Gabriel Aston
University of Cambridge
Abstract
This paper explores the reductionist project to reduce intentions to sets of beliefs and desires and a possible motivation for such a project. I outline an extant critique from Holton of the reductionist project before presenting my own novel critique which rests upon the principle of halfway proportionality and the notion of a strictly stronger phenomenon as expounded in the works of Vaassen and Yablo. In the course of this paper, I demonstrate the following. Granting that intentions can be reduced to sets of beliefs and desires, since sets of beliefs and desires are strictly stronger phenomena than and are screened off by intentions, intention is a valid cause of, and thus explanation of, action. Furthermore, I maintain that intentions have greater explanatory value than beliefs and desires, and thus that the reductionist project is shown to be undermotivated. As such, I conclude that the reductionist project merely obfuscates our understanding and that further work that takes intention as a basic mental kind alongside belief and desire is required, as explored in papers by Bratman and Holton.
I. Introduction
Within the notion of intention is a duality that aptly leads Bratman to describe intention as “Janus-faced” (Bratman 378). We describe actions as “intentional,” e.g., “I intentionally left out my copy of _Middlemarch_, so that I would remember it.” But we also think of intention as a state of mind, e.g., “I intend to get around to reading _Middlemarch_ someday.” In this paper, I will show how this duality may inspire attempts to reduce intentions to sets of beliefs and desires, focusing on one notable attempt at such a reduction in the work of Davidson (“How is Weakness of the Will Possible?”; “Intending”). I will then proceed to argue against such a reduction. I will outline how Holton’s work on resolutions can be taken as an argument for the incompleteness of the reductionist project before showing that Holton’s argument leaves too much space for the reductionist to adapt. As such, I will provide a novel argument against the reduction of intention to beliefs and desires grounded in the works of Vaassen and Yablo on causality and the proportionality of explanations.¹ This argument will subvert attempts at the reduction of intention to beliefs and desires by demonstrating the distinct explanatory role of intentions.
II. The Reduction of Intention to Beliefs and Desires
To begin, I will motivate and outline a reductionist account. As already made clear, the concept of intentions, even in a pre-theoretical context, is of an apparent complexity. As Bratman notes, we use inten-tions to characterize actions and mental states alike, and how these distinct elements are unified under the notion of intention is not immediately apparent. We may thus be motivated to form a reductionist account of intention in order to transform the complex notion of intention into more familiar terms.²
One well known example of such a reductionist account is that of Davidson, which aims to reduce intentions to sets of beliefs and desires. Davidson’s account distinguishes two types of judgments, _prima facie_ judgment and all-out judgment, and identifies intentions with the latter. A _prima facie_ judgment is a judgment that something is good (or bad) in so far as it has some particular characteristic. I may judge _prima facie_ that reading _Middlemarch_ is good in so far as it would make me seem well read to others but bad in so far as it may be dull. In this instance, my negative _prima facie_ judgment may be enough to outweigh the other positive one, and lead to my not reading _Middlemarch_. When we act, we demonstrate a further judgment that some elements of the action that we _prima facie_ judged as good were enough to outweigh those elements we _prima facie_ judged as bad, such that we were moved to perform the act. This secondary sort of judgment is an all-out judgment and is merely a product of those BDs that form our _prima facie_ judgments. To possess an intention to _Φ_ is merely to form an “all-out” judgment that doing _Φ_ is desirable and to act intentionally is merely to act under the influence of such an all-out judgment. The unity of the guises of intention is then explained just in terms of the relations between those BDs that constitute an intention, and those that constitute the motivation behind an intentional action.
III. The Reductionist Project as Incomplete
One strategy to challenge the reductionist project is to argue against its completeness. I will now outline one such argument from Holton in order to demonstrate that such lines of argument commonly leave the reductionist with too many responses to be conclusive. Holton identifies resolutions as a demanding subset of intentions. These intentions are formed with the aim of assisting us in resisting later temptations and maintaining an intended course of action. Holton then argues that the reductionist is unable to provide a model of BDs that accurately captures the temptation-blocking function of resolution. Thus, Holton can be taken as giving an argument that the reductionist project is incomplete.
