Intentions and Permissibility


This paper examines a number of issues arising from an analysis of the relationship between intentions and permissibility. This analysis is developed in response to the straw man position, ‘IIP’[1]; that the content of the intentions held by an agent in performing some act are irrelevant in determining the moral permissibility of that act. This is done intentionally as assurance that IIP is not confused for my own position.

The crux of this paper has its origins in my resistance to accept double effect reasoning as I understood it and the work of both Tim Scanlon and Judith Thomson on the matter. Examples regarding double effect reasoning, in cases such as euthanasia, can often tend towards unappealing results. You want voluntary euthanasia. Under most versions of the Doctrine of Double Effect, someone who intends to relieve your pain, foreseeing your death may permissibly euthanize you. However, someone who intends your death, though foreseeing this will relieve your pain may not permissibly euthanize you. That the permissibility is hinged on the agent’s intentions rather than, more plausibly, the subject’s consent, is dissatisfying. My aim, then, is to discover whether there is a method of cashing out the relationship between intentions and permissibility, specifically examining whether this relationship is less comprehensive or encompassing than traditionally thought. To do this, I will begin by offering some motivation for the initially implausible view that intentions serve no role in determining the permissibility of an act, and work up by providing plausible exceptions to such a position. Throughout I work under the assumption that permissibility is a binary term[2], insofar as an act is either morally prohibited or it is not. For this reason, positive moral duties or requirements are usually categorised as ‘at least permissible’.

Acting Intentionally:

Our ordinary use of the term ‘intentions’ is opaque; when we ask “what were your intentions?” there are several understood uses, which can be explicated as asking what actions the agent intended to carry out, what reasons for acting the agent held and how the agent (or some other group or standard) would evaluate the intentions the agent held. These are reflected by the questions “What did you intend to do?”[3], “Why did you intend to do it?” and “Were your intentions good?” The first two of these questions seem to cover the majority of our use[4]. Since only the “reasons for action” usage does not make use of the concept of intention in its explication, I will stick with it for the time being. Reductive accounts of intention to other mental states have been proposed (a mixture of or reduction to future directed beliefs, desires, reasons for acting, or belief-desire pairs). You can have and believe you have sufficient reason to x without intending to x (due to conflict of reasons). Similarly, intentions are not simply reducible to desires; you might desire to go to Paris and go to New York on Sunday morning, but you may not rationally intend to do both, since you know this cannot be done. The use I am primarily concerned with in this essay is the reasons held salient[5] by an agent for their past or future behaviour. These reasons will typically be expressed in terms of some other proximate or distant goal. I am not particularly committed to this view and am happy to treat intentions as used in moral contexts as irreducible. The actions I am interested in are those performed for some intention, and not just by someone possessing some intention. Intentional actions are roughly of the form “Agent A φ’s in order to bring about state of affairs S, for reason, R”

When there is a lack of unity between the intention behind the action and the action performed, we will believe that the action was unintentional, and will usually consider it accidental. An action cannot be intentional if the agent is unaware of performing it, even if there is, or the agent holds, some reason for performing the action.  However, awareness alone does not make an action, let alone an intentional one. The standard example here would be falling down the stairs. You cannot fall down the stairs intentionally unless you have some reason for doing so. Even so, it is still possible to possess a reason for falling down the stairs, and doing so, without this being the reason for which you fell[6]. Also, the agent may hold some other end or reason for the outcome of the action they perform, but so long as this is not the reason for which the agent acts, then the action was not intentional (in the way that matters). Somehow I have swallowed a ring, and I wish to throw it up. However, the stress of it all has caused me to have a headache, so I take some painkillers. The painkillers get lodged in my throat, which causes me to vomit. The vomiting recovers the ring. My vomiting was not intentional (in the narrower sense we care about) simply because I possessed some reason for throwing up. If asked the reason why I took the painkillers, I would say “in order to relieve my headache”. I did not take the painkillers in order to vomit, and I did not vomit in order to recover the ring.

Further, an act may be viewed as unintentional by the agent even when there is some unity between the action and intention. This would be when there is disunity between the intention and the way in which the act was carried out. This will typically happen when the achievement of the outcome is accidental or unexpected, but somehow causally dependent upon the agents acting. Even though the agent might say “I didn’t mean to do that”, we should not conclude that the action was unintentional. The outcome was as unintentional as unexpected for the agent, but the act was at least intentional in the sense it was performed with some intention. The outcome did not match the intention for which the action was performed. Intentional acts are merely those whereby we act meaning to do something. The intentional acts we are interested in are those in which some agent acts, meaning to bring about some x and succeeds or fails in doing this. Regardless of outcome these acts are still intentional.

Part I: Motivating IIP

My preliminary aim in motivating IIP is to show that intentions and permissibility do not track perfectly. Demonstrating this will at least show that intentions do not determine permissibility in every instance. What motivates this aim is the pair of widely-held yet plausible intuitions:

  • It is possible to do something morally impermissible with typically good intentions.
  • It is possible to do something morally permissible with typically bad intentions.

It could be the case that neither or only one of these intuitions holds true. However, entertaining such intuitions avoids violence (in this area at least) to our ordinary moral reasoning. My primary concern is with permissibility, and the instances in which intentions determine the permissibility of an act; I do not wish to claim that intentions are morally irrelevant; there is more to moral evaluation than permissibility. I hold that it is always morally better to do some given act with good, rather than bad intentions. I also hold that it is always morally better to do a permissible act than an impermissible act, even though the act may truly be better under some other evaluative standard. There are several positions that one could take with regards to the general relationship between intent and permissibility:

  • Intentions are irrelevant to permissibility.
  • Intentions are of far more limited relevance to permissibility than we ordinarily think, there are certain classes of acts where intentions determine the permissibility of the act.
  • Intentions largely determine permissibility of acts, but there are certain classes of acts where the intention an agent holds cannot affect the act’s permissibility.
  • The permissibility of an act is determined solely by the intentions an agent holds for acting.
  • As above, but in addition the intentions an agent possesses are themselves assessable for permissibility, even if they are never acted upon.

I will begin by motivating the first of these views. A promising method to show a discrepancy between permissibility and intent is as follows: First, there is an act in which the intentions of the agent are given. The act is assessed regarding permissibility. Then, the intentions of the agent in the case are modified[7]. The act is then assessed again regarding permissibility. If the permissibility of an act has changed, the most plausible explanation is that this has changed because of the substitution of the intention[8]. If this is the case, then intentions are relevant in this instance to moral permissibility; some reasons for action make certain acts impermissible. There may be some worries regarding whether this can demonstrate an actual link between intentions and permissibility. However, at the very least it shows there is good reason to thoroughly examine the cases in which such results seem to occur.

Preliminary cases:

  1. Dave gives up eating factory farmed meat in order to improve his chances of sleeping with a particular vegan co-worker. Dave would not abstain were the expected outcome he holds otherwise.

I don’t think it’s possible (in the world as it is) to motivate a general moral requirement to eat factory farmed meat. It seems fair to say that Dave does a permissible act for suspect reasons; reasons which illuminate his character. We may word this case differently to heighten or lessen our attitude towards Dave and the action he takes. We might describe Dave’s intention merely as improving his standing with his co-worker, and this would be true[9]. However, whether this description would be appropriate is irrelevant; there is no substitute intention that could make the behaviour of “not eating factory-farmed meat” impermissible. This might be seen as a bold claim; there are some pretty rotten or even evil intentions an agent can possess. But similarly, we do not want to falsely say, “It is morally impermissible for Dave to give up eating factory-farmed meat[10]”. Let’s consider some alternate intentions, without changing significantly any of the other features of the case; perhaps Dave gives up eating factory farmed meat in order to:

  • Demonstrate his support for animal welfare (but tentatively).
  • Fractionally reduce the inheritance of a mortal enemy, who is the child of a factory-farming magnate.
  • Act upon his longing for simpler times and hatred of industrialisation.

