If my claim is true that the problem of human agency is the problem of exclusion, then answers that have been given to the exclusion problem in the philosophy of mind might inspire answers to the exclusion problem in philosophy of action. ## References Alvarez, Maria, and John Hyman. 1998. “Agents and Their Actions.” *Philosophy* 73 (2): 219–245. Buckareff, Andrei A. 2012. “How Does Agent-Causal Power Work?” *The Modern Schoolman* 88 (1/2): 105–121. Bunnin, Nicholas, and Jiyuan Yu. 2004. *The Blackwell Dictionary of Western Philosophy*. Malden MA: Blackwell Pub. Campbell, Neil, and Dwayne Moore. 2009. “On Kim’s Exclusion Principle.” *Synthese* 169 (1) (July 1): 75–90. doi:10.1007/s11229-008-9337-3. Chisholm, Roderick Milton. 1976. *Person and Object: A Metaphysical Study*. Open Court Publishing Company. Clarke, Randolph. 2011. “Agent Causation.” In *A Companion to the Philosophy of Action*, edited by Timothy O’Connor and Constantine Sandis. John Wiley & Sons. D’Oro, Giuseppina. 2012. “Reasons and Causes: The Philosophical Battle and The Meta-philosophical War.” *Australasian Journal of Philosophy* 90 (2): 207–221. doi:10.1080/00048402.2011.583930. Davidson, Donald. 1963. “Actions, Reasons, and Causes.” *The Journal of Philosophy* 60 (23). Sixtieth Annual Meeting (November 7): 685–700. doi:10.2307/2023177. ———. 2001. “Agency.” In *Essays on Actions and Events*. Oxford: Oxford University Press. Goldman, Alvin I. 1969. “The Compatibility of Mechanism and Purpose.” *The Philosophical Review* 78 (4) (October 1): 468–482. doi:10.2307/2184199. Horgan, Terry. 2007. “Mental Causation and the Agent-Exclusion Problem.” *Erkenntnis* 67 (2): 183–200. doi:10.1007/s10670-007-9067-9. Hornsby, Jennifer. 1980. *Actions*. Routledge. ———. 2012. “Actions and Activity.” *Philosophical Issues* 22 (1): 233–245. Jackson, Frank, and Philip Pettit. 1990. “Program Explanation: A General Perspective.” *Analysis* 50 (2) (March 1): 107–117. doi:10.2307/3328853. Jacob, Pierre. 2002. “Some Problems for Recductive Physicalism.” *Philosophy and Phenomenological Research* 65 (3) (November): 648–654. doi:10.1111/j.1933-1592.2002.tb00228.x. Kim, Jaegwon. 1989a. “Mechanism, Purpose, and Explanatory Exclusion.” *Philosophical Perspectives* 3 (January 1): 77–108. doi:10.2307/2214264. ———. 1989b. “The Myth of Nonreductive Materialism.” *Proceedings and Addresses of the American Philosophical Association* 63 (3) (November 1): 31–47. doi:10.2307/3130081. ———. 1992. “‘Downward Causation’ in Emergentism and Nonreductive Physicalism.” In *Emergence or Reduction?: Essays on the Prospects of Nonreductive Physicalism*, edited by Ansgar Beckermann, Hans Flohr, and Jaegwon Kim. Walter de Gruyter. ———. 1998. *Mind in a Physical World: An Essay on the Mind-body Problem and Mental Causation*. MIT Press. ———. 2008. *Physicalism, Or Something Near Enough*. Princeton University Press. List, Christian, and Peter Menzies. 2009. “Non-Reductive Physicalism and the Limits of the Exclusion Principle.” *The Journal of Philosophy* 106 (9): 475–502. Lowe, E.J. 2001. “Event Causation and Agent Causation.” *Grazer Philosophische Studien* 61 (1): 1–20. Malcolm, Norman. 1968. “The Conceivability of Mechanism.” *The Philosophical Review* 77 (1) (January 1): 45–72. doi:10.2307/2183182. Mayr, Erasmus. 2011. *Understanding Human Agency*. Oxford University Press. Nagel, Thomas. 1986. *The View from Nowhere*. Oxford University Press. Sellars, Wilfrid. 1962. “Philosophy and the Scientific Image of Man.” In *Frontiers of Science and Philosophy*, edited by Robert G. Colodny. University of Pittsburgh Press. Shapiro, Lawrence A. 2012. “Mental Manipulations and the Problem of Causal Exclusion.” *Australasian Journal of Philosophy* 90 (3): 507–524. doi:10.1080/00048402.2011.601749. Smith, Michael. 1998. “The Possibility of Philosophy of Action.” In *Human Action, Deliberation, and Causation*, edited by J. a M. Bransen and Stefaan E. Cuypers. Kluwer Academic Publishers. Stoljar, Daniel. 2010. *Physicalism*. Routledge. Taylor, Richard. 1966. *Action and Purpose*. Velleman, J. David. 1992. “What Happens When Someone Acts?” *Mind* 101 (403) (July 1): 461–481. doi:10.2307/2253898. Yates, David. 2012. “Functionalism and the Metaphysics of Causal Exclusion.” *Philosophers Imprint* 12 (13) (July). [^1]: “New” is perhaps too bold, “renewed” seems more appropriate. Philosophy of action has seen a discussion on exclusion of causal explanations of action (cf. Malcolm 1968; Goldman 1969; Kim 1989a). Reviving and moving the discussion from causal *explanation* to causation might be news. At any rate, the recent debate on causal exclusion in philosophy of mind calls to revisit the issue in philosophy of action (cf. Horgan 2007; Campbell and Moore 2009; List and Menzies 2009; Yates 2012; Shapiro 2012). [^2]: Some restricting qualifications apply. First, for the purposes here, I am concerned only with a part of the problem of human agency (see footnote 4). Second, this project presupposes causalist conceptions of agency. Hence, I will only focus on causal exclusion. For non-causal