Holton’s argument is essentially the following: 1) A BD model must characterize our resolute action as the result of the pull by some desire to act in the resolute manner. 2) The experiences of acting because we are pulled by our desires, and acting because of a resolution, are significantly different. 3) We thus have reason to doubt the accuracy of the reduction from one into the other. Holton asks us to imagine that we prefer:
“A. I give up smoking for good soon.
to
B. I don’t give up smoking for good soon.” (Holton 43)
We are then also asked to imagine a circumstance in which we recognize that for each single cigarette we smoke, the harm it does to us would be outweighed by the pleasure we take from it. Thus, we also prefer:
“C. I don’t resist this cigarette.
to
D. I resist this cigarette.” (Holton 43)
If we just followed our strongest desires, we would always put off quitting smoking until after our next cigarette, with no inconsistency between smoking this specific cigarette and quitting soon. We can imagine the smoker escaping the cycle by forming a resolution to stop smoking from time t, and thus being made to give up the pursuit of desire C after t to avoid irresolution. To replicate this function purely in terms of BDs, we must show how the addition of further desires and beliefs could make the pursuit of desires A and C inconsistent.
Holton outlines how this could be done. Firstly, we could posit further desires on the part of the smoker. The smoker, due to his desire A, may utter “I will quit smoking from tomorrow.” Given the smoker has a desire not to contradict his assertions, he will be driven to quit smoking after tomorrow. Secondly, we could posit further beliefs on the part of the smoker that would cause him to view the fulfillment of desire A as inconsistent with the fulfillment of desire C, e.g., a belief that only if he can resist this specific cigarette in front of him will he ever be able to quit.
Holton casts doubt on the successes of each of these strategies and further notes that regardless of which of these options the reductionist supports, the smoker’s quitting is ultimately a product of one of his desires triumphing over his desire to smoke, whether it be the extra desire posited or the desire to quit sometime soon. Holton argues that in our actual experiences of being resolute, sticking to our resolutions is distinctly effortful in a manner that merely doing as we most strongly desire is not. When we are led by our desires, there is not the same mental struggle as in maintaining a resolution. Thus, we have reason to doubt the accuracy of the reduction of resolutions to BDs, and in consequence reason to doubt the completion of the reductionist project.
IV. The Reductionist Response
I am sympathetic toward Holton’s argument; however, it appears to leave the reductionist with too many counter arguments to be damning. I will outline one such response open to the reductionist, arguing that the felt experience of resolution is also insufficiently explained on Holton’s non-reductionist account.
The reductionist may argue that even if we admit intention as distinct from BDs, we still cannot fully explain the distinct feeling of effort that is involved in maintaining our resolutions. Holton posits that alongside intention as a distinct mental kind, we also possess a distinct faculty of “will-power” that must be exercised in order for us to maintain our resolutions (39). The exercise of this faculty is held to explain the feeling of effort in maintaining a resolution. However, as Holton notes, we can decide not to attempt to be resolute and choose not to undertake the effort of exercising the will-power. In fact, Holton suggests that in the majority of cases, whether we are resolute or not is a matter of our willingness to use up our will-power. Holton thus maintains that in order to be resolute, we must have sufficient motivation to exercise the will. The reductionist may question whether this motivation is not merely a desire to be resolute. If this is a desire, then by Holton’s own description, determining the extent to which we exercise the will to maintain a resolution should have an automatic pulling feel. The reductionist could then complain, similarly to Holton, that this does not reflect the felt experience of maintaining a resolution; deciding the extent to which we are willing to employ the will is itself imbued with that same feeling of effort as the use of the will. Thus, it seems that the non-reductionist account is not complete. Some other faculty, e.g., a super-will, may need to be posited, but this seems to merely move the problem up a level. Thus, the reductionist can argue that if the felt quality of resolutions is a problem for their account, it is also a problem for the non-reductionist account.