In all of these instances he gives up eating factory-famed meat. In none of these instances does this action become morally impermissible. “So what?” one might think, “It’s only cases concerning harms or rights violations that we should be concerned with intent in this way”. Now, at a stretch, this act is merely a resolution regarding future acts of omission, Dave intend not to do something. If presented with factory-farmed meat (especially in front of the object of his desire), he will decline nearly any invitation for him to consume it. It may be the case that there is some unique moral restriction regarding resolutions. Instead, let us consider a case where Dave does not refrain and intentionally does something:

IIa.      Dave buys rat poison in order to spend some money. Is this purchase permissible?

IIb.      Dave buys rat poison in order to kill Tom’s wife. Is the purchase permissible?

In both instances it would seem so. If we consider the purchase of rat poison permissible in the first case it seems bizarre to think otherwise regarding the second. We are evaluating whether or not Dave may, permissibly, purchase the poison. We are not considering whether Dave may kill Tom’s wife, nor whether Dave is a bad person[11]. It is very important to note that judgments about a person’s character are separable from and need not track judgments about the permissibility of a person’s actions.

  1. Dave knows[12] Tom intends to kill his wife with rat poison, but does not have any. Dave gives Tom some rat poison. Is Dave’s action permissible?

Even though the result of Tom’s actions are epistemically open, if Tom went on to murder his wife, it would be true to say Dave had foreseen this as a possibility. Dave facilitates harm by acting as he does, and certainly this seems impermissible in view of the foreseeable outcome (presuming she is remotely innocent). Let us consider some substitute intentions and examine whether they affect the permissibility of Dave’s actions:

  • Dave wants to please Tom.
  • Dave hates Tom’s wife.
  • Dave intends to satisfy his necrophilia.
  • Dave wants to visit Tom in prison because Dave has never visited a prison inmate before.

It seems here that regardless of intent Dave’s action is impermissible. The intentions certainly flavour the action but only to change our opinion of Dave’s character. Dave is now “spineless…and does wrong” or “necrophiliac…and does wrong”. Here at least the judgements of character seem vestigial to the assessment of permissibility. What Dave aims to do when he hands Tom the poison primarily matters for determining how much and for what we blame him, but not whether he may do it. Dave’s action would also seem impermissible if Tom was only pretending, to gauge how committed to him Dave was.

IV.       Patario rescues Imhotep from drowning in order to slowly torture him to death later[13].Is Patario’s rescuing Imhotep permissible?

This is simply a pumped version of IIb. Ordinarily, saving a life is at least permissible. Patario’s character is cause for concern, but it is clear that this does not affect what he may or may not do here and now. It is at least permissible for Patario to save Imhotep. It is also impermissible for Patario to torture Imhotep to death. We can agree that, Imhotep will be worse off for being saved if Patario does what he intends, but our assessment is regarding what he may do now, not what he may do in future. Neither purchasing poison nor rescuing somebody commits us to killing them. We do not wish to be morally committed to prohibiting anyone from rescuing others from mortal danger.

V.        Imhotep is drowning in the drowning pool of Patario’s torture palace. Patario presses the button which activates the robotic arm which grabs Patario’s victims from the drowning pool and places them in the already running torture machine. This is all the arm does. Is Patario’s pressing the button permissible?

I have no problems in saying that this is impermissible regardless of intent. This is because this action may not truly be described as ‘rescuing’ or ‘preventing’ so much as ‘putting in a torture machine’. You can prevent someone from drowning by shooting them in the head so that they die, but you cannot rescue them by doing this. Patario lacks any form of control in preventing the torture from taking place once he presses the button, unlike when he rescues Imhotep himself. Motivating such conclusions is some version of the following:

IIP:      The content of the intentions an agent possesses at the time of performing some act play no role in determining the permissibility of that act.

Under IIP, there exist no intentions an agent could possess for any permissible act that would, if substituted, render that act impermissible, nor any intentions an agent could possess in performing any impermissible act that would, if substituted, render the act permissible. ‘A pure heart cannot make a wrong act right; neither can an impure heart make a right act wrong’[14]. Intentions are not morally vestigial under this account; this principle makes no claims regarding forms of moral evaluation besides permissibility. One plausible role is that the intentions an agent possesses allow us to determine what sort of person they are (for one, they are at least the sort of person who would intend to perform some act). In Scanlon’s terms, intentions have predictive significance[15]. We believe that those who act with malice even only in permissible ways, will be more likely and successful in performing acts that are typically malicious. If this is the case, then this view is not falsified if an act is shown to be morally worse than another equally permissible act in virtue of the intention for which it was performed; the moral worseness will be a product of character.

Such a role for intentions may seem obvious, and is certainly widespread. We think it is morally worse to perform an act for bad intentions instead of good ones. Particularly, we think that the intention in performing an act will determine whether the agent is blameworthy. We often do not consider accidents to be blameworthy (though we may do when we consider them negligent). Our assessments of character alter when we entertain and substitute the intention an agent is said to possess in performing some action.

To say ‘you ought to not act for that reason’; we are not necessarily making any claims about permissibility. This utterance is easily and habitually mistaken for a claim about permissibility. We are saying your reason for acting does not justify your action, but not that ‘you ought not to act at all’. The facts may justify your action. One ought to act for the best reasons, but one may permissibly act for others. Impermissibility does not entail blameworthiness nor does blameworthiness entail impermissibility. Doing kind of act x is permissible, but one should not act for some sorts of reasons if one is to be considered a good moral agent.

Even if we think that the acts an agent performs determine our assessment of their character, usually negative, it will be because we expect a unity between the act and the intentions. We may believe that someone spilled some ink on our carpet out of maliciousness (even if it really was an accident). It is what we believe their intentions to be that does the work in our assessment of them. We take intentions as more important than acts and their consequences for evidencing how an agent reflects upon and respond to reasons.

In opposition to IIP, we might say that we can imagine instances whereby merely forming an intention is impermissible, say, when one has rigged a bomb to explode in some crowded area whenever you form the intention to buy ice cream[16]. Or consider a case where someone else has, unbeknownst to you, rigged this bomb to your intentions? There is a sense in which you ought not to form this intention[17], while rationally you clearly ought to. Such cases are not our interest, our concern is whether permissibility can hinge upon the content of intentions rather than the other concrete features of the case (where clearly here it does so, the impermissible intent is whatever one you have programmed the bomb to).

There are some doubts regarding the different kind of descriptions an action might have. The worry is that reports of intentions are either not fine grained enough or that intentions just aren’t fine grained entities. The agent my think to themselves “I will run him over” rather than “I will run him over with car x”. You are planning to run someone over, but you’ve stored the car you’ve chosen for the job elsewhere. Whilst driving to pick up that car, the person you are planning to run over walks out in front of your car. There is a sense in which hitting them then and there was unintended, or even accidental. It seems obvious, though it is worth pointing out, that actions can be described in many different ways, and the locus of description can alter our assessment of permissibility[18]. I take it as given that some descriptions will not omit any morally relevant-features, and will hopefully appeal to these.

Some Initial Objections:

If the unintentional twin of an impermissible, intentional act (typically what we would call an accident) is permissible then this might be taken as evidence that intentions determine permissibility.

P1.       A certain intentional act is impermissible.

P2.       The unintentional version of this act is permissible.

C1.       So the impermissibility of this act was determined by the intention the agent possessed.

This is not an argument I will appeal to. The fact that the second act was unintentional means it was no longer the same kind of thing, nor comparable in the same way. The agent no longer has any beliefs or desires directed toward the act or the consequences the agent believes[19] the act will have. Unintentional is not substitutable with possessing a differing intention. You may unintentionally fall down the stairs and break someone else’s vase doing so. You may not permissibly intentionally destroy their vase if you make it look like an accident. That an act need only be intentional to be impermissible determines nothing regarding whether the specific intention in question determined this; since not even accepting IIP, any intention would seem substitutable with the same result. More importantly, whether this could function as an argument against IIP is dubious; IIP only claims the content of intentions is substitutable (what they refer to, not that they refer). A more promising argument against IIP is as follows:

P1.       A certain intentional act is impermissible.