theories of action, my thesis naturally extends to explanatory exclusion. Finally, I assume that agential states and mental states of agents are multiply realizable. [^3]: Strictly speaking, agency is only a manifest and not a scientific notion (cf. Sellars 1962). The “scientific story of agency” should be understood as the neurophysiological correlates to the manifest notion of agency. This, of course, assumes that such a reduction from manifest notions to neurophysiological accounts exists. [^4]: The problem can be seen as a tension between three theses. It is the question of how a conception of agency can meet three desiderata simultaneously (cf. Mayr 2011, chap. 1). I focus only on two of them and leave the third one aside. [^5]: Again, for convenience this formulation restricts us to causal theories of action. The exclusion problem arises also for non-causal theories of action. [^6]: For example, consider this passage from the entry on “Agent” in The Blackwell Dictionary of Western Philosophy “The change caused by an agent is called agent-causation, in contrast to event-causation in which one thing is caused externally by another.” (Bunnin and Yu 2004) [^7]: Some people hold a view, similar to Davidson (2001), that agent-causation reduces to event-causation (see Lowe 2001). Such a position might be taken to violate the *activity* condition below. However, it also recognizes that this dichotomy is usually drawn. [^8]: Classic agent-causalist accounts are Taylor (1966) and Chisholm (1976). Alvarez and Hyman (1998) and Mayr (2011) are contemporary examples. [^9]: Younger accounts of agent-causation explicitly seek to reconcile agent-causation with event-causation while upholding the view that agent-causation is different in kind. [^10]: Roughly speaking, a conception of agency is a set of positions on the metaphysics of actions including causation, how they relate to events and reasons, and on what makes them intentional. [^11]: The specific dependence can be left open. Supervenience would be one example for a modal dependence. I assume that mental properties are not physical properties. This assumption can be weakened such that a mental property is both a physical and a non-physical property. The conclusion can be maintained by strengthening clause (ii) of *Completeness* below to “*c* instantiates *only* physical properties”. [^12]: Furthermore, nothing hangs on the property formulation of *Activity*. In a different formulation the problem of human agency is simply a different formulation of the exclusion problem. [^13]: Here I give Davidson the reading of a token-identity theory. Before the backdrop of anomalous monism it can also be interpreted as a position of non-reductive physicalism (cf. Jacob 2002). [^14]: Note that under the formulation above, the problem of human agency seems to be the problem of mental causation. Generally, the problem of human agency is considered broader than the problem of mental causation. The coincidence here is because I focus only on a part of the full problem of human agency (see footnote 4). [^15]: Indexing for time is required. Otherwise by transitivity of causation exclusion is false (cf. Stoljar 2010, chap. 11). [^16]: Note that *Exclusion* is weaker than causal closure principles also known as ‘closure of the physical realm’. They demand that, roughly, any effect has *only* physical causes. In what seems to be an analogous way, Alvarez and Hyman (1998) distinguish between a strong and weak principle of causation. Note that the *Completeness* condition below is weaker than such closure conditions. [^17]: “Realized” indicates a functionalist view of the mental. I am committed to this view only in so far as assuming an identity theory of the mental would immediately lead to a violation of *Activity*. [^18]: Note that a similar result is obtained in Buckareff (2012), who holds an exclusion principle against agent-causalist conceptions of agency in general. [^19]: It seems also compatible with some versions of dualism. I leave this aside. [^20]: Exclusion principles seem to be false for some conceptions of causation (Jackson and Pettit 1990; List and Menzies 2009). [^21]: For a position similar to this effect see Velleman (1992). A weaker version of this objection rejects merely property causation in favor of substance causation. It seems that would not avoid the problem of exclusion (Buckareff 2012). [^22]: Furthermore, my thesis extends naturally to non-causalist conceptions of agency. While causalist conceptions of agency face a problem of *causal* exclusion, non-causal conceptions of agency face a problem of *explanatory* exclusion. [^23]: Setting aside a dual-explananda strategy, which argues, roughly, that the manifest story and the scientific story are stories about two fundamentally different things. [^24]: The early position of activity as trying (Hornsby 1980) seems to have given way to criticism. Amendments to this view defend that activities are an ontological category sui generis but do not give a functional definition (Hornsby 2012).