V. Proportionality of Explanations and the Explanatory Power of Intentions
Novel argumentation that undermines the reductionist on the grounds of their own assumptions, and thus leaves the reductionist with fewer responses, is therefore warranted. I will argue, drawing on the work of Yablo and Vaassen, that even if we accept that intention can be reduced to BDs, intention remains not only a valid but in fact a superior explanation of behavior than the BDs that it is reducible to. As such, I will conclude that the reduction of intention to terms of BDs in fact is a reduction to terms of inferior explanatory power, and as such weakens our understanding of mental faculties.
It is firstly worth noting that both Yablo and Vaassen frame their discussions in terms of causes. For the purposes of this paper, I will reframe their arguments in terms of explanations. The arguments they present translate without difficulty. If we show that X is a cause of Y, we, qua cause, show that X is an explanation of Y. Given that the focus in this paper has been on the explanatory motivation for reduction, operating in the language of explanation will enable us to challenge the reductionist account.
The argument that intentions are a permissible explanation of behavior, even granting reduction, will go as follows:
(1. The principle of halfway proportionality) P explains E iff for any P+ strictly stronger than P, P screens off P+ from E.
(2. The reductionist conclusion) Intentions supervene on BDs.
(3. Premise) Intentions can be realized by multiple distinct BDs.
(4. By the definition of strictly stronger phenomena, 2, and 3) BDs are strictly stronger than intentions.
(5. Premise) Intentions screen off BDs from actions.
(6. Conclusion—by 1, 4, and 5) Intentions explain actions.
Firstly, a thorough definition of the strictly stronger relation must be provided. Vaassen identifies that a phenomenon P+ is strictly stronger than another phenomenon P if the possible conditions of the world required for there to be P+ are a proper subset of those required for there to be P. Thus, an object being scarlet is a strictly stronger phenomenon than that same object being red. The “strictly stronger than” relation is more thoroughly developed by Yablo. P+ is a strictly stronger condition than P iff for some X, to be of kind P+ is for it to be of kind P, “not simpliciter, but in a specific way” (Yablo 252). In other words, if P+ is strictly stronger than P, then there is an asymmetric necessity such that, necessarily for all X, if X is of kind P+, then X is of kind P, and such that it is possible for some X to be of kind P and not be of kind P+. As Vaassen notes, two phenomena standing in the strictly stronger relation must have distinct identity conditions and distinct explanatory properties. An object may be red without being scarlet, therefore the phenomena cannot be identical. Furthermore, my liking of your jumper may be explained by the fact that it is scarlet but not by the fact that it is red. I may have hated it had it been maroon.
With the “strictly stronger than” relation defined, I will now outline the principle of halfway proportionality. The principle of halfway proportionality, reframed in terms of explanations, states that P explains E iff for any P+ strictly stronger than P, P screens off P+ from E (Vaassen). “Screens off” is defined as follows: for any three non-identical phenomena P, Q and E, P screens off E from Q iff E would still have occurred if P had occurred without Q. That this principal fits with our typical understanding of explanations can be seen when we consider cases in which we add further details to our causal explanations. Yablo provides the following example: Imagine we have a chicken that is trained to peck at red objects. If we placed a scarlet disk in front of this chicken the chicken would peck. We wish to say that “the chicken pecked the disk because it was red” is true. However, we further wish to allow that “the chicken pecked the disk because it was scarlet” is true. We could keep doing this down into extreme levels of detail, e.g., “The chicken pecked because Γ,” where “Γ” is all the information regarding the chicken as a physical system. We still want to permit our ordinary explanatory talk such as “the chicken pecked the disk because it is red.” If not, our ability to make general explanatory claims such as “my chicken pecks because of red disks” would be removed. However, we clearly still need a limit for what may count as an explanation, strictly weaker conditions need to “earn their keep,” and the halfway proportionality condition ensures this (Vaassen 19). “The chicken pecked the disk because the disk is colored” is a more general condition than “the chicken pecked the disk because it is red.” However, it is not the case that the disk that is colored screens off the disk that is red, for if it was colored and not red, the chicken would not peck. Hence, the principle of halfway proportionality ensures that strictly weaker conditions can still be considered valid explanations, whilst ensuring anything dubbed a valid explanation still has explana-tory value.