P2.       This act with a substitute intention is permissible.

C1.       So the permissibility of this act was determined by the intention the agent possessed.

C2.       So intentions are sometimes relevant to permissibility; IIP is false.

Whether any cases fitting this argument can be presented is another matter; this argument mirrors the method I used to motivate IIP. It may be objected that simply changing the reasons an agent has for performing some act changes the cases too radically to draw any useful conclusions, as by changing the reasons we are forced to change a large number of the agent’s beliefs in order for them to be coherent. Certainly, an agent’s beliefs would seem to change the meaning of their actions significantly enough to change sometimes whether it is appropriate to call them the same kind of action in all relevant respects. That said we see no sort of hesitancy, and I think rightly, employing such methodology in support for the view that intentions are relevant to permissibility; DDE cases will typically offer contrasts by substituting the reasons for which an agent acts. What follows appears a more immediately problematic case for IIP:

VI.       Antonio & Bernardo are involved in some competition to win some prize. Bernardo is slightly more likely to win than Antonio. In the past Bernardo has committed some minor crime. Antonio knows this. Antonio also knows that by the (irrelevant) specificities of the crime he could get Bernardo kicked out of the competition by reporting Bernardo for the crime. By doing this, Antonio would win the prize. Antonio wants the prize, so Antonio reports Bernardo.[20]

The thought is that, since it seems true that it is normally at least permissible to report a crime then it seems Antonio cannot have done wrong (especially if IIP is true).  But, we are inclined to hold, that Antonio has done wrong.  It then seems to follow that intentions (at least here) must count, and if this is true, then IIP is false.

In examining this, we must consider that the crime is both minor and irrelevant to the competition and that the judges act unjustly in kicking Bernardo out of competition. What Antonio does is beyond facilitating wrongdoing. If indeed the judges’ actions are static to the extent Antonio knows they will kick out Bernardo if Bernardo is reported then Antonio will be held as responsible[21]. Patario does not merely facilitate harm or wrongdoing by putting Dave or Imhotep in the torture machine.

However, it is not always at least morally permissible to report a crime[22], in the way that it is not always at least permissible to prevent someone from drowning[23]. Principles such as this are open to overruling reasons (it would be possible to ethically justify civil disobedience, even if it were legally prohibited). Firstly, we may substitute Antonio’s intentions to see if they are reasonably the source of the impermissibility. Antonio might report Bernardo:

  • So that another competitor, Cervantes, might win.
  • Because Bernardo is a Christian. Antonio hates Christians.
  • Because Antonio equates the law with Justice.
  • Because he thinks it’s unfair that someone slightly better than him should be allowed to compete.

There is no way in which Antonio’s action does not violate Bernardo. It is, however, not the intention in these cases that grounds this violation. Imagine that Antonio was not even involved in the competition (seeking neither his own advantage nor schadenfreude), but reported Bernardo to the judges anyway, knowingly removing his chance of competing, instead of waiting until the contest was over. It must be something special about the nature of competition that restricts the permissibility of the act “reporting for a minor crime”. In the context of contests, there is something wrong with sabotaging your potential competitors no matter the reason. Some principle akin to the following is at work:

Fair Play:         One must not act in ways outside of competition which will foreseeably[24] undermine the participants’ ability in the competition (unless there is/they have a sufficiently serious reason to do so)[25].

This might cover why it is wrong to schedule a wake-up call to a competitor’s hotel room at 3am the night before the contest, whether they are a rival or not. To be reported for a minor crime, the result of which will remove you from competition certainly undermines your ability as a competitor. To be reported after the competition does no such thing[26]. It seems epistemological considerations are far more central than intentions here. If Antonio does not know Bernardo is also involved in competition, and Antonio reports Bernardo, it seems that whatever reasons they hold they have done nothing impermissible, or at least what they have done is not blameworthy.  Now if Antonio does know who Bernardo is and, not wishing to be thought a scoundrel, asks Cervantes to report the crime, Antonio does wrong insofar as they facilitate harm (whether Cervantes does wrong or is blameworthy depends on whether he knows what Antonio knows). If before acting Antonio discovered that Bernardo was involved in the same competition, the meaning of Antonio’s action (between Antonio & Bernardo) is changed. Antonio knowingly violates Bernardo. The morally non-problematic thing to do would be for Antonio to wait until the competition was over before reporting Bernardo. And that such an option exists means plausibly Antonio ought not to do the potentially problematic thing.

Acting after a contest is unproblematic. Acting before or during are increasingly problematic. Even if he is reported after the contest, and stripped of his title, Bernardo will at least know he was able to win on his own merits (in most people’s eyes, he will still have won, since his title was stripped unjustly). The suggested principle would not rule this action impermissible. Antonio or anyone may report him for a major crime, regardless of intent (Antonio might not care at all about the victims, and just be interested in winning). This is because the reason to report the major crime is stronger than the one not to undermine the Bernardo’s performance[27]. In cases of exclusive choice, we should go with the stronger reasons.

VII.     Dave and Parfit are in competition to be awarded PhD funding. Each offers to take the head of the funding committee out for dinner. Dave does this with the intent that it will improve his standing in the running. Parfit does it because she thinks the head of the committee has done a great job elsewhere. Neither are explicit to the head in their reasons[28].

It’s clear that what Dave does is bribery; however, the act of taking someone out for dinner does not seem the sort of thing which could generally be impermissible. But if IIP is true, then Dave’s bribery (something we believe is impermissible) is of the same stripe as what Parfit does. If something like IIP is true, then neither acts permissibly or both do. My proposed solution within this framework is that neither does, and for the same kind of principle that prohibits Antonio from reporting Bernardo’s minor crime. One student is more blameworthy than the other, but what neither does is permissible (or even sensible). Both at least foresee that this may unduly affect their chances yet continue to act; such action is blameworthy regardless.

Part II: Attempts

One line of inquiry which shows promise in illuminating IIP’s viability is in examining the nature of attempts. In law most crimes have an analogue, less seriously punished, attempt version. ‘Inchoate crimes’ are those which cause no actual harm, but aim at some of the kinds of harm the law seeks to prevent. This includes failed attempts to do the impermissible, solicitation of someone else to commit a crime, and conspiracy to commit a crime. It’s possible that a moral analogue exists.

There is the distinct possibility that attempts, insofar as this refers to failed actions, are not impermissible. This would probably be the case if IIP is true. By this account, we might offer the explanation that our beliefs on the matter are simply a distorting effect of our character judgement on the agent for the manifestation of an intention failing to do the morally impermissible. The agent of an attempted murder will have possessed the same intentions as the agent in a successful murder (other factors will explain why the one failed while the other succeeded). Since we distrust the two for their intentions in much the same manner, we mistakenly consider what both do as impermissible; when perhaps only the latter has acted in a way warranting this assessment. A similar phenomenon explains our hesitancy in granting permissibility verdicts in pumped cases (such as Patario’s rescuing).

This area I have otherwise considered a problem for Consequentialist positions (Consequentialist positions gel naturally with IIP[29]; intentions only mattering derivatively). When an attempt to bring about some significantly worse state of affairs fails, in what are we to ground the perceived wrongness? Is an inept assassin who scares their mark by forgetting to use a silencer significantly worse than an inept assassin who remembers? One who scares a room full of people as opposed to one? It seems that they are worse, but the degree of moral worseness is not significant. What matters is not the manner or margin by which they fail and bring about other comparably minor negative consequences, but the fact that they have attempted to kill someone who does not want to be killed.

Pure attempt crimes appeared in western law remarkably late[30].  A brutal assault where the perpetrator intended to kill rather than merely injure has at least the violence to ground the criminal wrongness in, the attempt was partly constituted by completed wrongdoings. It is very easy to construct cases of attempts to do the morally impermissible whereby the intended victim is unaware of the failure of the attempt against them. There is no harm to appeal to in these cases to ground the wrongness we perceive, nor is it an appealing option to declare or be committed to the view that the wrongness we perceive in these cases is illusory (whilst real in those where the intended victim is aware of the attempt, where it would merely be excessive). Attempt law is a particularly tricky and unusual domain:

“As Wayne LaFave explains, ‘‘The mental state required for the crime of attempt, as it is customarily stated in the cases, is an intent to commit some other crime.’’ The other crimes, however, often are not defined in terms of a specific intent, and this gives rise to moral anomalies. As LaFave points out: [I]f A, B, C and D have each taken the life of another, A acting with intent to kill, B with an intent to do serious bodily injury, C with a reckless disregard of human life, and D in the course of a dangerous felony, all [four] are guilty of murder because the crime of murder is defined in such a way that any one of these mental states will suffice. However, if the victims do not die from their injuries, then only A is guilty of attempted murder.”[31]

We can imagine a standard trolley case where the agent has no intention to save the five, just to kill the one. They have the intent to see only one person die. They could not achieve their goal without the one dying. If they turned the train and it derailed, killing no-one, we would presumably have some case of attempted murder. Because it doesn’t derail and the person does die, we have an act. If, as some believe, it is at least permissible to turn the trolley and IIP is true, we have a permissible act. Legally, it’s a murder. If IIP is true, then it’s a morally permissible case of murder.

Linguistically, ‘attempt’ entails ‘intend’. That is, one cannot attempt to do something without intending to do it, though one can intend to do something without attempting to do it. ‘Attempts’ necessitate ‘intends’. Attempts to x necessarily require not just that one intends, but that one intends to x. My claim is that this relationship is not a contingent truth about the meaning of the terms, but a fact about that nature of attempting. On this view, unintentional acts (or intentional acts’ unintended consequences) cannot be considered attempts, since they lack the necessary ‘in order to’ feature. A challenge, then, would be to present a case of someone attempting to do something without intending to do it[32]. Any account of attempting we do offer will have to reflect this link; if someone didn’t intend to x, then they did not attempt to x.

It might be objected on behalf of IIP that ‘attempt’ does not entail intend. The reason we hold for performing an action can be other than that of completing the action we attempt[33]. Someone might apply to jobs not to get the job but to practice job applications, even if they’d take any of the jobs if offered. But this is not an adequate counter-example; they do possess a reason to succeed (they want a job), but this is not the reason for which they act (they want to be well practiced for when a job they love comes along). The attempter will be fully satisfied with their actions whether they get job offers or not. As a result the agent is attempting, but not to get the job[34].

You might try to lift something you do not believe you can lift, in order to demonstrate that it cannot be lifted. The fact that in one variant it turns out to be a copy made of Styrofoam and you do lift it is irrelevant, as is the fact that your act would be physically identical to someone who believes they can lift it. The difference is you try to lift it, without attempting to lift it (for you lack the required prerequisite mental state). We can attempt to y via trying to x.

Reconsider and hold fast the agent’s beliefs and desires, but remove the intention for which they actually act. They are now not attempting to demonstrate they cannot do something. They are attempting to lift the rock (while they believe they cannot lift the rock and desire to show that they cannot lift the rock). Substituting an intention in which is contiguous with attempting to lift the rock, believing they cannot lift the rock and desiring to show they cannot lift the rock is not easily done without also attributing irrationality to the agent. This is not the same as believing it is a long shot (going to the shops to get some bread on the off chance they have some left, even though you would bet against it, because you really want some bread). The agent believes they cannot do it; and the only intentions that can make sense of this along with the agents’ desires are the intentions to show they cannot lift it. It is not irrational to attempt long shots and it is not irrational to attempt something that looks to the uninformed observer like you are attempting to do something else. These agents do not even broadly intend to x as their belief blocks them foreseeing their x-ing.

Parfit Cases:

i)Parfit intends to kill Dave. She lies on a roof opposite his workplace and, with her rifle, aims at his head and pulls the trigger. The rifle is not loaded. Cursing her carelessness, she sighs and goes home.[35]

Parfit believed her rifle was loaded, and acted on this assumption. If we hold what Parfit does (say, pulling the trigger with the intention of killing Dave and the belief that her pulling the trigger will expedite this) as morally impermissible then this is best explained by appeal to intention. If IIP cannot account for this, this would seem to entail a plausibility cost in holding IIP. It’s uncontroversial that Parfit should be negatively assessed for this action[36], if only because the action goes beyond predictive significance (the action with that particular intent clearly demonstrates her to be a dangerous person). There are no negative consequences besides her irritation to latch onto. Pulling the trigger on an empty rifle just isn’t harmful. To illustrate:

ii)Parfit does not intend to kill Dave. She lies on a roof opposite his workplace and, with her rifle, aims at his head and pulls the trigger. She is play acting. She uses a deactivated rifle. She utters “Tango down.” and goes home.

Here the action remains largely static[37] but for the intention. What Parfit does here is permissible. The obvious candidate to explain this asymmetry in our judgements is the possibility that intentions play a determinate role. The intention also seems to change the meaning of the action; what kind of action we are discussing. We can call the first case an attempt, whilst calling the second a successful act. We might be a little dubious over this Parfit’s character here, but not because we believe this to be a case of attempted murder.

However, we cannot simply conclude that attempts are a class of actions where intention affects permissibility. Attempts are not easily defined. They rarely consist of singular actions and are multiply-realisable. It seems that attempts that have been abandoned early on are not relevant to permissibility in the same way as those that have failed due to moral luck. To illustrate:

iii)Parfit intends to kill Dave. She opens her front door to go to the building opposite his work so she can shoot him, but it is raining. She stays home.

It seems Parfit’s actions in ii) & iii) were permissible, and if this is correct, it is not the case that an attempt of any kind combined with an intention to do something morally impermissible, is impermissible. If these actions are impermissible, then the impermissibility cannot be a product of harm alone, since these actions harm Dave in no sense (though other failed forms of attempted murder may harm by distressing the victim).

Now, it might be objected that we should not appeal to this case in our analysis because Parfit iii) has not properly attempted anything. I would hesitantly agree; whatever she has attempted here (perhaps facilitating her later killing Dave), she isn’t attempting to kill Dave. Reasonably, an attempt might be required to include either everything to agent believes is necessary to carry out the action they intend, or everything actually required (from their end) to carry out the action they intend. If this is correct then Parfit has attempted to leave her house and failed, and there is nothing morally impermissible about this, regardless of intent.

In i) Parfit does something impermissible, even though she holds the same intentions as in iii). It seems that what would be considered the final stage/s of an attempt (pulling the trigger etc) to do something morally impermissible is dependent upon the agent’s intentions for its moral permissibility. Though this suggests an immediate boundary problem, it at least shows that there exists some possibly defined class of actions whereby intentions are relevant to permissibility; these I call ‘final act attempts’.

“Both D1 and D2 want to Rob the bank. Both D1 and D2 want ribs for dinner. Both drive to the corner with the bank and the grocery store. D1 drives with the intent to rob the bank, D2 with the intent to buy ribs.[38]

Both facilitate their own plans in so acting, but since neither of these acts constitutes a final act attempt it seems incorrect to label one as attempted robbery and the other as attempted rib purchasing. If we think of the cases examined previous to i), we see that some are clear incidences of ‘early’ attempts. Tom is in some sense attempting to murder his wife. Patario rescuing Imhotep is part of an attempt by Patario to torture Imhotep.  But rescue or transactions alone are far from sufficient in constituting or guaranteeing successful torture or murder; they do not constitute the final act of such an attempt. As such the intentions do not influence the permissibility, but would had they manifested themselves in the form of final act attempts to murder or torture.

Character Judgements:

You can always make a character judgement based on the intentions, beliefs and desires an agent possesses. You cannot, however, make character judgements with the same degree of reliability in isolation of this information.  The judgements made in these circumstances will be both speculative and open to revision based on new information. They take the general form of “If agent possessed x intention, then agent is …” We assign some likelihood to which intentions an agent possessed in so acting, based on the general association of acts of that kind.