Now that we have these resources, consider again talk of intentions and BDs. I argue that given the reduction of intentions to BDs, an agent’s being in a state of intention is a strictly weaker phenomenon than an agent’s possessing the BDs to which that intention has been reduced. Yablo highlights that if one phenome-non supervenes on, and is multiply realizable by, another set of phenomena, this ensures that the supervening phenomenon is strictly weaker than those phenomena that it supervenes on. By the definition of supervenience, X supervening on phenomena of type Y means that the occurrence of some correct phenomena of type Y necessarily entail the occurrence of phenomenon X. X’s being multiply realizable by phenomena of type Y also ensures that the occurrence of phenomenon X does not entail the occurrence of any specific phenomenon of type Y. This asymmetric necessity is clearly that which is involved in the definition of the strictly stronger relation.
BD1. The beliefs that this is the First World War, my brother is German, I am French, and that all contrary evidence once I leave the hypnosis is a lie. The desire that I capture any enemy nationals I discover.
BD2. The beliefs that this is the First World War, my brother is French, I am German, and that all contrary evidence once I leave the hypnosis is a lie. The desire that I capture any enemy nationals I discover.
Either of these BDs would perform the same functional role as my intention; either way I attempt to capture my brother. It would seem that a given intention can not only supervene on, but be realized by, several different groups of BDs. As such, by the definitions provided, possession of a given intention is a strictly weaker phenomenon than the possession of the BDs into which it is reduced.
I have argued for 1–4 of the given argument and will now argue for the final premise: Intentions screen off the actions that they explain from the BDs that they are held to reduce to. Consider again the definition of screening off. For any three non-identical phenomena P, Q, and E. P screens off E from Q iff E would still have occurred if P had occurred without Q. If I still had my intention to capture my brother, but I did not have BD2 because I had some alternative BDs realizing that intention, then I would still attempt to capture my brother. As such, we can see that for any intention that is multiply realizable by BDs, this intention screens off those particular BDs to which it is reduced from the behavior that they explain.
More than this, however, it can be argued that intentions have a greater explanatory value than BDs. We have a legitimate preference for more general explanations, favoring a strictly weaker explanation over a strictly stronger screened off explanation, in that the screening off of the strictly stronger explanation appears to reveal that the extra information in that explanation is not relevant to explaining the event. Consider again the case of the chicken pecking the disk—that the disk was red feels like an appropriate explanation, capturing all and only the relevant facts. The set of details of all the facts about the chicken as a physical system appears wildly disproportionate and generally irrelevant to explaining why the chicken pecked. Part of the motivation for relevance is robustness of explanation—the less we must specify for our explanation to be sufficient, the more probable that it remains correct in light of new information. Another element of this preference may be the assistance in our ability to draw general explanatory principles. Compare “my chicken pecks because of red objects” to “my chicken pecks because Γ” where “Γ” gives all the details of my chicken as a physical system relevant to my chicken pecking. To give an example with intention, imagine I am trying to explain to a friend of mine why I have taken up running. My best explanation is that “I intend to get fit.” The explanation where I state all the beliefs I have regarding how running will get me fit, how I desire to be fit, and how being fit corresponds to all my other desires, etc., appears greatly out of proportion. Due to this preference of explanation, even if we do accept the reduction of intentions to BDs, explanations that utilize intentions must still be taken not only as permissible but in fact superior to those utilizing BDs. As such, the motivation of the reduction is undermined as the reduction can only obfuscate our understanding of mental faculties.
VI. Are Desires and Beliefs Proportional Enough?
It may be tempting to interject at this point that BD explanations are nothing like the physical system style explanations given in the chicken case. BD explanations may be a strictly stronger explanation of why we act as we do in the fraternal capture case than intention explanations, but it still seems that BD explanations reveal something important to us in that case. It may be objected that a more fitting analogy is like that between the explanations “Socrates died from drinking a deadly poison” and “Socrates died from drinking hemlock” (to borrow Yablo’s example). The second is clearly a strictly stronger explanation and is screened off by the first, but the second explanation is still proportional enough to that which is explained that we do not find its details irrelevant and may like it as much as the first explanation. Thinking particularly of intentions, it seems clear that there are scenarios where strictly stronger explanations are more useful. For example, the finely detailed talk of the chemical and electrical reactions in the brain of an individual performing some action Φ may be a much more appropriate explanation in a neurochemistry lecture than talk of the agent’s intentions.