This all said, we notice the character judgements we do make regarding those we know, or believe we will encounter as being greater in intensity from those we do not. Hearing of friends’ or friends of friends’ infidelity strikes us with greater force than hearing of those we do not know. There is also a large bias in the judgements I do make which is a function of how much they are likely to affect me. Having character judgements about people you never meet does not obviously serve such a purpose. However, judgements about a person you have not yet met do; if true, they ensure you are not at a disadvantage when you meet them. Since there is no reliable way of distinguishing between people you will never meet and people you will meet, then forming character judgements about people you have never met is safer than not. This, however, would not explain our tendency to blame the dead. We might hear of Blackbeard’s deeds believing we will never meet him, but we do not refrain from making any character judgements regarding him. Presumably, this is in part because we have little to no control in making these judgements.

Knowing an agent’s poor intentions in performing permissible acts gives us reason to be wary of them elsewhere. Predictive significance means this sort of agent will try to do these things, not that they will do them. We negatively assess two agents, inept and capable, possessing intent to do some wrongdoing in the same manner. The perceived wrongness then does not come from the risk that an agent might succeed (for we may believe the inept could not succeed), but from the fact that trying to perform the act is itself wrong.  If that’s the case then we look badly on their character not because we believe they will try to do wrong, but because we believe they are going to do wrong (insofar as they try to do something, which is impermissible whether they succeed or not). Ineptness does not alter the permissibility of attempts, nor temper the appropriateness of blame. In a case of attempted murder, that someone misuses the firearm due to a lack of preparation or understanding, and shoots themselves has no bearing on the moral value of their attempt. This is obviously not the whole picture. Some of the inferences we make regarding other agents are presumably unwarranted, erring on the side of caution.

Factually Impossible Attempts:

There are, however, other more exotic classes of attempt. These should be of interest to us in finding further flaw with or defence of IIP. Douglas Husak writes:

“Suppose that our would-be arsonist believed that he had made even more progress toward his objective; he tries to burn the hay by holding his lighter to the straw. In fact, a shrewd policeman has switched substances in the lighter, replacing butane with a non-flammable substance. To my mind, the policeman has done a commendable job of gathering incriminating evidence against the defendant while preventing a risk that harm will actually occur. Any satisfactory account of attempts should hold this defendant to be guilty of attempted arson. Were the facts as he reasonably believed them to be, he would have succeeded in setting fire to the hay.”[39]

However, Husak goes on to say:

“Defendants should not always be guilty of a criminal attempt simply because they would be guilty if the facts were as they believed them to be. A number of cases reveal the kernel of truth in theories that require an objective risk in order to impose liability for attempts. When a defendant’s act fails to engage in the world in the appropriate causal way—as when he sticks pins in a voodoo doll in an effort to poison his enemy—we should not convict him of a criminal attempt.”[40]

It’s appealing to take such a line, though not obvious to pin down what the appropriate causal way is. We may imagine one person attempting to commit arson[41] by using such a lighter (the lighter fuel has been replaced with water). All the while[42], another person uses a working lighter to burn down a voodoo mock-up of the same building; their attempt is to burn the real building down via this method. Both intend to burn down a real building and believe this will likely happen as a result of their actions. Such actions are known as Factual Impossibility Attempts. A factual impossibility attempt has occurred when, at the time of the attempt, the facts prevent the possibility of the completion of the act attempted, though the agent is unaware of this fact[43]. Historical legal cases include:

  • A woman put sugar into her husband’s coffee in order to kill him, believing it to be arsenic[44].
  • A man shoots at a hole in his roof, in order to kill the policeman he believes is observing him from the loft, who was at the hole mere moments before[45].
  • Two men believed they were raping an unconscious woman, who was in fact dead at the time[46].

It’s impossible to poison your husband with sugar in his coffee, shoot someone who is not in the path of your bullet or rape a cadaver. They are not guilty of the ‘parent’ crime because they did not meet all the conditions, though they possessed the same intentions as someone who does. Little effort is required to justify our ascriptions of praise and blame in these cases; since you can be morally evaluated for all the things you try to do and you cannot do all the things you try to do; you can be morally evaluated for trying things you cannot do.

Regarding these cases, there are two reasonable options.  Firstly, if there are impermissible instances of attempts, then even attempts to do something impermissible that are impossible, are impermissible also. Alternatively, there is something about these impossible attempts that means they are always permissible. This may be some externalist criteria, perhaps that they are not ‘real’ attempts because there is no objective risk that they could succeed. However, for most species of determinist, the risk any ordinary failed act posed was only ever epistemic; a species of risk that doesn’t seem to adequately satisfy our intuitions regarding the impermissibility of attempts in general.

But we think Husak’s tampered-lighter case is an impermissible instance of attempted arson. It’s worth remembering that the first Parfit case counts as an impossible attempt in the same manner. If so and if we are to pursue this line, then there must be something that differentiates them from, or at least gives us pause over the Voodoo case. The would-be arsonist fails because of some external factor (things are not as they believe; they had no chance to start a fire with the water filled lighter) alongside the Voodooist (things are not as they believe; they had no chance to commit arson with voodoo). If answering truthfully after the fact, both would claim they were attempting to x, and I do not wish to shy away from this usage by suggesting their attempts were not ‘real attempts’. If it is the case that there is a significant difference between attempted Voodoo Arson and the attempted Arson, then this would suggest that intentions are less relevant with regards to permissibility as IIP suggests; since it would not be the intentions (which in these cases are practically identical), but some other features of the case that would determine this.

Belief would seem essential. I may genuinely desire that you might die as the result of me stabbing pins into a doll, but knowing full-well that there is no causal connection, I cannot be said to be attempting to murder you, even if I would not change my behaviour were reality otherwise (and I knew I could kill you this way). This is because there is no sense in which I am trying to achieve your death by stabbing the doll. I can achieve all I am trying to do (presumably express frustration) without your death.

Now, I have no problem admitting the Voodoo and mundane arson cases are both instances of attempts, and as such are morally blameworthy and morally impermissible in the same manner. I don’t view this as a cost; each intends and attempts to do something impermissible and fails. Possibility of success isn’t a condition of something being an attempt. A case might be made that the deluded are not morally culpable in the same manner. Culpability is not a function of moral permissibility but blameworthiness. It should be noted that both cases agents are motivated by false beliefs; that the lighter works, and that the Hollywood-Voodoo works. Both are instances of a false belief being pursued rationally. We do not generally rule out acting under a false belief as grounds for non-culpability. However, one suggestion is that the arsonist only has a single relevant false belief, but the Voodoo practitioner, in the relevant sense, possesses more or even systematically untrue beliefs.

The discrepancy we experience between the cases may be that Voodoo-doll-houses are not the sort of thing that can have remote effects when used by their ordinary methods, whereas lighters are the sort of thing that can set fire to things if used normally. Unfortunately, this response is too quick; water filled lighters are not the sort of thing that can start fires (and a burning voodoo house could be used to commit arson). However, an ordinary lighter is the sort of thing that, if not interfered with, is a sufficient tool for arson. An ordinary Voodoo doll house is not the sort of thing that, if not interfered with, is a sufficient tool for long distance arson. Perhaps the attempt cases we truly worry about are those in which we believe ‘If they tried it again, knowing the facts, they could succeed by their original method’. The arsonist would put butane in the lighter. The forgetful marksman would check their weapon was loaded. The Voodooist is out of luck, and would have to abandon their original method altogether. I think this explains our attitudes. We worry about the Voodooist, but not in the same way as the would-be arsonist or marksman. When we as evaluators also believe in witchcraft or voodoo, we will treat it the same seriousness as we would treat a ‘mundane’ attempt to bring about that consequence. Strangely, there is a sense in which fire-by-Voodoo is somehow more impossible than fire by water-filled-lighter. We ordinarily worry that the Voodoo practitioner is not remotely warranted epistemically in holding the beliefs they hold, even if they are fully rational in pursuit of them. However, the impermissibility of their actions does not hinge on any of these worries.

To illustrate, legality is typically externalist. If someone falsely believes the age of consent in the UK is 18, and believing this has consensual sex with a 17 year old, then they are attempting to commit an imaginary crime[47]; they believe they are violating some statute that does not exist. You can attempt to do something you believe to be illegal, without making a legally impermissible attempt. The moral analogue is obvious. Further, it does not matter whether someone is aware of the law that they can break it (though it may reduce the severity of punishment). If you attempt to do something that is a crime[48], whether or not you believe it is, you are committing a crime of attempt. If you do something you believe to be a crime, which is not a crime, you are not committing a crime of attempt. The law is generally sensitive to your beliefs, but it is not sensitive to your beliefs regarding the law.

In addition, if your false beliefs mean what you do would, if true, constitute a crime (you have sex with someone who is eighteen, believing them to be fifteen), then your actions constitute an impossible attempt crime. With relevance to moral permissibility, that the agent does or attempts to do something they believe to be morally permissible or impermissible has no impact on whether this is in fact morally permissible. Possessing intentions to do something you believe to be impermissible is not the same as possessing intentions to do something impermissible.  The moral evaluation of an agent’s intentional act is done in isolation from the agent’s beliefs regarding the act’s moral value.

Returning to Husak’s case, the Voodoo practitioner attempts arson; ordinarily morally impermissible. Whether they believe it to be so is irrelevant. If they attempt by Voodoo to perform acupuncture, believing the practice of all Voodoo is morally outlawed, they commit no wrongdoing because there is no moral prohibition outlawing Voodoo or acupuncture. It is essential for a verdict of impermissibility via attempt that one actually attempts to do something wrong, rather than something believed by the agent to be wrong. Put otherwise, you cannot bootstrap moral impermissibility by merely attempting to do something you believe to be impermissible, in same way you cannot bootstrap permissibility by ignorance of the impermissibility of some act you attempt to perform. It’s important to remember in this that Husak has some legal prudence concerns motivating his view, and his discussion is aimed at legal and not moral permissibility; plausibly, we cannot (and probably should not) convict everyone with insane beliefs for many prudential reasons.

Defining Attempts:

We are now, I think, in some position to offer an account of attempting which does not struggle in accounting for impossible attempts.

Attempted:      Y attempted to X if, at the time of acting while intending to X, was it the case that the facts were as Y believed them to be, Y could[49] (and in nearly all cases, would) have X’d.

Thus the mundane and Voodoo arsonists were genuinely attempting to commit arson. However, after considering fairly standard double effect cases we realise that this definition alone proves inadequate. Two standard trolley cases: Jill believes that turning the trolley onto the track where Jack is located (in order to save the five) will kill Jack. Jill+ also believes turning the trolley onto Jack+ (in order to kill him) will do this. It’s clear that Jill foresees Jack’s death as the result of her actions, and in this sense broadly intends to kill him. However, under this proposed definition, Jill and Jill+ might each be attempting to kill one Jack. We want to say Jill+ acts under a much narrower sense of intent; what she attempts (killing Jack+) is one and the same as her goal. For Jill this is not so, she only broadly intends Jack’s death. Appealing to the intended-foreseen distinction, and to isolate this sense of narrow intent, we add the amendment:

Amended:        Y did not attempt to X if Y could have achieved their end by their chosen method without X-ing.

Thus, while Jill is acting intentionally when she switches the trolley onto Jack, Jill did not attempt to kill Jack in the same way Jill+ attempts to kill Jack+. Considering cases in no-one is harmed (the train derails, perhaps) helps to illustrate this difference. Someone can apply for jobs they want while believing there is a very slim chance they will get them. They are genuinely attempting to get the jobs[50], for if the facts are as they believe them to be, they could get the job. The amended definition is compatible with cases of success and failure in this sort of attempt.

In relation, the person who applies for the jobs, believing them to be long shots does genuinely attempt, since they do so with narrow intent. They achieve all they intend by getting an interview. The person who applies to the long-shot jobs (jobs they would take if offered, but do not believe they will be) solely to demonstrate they are applying to jobs so as to continue to receive their unemployment benefits is not[51]; getting the interview would be broadly intended, a positive side-effect, but not the reason for which they act.

Patario’s rescuing Imhotep was not an attempt to torture Imhotep to death, though it is a necessary requirement for Patario’s later doing so. Patario can’t torture a dead man (and this is exactly why he rescues him). It fails to constitute an attempt itself because it fails to meet the necessary requirements, but it is part of an attempt. If Patario was pulling Imhotep from the river into a torture machine, then ‘rescuing’ Imhotep in this manner would be impermissible. It is not permissible to knowingly put someone into a torture machine, but this is not because of the intentions the agent possesses. Considering the fat-man variant of the typical trolley case, in pushing the fat man the agent needn’t require he dies. If the fat man stops the trolley but is either dead, seriously injured or unharmed, the agent has achieved their goal. Under this principle we do not intend the fat man dies; we may be happiest if they were uninjured, even though we foresee this will not happen. The same reasoning, if this principle is true, holds for the loop trolley case. Now, if these actions are impermissible, then it would seem that the impermissibility does not stem from the fact that they are attempts.

Other Avenues:


There are cases whereby it is clear the action the agent performs, regardless of the seriousness of the outcome, is a case of serious wrongdoing, even though their intention ensures they were attempting some less severe form of wrongdoing, or no wrongdoing at all. Someone fires their gun into the crowd at a rival gang member’s house party[52]. Their intent is merely to scare, though they do not care if people die as a result. However, they are not attempting to kill even though they believe their action may well result in the death of some party-goers; they would achieve their narrow aim were none of the party-goers to be killed or injured, just so long as they fear for their safety. Now, imagine the bare minimum is achieved, none are killed or injured, and all are scared. The agent was certainly not warranted in believing his action would not harm anyone (in fact they believed it was unlikely to be the case), they assigned some significant probability to this outcome. Putting aside issues such as property damage[53], the impermissibility here seems to stem from a number of factors.

You may imagine the rival gang have been made aware of the gang member’s intent, and have set up mannequins to give out silhouettes and audio imitating a party[54], whilst holding the actual event elsewhere. When the gang-member shoots into the ‘party’, believing it to be full of innocent bystanders, they neither injure nor kill anyone. Their mental state is relevantly similar to someone who does. Neither is justified in believing their action does not endanger anyone (both believe it does).

What the gang member does in these cases is not as clear cut as the arsonist with the faulty lighter. But they are not clearly cases of attempted murder, even though the agent has assigned some probability in each case to killing some of the absent bystanders. Their intent does not commit them to a course of action that will kill people, but one they believe might (but not will). This stuff is wrong in the same way throwing bricks over the fence into a schoolyard at noon without looking is; it’s risky whether or not anyone gets brained.

I might aim to crush your brain, without necessarily intending to kill you (even though I believe crushing your brain will kill you).When I do attempt to crush your brain, we are inclined to say I am attempting to kill you[55]. Likewise, it doesn’t seem like the strict content of my intent matters all that much when, were I to succeed, my crushing your brain would kill you. This differs from the gang case. There is a way in which they could succeed without killing anyone. Nomologically, one can shoot into a room full of people at random without significantly harming any of them. However, one cannot crush a living person’s brain without seriously harming or killing them. One cannot shoot into a room full of people or crush someone’s brain without at least seriously endangering someone. The gang member’s action is impermissible insofar as it significantly endangers the lives and wellbeing of every person present at the party without a sufficiently strong reason for doing so. It is reckless. Crushing the brain will kill someone who doesn’t want to be killed; and this is reason enough against so acting. Anyone who attempts to crush your brain is, or ought to know they are, committed to killing you. If they do not have a sufficient reason for behaving in this manner then this act is impermissible. The gang member outside the silhouette house attempts a form of reckless endangerment without good reason (a straight up impossibility attempt).