This objection requires greater discussion, and a thorough defence of the value of more general explanations is warranted and worthy of a paper of its own. My response at this time is as follows: Ιt is true that sometimes explanations with a greater level of detail can be preferable to those of less detail—particularly if we have a specific purpose in mind (e.g., if we are a doctor trying to save a patient’s life, knowing which poison they drank may be of use). However, in the specific case of BD explanations, I maintain that the BDs are still perceived as explanatorily relevant, not because they serve some specific purpose for which detail is needed, but rather because our under-standing of agent behavior naturally involves explanations in which an agent’s beliefs, intentions, and desires interact with one another as distinct mental kinds. In our natural discussion, we talk of what an agent intends but we also talk of an agent’s BDs and how they interact with this intention. We may say, “the man has some delusional beliefs and he intends to act upon them,” or “I know he desires to quit smoking, but he is intentionally putting himself through great temptation.” On the natural view, BDs are on the same level as inten-tions; they are mental kinds of a comparable sort both explanatorily and ontologically. From a reductionist viewpoint, intentions and BDs are locked in a hierarchy, where the intention is placed on a level beyond the “simple” BDs as a complex formed out of them. This restricts how we naturally wish to give some explanations in that it limits the information we can include on a single explanation to information of either intentions or BDs.
VII. Conclusion
The reductionist project to analyse intention as BDs thus appears to be ultimately self-frustrating. Unlike other anti-reductionist arguments, I have not attempted to refute the possibility of the reduction directly. Rather, the acceptance that intention can be reduced into BDs has been shown to ensure, given the principle of halfway proportion-ality, that explanations involving intentions are still explanatorily permissible, and in fact are preferable to explanations involving BDs. Furthermore, the reduction has been shown to introduce an explana-tory hierarchy that is frustrating, blocking our natural desire to provide explanations that involve beliefs, intentions, and desires and instead forcing us to utilize BDs or intentions exclusively within a single level of explanation. The reductionist project has thus been shown to obfuscate our understanding and frustrate our explanatory talk. As such, further work in the spirit of Bratman and Holton, aiming to develop an account of intention that takes it as a basic mental kind alongside belief and desire, is warranted.
¹ Special thanks to Dr. Will Hornett for pointing me to the works of Vaassen and Yablo and for his invaluable guidance as a supervisor.
² Providing thorough definitions of either ‘belief’ or ‘desire’ is itself a subject of some controversy, see Schwitzgebel and Schroeder for an introduction to the relevant debates. Whilst providing precise definitions of the terms “belief” and “desire” may prove useful to the discussion at a later date, a precise definition is not necessary to the current discussion and is not utilized by Davidson in the papers cited here. I rely simply on the reader’s good-will to understand them in the common senses.
Works Cited
Bratman, Michael;. “Two Faces of Intention.” The Philosophical Review, vol. 93, no. 3, July 1984, pp. 375–405.
Davidson, Donald. “How is Weakness of the Will Possible?” Essays on Actions and Events: Philosophical Essays Volume 1, Clarendon Press, 2001, pp. 21–42.
---. “Intending.” Essays on Actions and Events: Philosophical Essays Volume 1, Clarendon Press, 2001, pp. 83–102.
Holton, Richard. “How is Strength of Will Possible?” Weakness of Will and Practical Irrationality, edited by Sarah Stroud and Christine Tappolet, Oxford University Press, 2003, pp. 39–67.
Schroeder, Tim. “Desire.” The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, Summer 2020 Edition, plato.stanford.edu/archives/sum2020/entries/desire/.
Schwitzgebel, Eric. “Belief.” The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta and Uri Nodelman, Spring 2024 Edition, plato.stanford.edu/archives/spr2024/entries/belief/.
Vaassen, Bram. “Halfway Proportionality.” Philosophical Studies, vol. 9, 2022, pp. 1–21.
Yablo, Stephen. “Mental Causation.” The Philosophical Review, vol. 101, no. 2, Apr. 1992, pp. 245–80.