However, consider again the gang member outside the silhouette-filled house. This time, they believe the silhouettes to be imitations and decide to get in some target practice. Even though they are correct, they are not warranted in believing this for some barn country type reason. If they went into the building and saw the mannequins, they would be warranted. We would consider their target practice (without checking) hugely reckless, given that they are not warranted in ruling out a party going on inside, even though their action seriously endangers no-one[56]. Nor does their intent even commit them to a course of action where there is actually a significant possibility of harming anyone (since they believe it to be full of mannequins). We might find some similarity with drink driving cases; it is not often people drive drunk with the belief that they are significantly endangering people[57]; they often believe that it is not only not dangerous, but safe for them to drive. Epistemic recklessness here seems to lead to morally relevant recklessness. There is a parallel with impossible attempt crimes; the agent here is not endangering anyone due to the way the world is, but their own evidence, if they weighed it correctly, ought to lead them to believe they were. It’s otherwise clear we ought not to do things we should believe will endanger others without good reason. Perhaps we ought not to do things we ought to believe will result in some morally negative consequence without sufficient reason.

There are some cases where the agent believes more is necessary than actually is to bring about their goal. Your enemy believes you to be a vampire, and that you may only be killed by decapitation. They wish to kill you. You are not a vampire. Believing that vampires are highly resistant to drugs, they slip you hundreds of times the recommended dose of sleeping pills, in order to incapacitate you long enough to decapitate you. This kills you. They then decapitate you. The wrongness in this case consists not in your decapitation, but in them giving you sleeping pills. In giving you the sleeping pills they weren’t attempting to kill you, but they did. The seriousness in the wrongdoing seems to extend beyond the attempted incapacitation, and facilitation of murder. Another variant, and further reason to show that intent is not relevant, would see them acting with concern for your vampiric insomnia[58]. The reason for which they unintentionally mislead you into killing yourself does not alter the impermissibility.

Consider instead that they gave you this redosed sleeping pill, but you for some reason discarded it. We may wish to say that in this they made a failed attempt not only to poison you, but to kill you. But since they never intended to kill you by this method, why are we tempted to think this? This is again a case of ‘the beliefs they ought to have had’. Had they weighed the evidence correctly, they ought to have realised that giving you the sleeping pill commits them to a course of action (killing you), without having sufficient reason for doing so. This second case may be more reasonably explained as a case of blameless wrongdoing[59]. You give a sick relative some medicine which, unbeknownst to you, they are allergic to. They die. As with the would-be vampire hunter, you were not trying to kill the relative. But the ‘out there’ nature of the vampire hunter’s beliefs makes them much more akin to the Voodoo arsonist mentioned earlier; but as though instead the Voodoo arsonist believed the home must be ritually covered in flammable material and set alight before the Voodoo fire could properly do its work. What the agent was attempting to do in these cases seems a red herring. What they do is what is evaluable; their attempts to do otherwise themselves constitute acts (that happen to realise their attempts by a method other than they intend). Regardless in your belief that the effects are illusory, it’s impermissible to set light to a house you have just covered in flammable material, as it is to intentionally give a lethal dose of anything to someone believing it will only incapacitate them; so long as you merely ought to believe that the dose is lethal or flammable material burns.

Another possible worry for understanding attempts is ‘and/or’ intentions, where the agent would optimally achieve both aims, but either or both would be satisfy the agent’s intent. In an altered trolley case the track has two forks; one is an empty spur, the other leading to another fork. On the additional fork there are two persons tied up, one to each track. Jill+ turns the trolley at the first junction to the second fork. She does not know which way the second junction is set, but changed the trolley’s progress at the first junction in order to see at least one person die. She will be satisfied if either person is killed.

If these exist then the earlier amendment to the definition would seem to lead us into falsehood; agents could be said to attempt neither x nor y since neither is necessary to them achieving all they wanted to do. But ‘all they wanted to do’, would seem to imply that the conjunction was the optimal solution, even though any other outcome besides satisfaction of no condition is satisfactory. Since you cannot rationally intend to do all of a set of outcomes you believe to be exclusively achievable, then it seems unlikely that inclusive or intentions can exist under such circumstances. If this is so, then it might be that the agent really intends one outcome (of a or b or a+b), but possesses desires regarding the others tempered by beliefs of what is achievable.

It might instead be cases whereby the agent attempts to A, or A+B. In such cases, the agent usually has a desire for A, and a desire for B. However, they do not achieve all they set out to do if they only achieve B. If they achieve only A, there is a sense in which they do not do all they intend, but that they are satisfied with the result shows this to be misplaced, even if they would prefer B also. This is because intentions are not the same as desires. The agent really does achieve all they intend if they achieve A, and they achieve all they desire if they A+B.

This whole worry is, I think, misplaced. An ‘inclusive or intention’ maps out as a list of conditions (a, b or a+b) of which one must be satisfied for satisfaction of the intent. The focus of these investigations has been too narrow. Y didn’t attempt to X if they could have done all they intended without X-ing still holds true. Y indeed did not attempt to a or b or (a+b), because they intended to X [a or b or (a+b)].

One final possibility against IIP and weaker formulations is that possession of certain intentions is itself impermissible. That is, intentions need not be married to acts to be assessed for permissibility. Legally, possessing such intent in certain situations, even without making a last act attempt, can be criminal. There exist ‘with intent’ crimes that require no action; possession with intent to supply is a distinct crime from either possession or supply (or their attempt variants).

A possible candidate from the Prevention of Crime Act 1953:

“ […] and “offensive weapon” means any article made or adapted for use for causing injury to the person, or intended by the person having it with him for such use by him or by some other person.”

As soon as you intend to hit someone with your bag of ice cubes, your bag of ice cubes has transformed into an offensive weapon, and you have in virtue of your intention committed the offence of possessing one without reasonable excuse[60]. Here we shouldn’t import too much from the law; the category as a whole would seem motivated by the law’s preventative aim of deterrence; penalties for possessing intentions seem sensible in achieving this insofar as intentions are predictive. However, there are no legally impermissible mere possessions of intent. These ‘with-intent’ crimes are of the same broad stripe as attempt crimes, including inchoate crimes such as Conspiracy. They require a certain (even if minimalistic) set of physical circumstances beyond mere possession for the impermissibility verdict to be granted.

Mere possession of an intention as some reason to perform some future impermissible act incriminates character, but is not obviously itself impermissible. Why Parfit walked to her door before closing it doesn’t matter. However, as with criminal acts, possessing an intention to perform some act gives you a necessary component of carrying that act out. It is this reason, primarily, that we are wary of those with typically poor intentions. People who lack such intentions are not free to harm us in the same way as those who possess them; it is only possible for them to do so accidentally, unless they form the intention, or act in a way that foresees this harm. However, we might think that an instance of impermissible possession would not falsify IIP, as IIP is a principle regarding the assessment of acts. Impermissible possession would seem insensitive to this, since possessions are not acts. However, any act married to this kind of intention would be impermissible by extension, at least to generate an impermissibility verdict if the act is assessed, and as such generate infinite counter examples to IIP. The problem is they don’t exist.

Concluding Remarks:

We determine the act an agent attempts to perform by examining the act the agent intends to perform; an agent can attempt to lie or inform while expressing the same information. We clarify which features of the act the agent requires to deem their acting a success. We then assess the act the agent intends to perform. If this act is impermissible, then the agent’s attempting (as long as it consists a final act attempt) to perform this act is morally impermissible. Were the agent to succeed, the agent would also be acting in a morally impermissible manner. If it is morally impermissible to succeed in performing some act, then it is morally impermissible to do what you believe is sufficient to succeed in performing this act. If it is morally impermissible to murder, then it is morally impermissible to do what you believe is sufficient to murder, even if you are incorrect in your estimation of what will be sufficient, and even if you do not believe murder is wrong. As such, impossible final act attempts hold the same moral status as their possible moral counterparts[61].

There might be a worry regarding such a principle that it extends to absurdity. If it is morally impermissible to attempt to perform some act, it follows that it is morally impermissible to attempt to perform this attempt (and so on). This worry is illusory; uses of “I attempt to [attempt to x]” must contain a shift in the use of ‘attempt’ to remain at all meaningful. We can only attempt actions. Since attempts are themselves a species of action, it is not immediately contradictory to say we attempt to [perform some action which happens to be an attempt to perform some other action], much in the same way as we might “remember to remember” something. The relation does not seem obviously transitive. However, whether such creatures exist in the wild is doubtful. What attempting to attempt to bowl a strike would consist in (besides attempting to bowl a strike, under the current definition) is beyond me.

If correct, IIP, while appealing, is false. You cannot have an attempted-impermissible act without an agent possessing the intentions to perform this impermissible act. This attempted act is distinctly assessable from the successful performance of the act. However, successful attempts are superseded in evaluation by the acts they entail. These acts may be assessed regarding permissibility without necessary recourse to intent[62].

One who holds IIP might claim that this is not at odds with IIP. Intentions are irrelevant to determining the permissibility of acts, because once we know what act an agent is performing, we can evaluate them. Attempted murder is wrong, while play acting is not. We do not need to know the intentions in order to assess these acts, but we do in order to know which of these acts the agent was performing. Obviously this is not a satisfying defence of IIP, since knowing the content of the intention is necessary to determining permissibility in these cases, an intention to x in attempts to x cannot be substituted out with the action remaining an attempt to x. Defending an alternative principle, PIP; that Physical states are irrelevant in determining whether an act is permissible, emphasises the implausibility of this view; we need the physical features of an act to determine what the act is, but once we have them we do not need to know how the gunman was standing in order to evaluate his action. This does not entail that these features are irrelevant, far from it (there are a limited number of physical features that can instantiate most ordinary kinds of attempt to x, barring those in hallucinations).

However, the method used to motivate IIP plausibly motivates the view that intentions are of limited relevance to determining permissibility of acts generally, and are irrelevant in a large number of cases. This view is somewhat at odds to our folk understandings of morality, though it was our folk understanding that allowed the foot in the door for this position; that it is possible to do something wrong with the best of intentions. Why then, if intentions are of limited relevance, are we so willing ordinarily to accept the poor intentions some agent held as a satisfactory explanation of the impermissibility of that agent’s actions? Candidate moral theories should be able to explain why acts with near identical physical features that are performed with wholly different intentions result in us assigning them totally different moral evaluations. If intentions are of limited relevance, then this might seem not a particularly useful practice. However, intentions do track permissibility rather well, but the relationship, I hope to have shown, is one of correlation and occasional causation. We never have the sense of ‘it doesn’t matter’ about an agent’s intentions, and since intentions track character almost perfectly, it is in our own interest to be concerned with the intentions of others (and by extension the intentions we display). Insofar as unpleasant character traits tend to cluster together our concern with character only rarely fails to be beneficial for us (we may easily be misled into treating other agents badly by exposure to false information about them). Agents with bad intentions will plausibly (and be expected to) tend toward bad acts insofar as they are rational and capable of acting upon their intentions.

My argument has relied on permissibility being binary. My own thoughts on systems to the contrary is that any proposed evidence for them would just invoke a ‘more or less objectionable’ reaction from us, rather than indicating the acts are more or less impermissible. Donating £10 to charity doesn’t seem somehow more permissible than donating £5[63]. I don’t deny that the permissibility of acts might be epistemically indeterminate, but do claim that there are no possible acts that are neither permissible nor impermissible.

I believe there are more classes of actions than attempts to do the impermissible that show IIP to be false. Possible candidates are cases of deception where the agent aims at some good, acts within projects that aim at some good but themselves resemble or commit the agent to impermissible actions or what Scanlon calls ‘expression cases’[64]; such as calling your dying great aunt to express some fake concern in the hopes this will ensure your being written into the will (though these may collapse into deception cases).

Those more enthusiastic in embracing this conclusion that intentions have rather limited application in determining permissibility might reason as follows: in cases where the impermissibility of an act seems unclear, we might substitute the intentions of the agent to get a clearer understanding of the original case, so long as we do not change the kind of act performed (importantly not making it a morally impermissible attempt). If the modified act is of the same stripe as the original, we are to hold the permissibility of the original as matching that of the modified case. There are obvious pragmatic reasons not to adopt this strategy, the foremost of which is that we tend not to be very good at this sort of thing. Another major reason is that the other classes of acts whereby permissibility is responsive to intent still need detailing; without which would lead to incorrect application. Final Act Attempts are, I believe, the largest class of these intention-responsive-in-evaluation type actions.

What this view does generate for us is more work. If we accept some form of it we should refrain from accepting any statement of an agent’s intentions as satisfactory in explaining the proposed permissibility of the act, even if we agree with this judgement. We are given reason to re-examine other factors we may rashly dismiss when given an appealing intention-based explanation. The kind of intentions held in performing some act can be substituted to alter our character judgements of the agent, as better or worse under all kinds of evaluative standard[65], while only rarely changing the permissibility of the action they perform in addition. Obvious import for this view is in areas typically associated with double effect reasoning; most formulations of double effect would not function if the view I have expressed is true. Either both of the contrasting intent cases will be permissible, or they will both be impermissible (so long as the contrasting cases are relevantly similar in their other features)[66].




Hasnas, John:                          Once More unto the Breach: The Inherent Liberalism of the Criminal Law and Liability for Attempting the Impossible, Hastings Law Journal (2002) 54; accessed at

Husak, Douglas:                      The Costs to Criminal Theory of Supposing that Intentions are Irrelevant to Permissibility, Criminal Law and Philosophy (2009) 3:51–70

Rachels, James:                        More Impertinent Distinctions and a Defence of Active Euthanasia, (1994) in Killing and Letting Die, 2nd edition, Steinbock and Norcross, 139-54

Scanlon, T.M.                          Moral Dimensions: Permissibility, Meaning and Blame, (2008), Belknap Press of Harvard University Press

Walen, Alec:                            Comments on Doug Husak: The Low Cost of Recognizing (and of Ignoring) the Limited Relevance of Intentions to Permissibility, Criminal Law and Philosophy (2009) 3:71–78

Yaffe, Gideon                          Attempts: In the Philosophy of Action and the Criminal Law, (2010), Oxford University Press


[1] The name comes from Douglas Husak’s discussion (2009). He holds that some form of it is the received position amongst non-consequentialists who reject the Doctrine of Double Effect. As far as I know, no-one holds IIP. Alec Walen (2009) also holds that IIP is a straw man. I have clarified IIP to explicitly rather than implicitly refer to the content of intentions.

[2] I am not denying the possibility of epistemically indeterminate permissibility.

[3] Might be rephrased as “what action did/do you believe you had/have most reason to perform?”

[4] It’s possible I’ve missed out some other uses that aren’t covered by these three.

[5] An agent may have conflicting internal reasons.

[6] You are walking up the stairs so you can fall down them from the top, and film it to send into ‘You’ve Been Framed’. You slip on the third stair and fall the two feet to the ground.

[7] If the act is permissible, the modified act’s agent is given typically bad intentions. If the act is impermissible, the agent is given typically good intentions.

[8] Other candidates include beliefs, since it seems difficult to plausibly substitute intentions without fairly dramatically altering the subject’s other beliefs.

[9] Unless Dave is very delusional.

[10] Though it would not be contradictory to also hold “Dave should not give up eating factory farmed meat, for that reason”.

[11] He almost certainly is.

[12] Whether an agent knows what they do is an essential part of our moral evaluation. If Dave did not know because he was simply negligent will lead us to think worse of Dave than if Tom had deceived Dave.

[13] This example is known as Mill’s tyrant, but its author is Rev. J. Llewellyn Davies, a critic of Mill.

[14] Rachels, p142

[15] Scanlon, p13

[16] Husak, footnote p53

[17] This is the issue of different uses of ‘ought’.

[18] “You wanted them dead and you killed them”

[19] Or doesn’t, as is more likely.

[20] This case, I believe, originates from Joseph Raz.

[21] The judges are also responsible. A funny thing about responsibility is that there are no problems of overdetermination type phenomena regarding it.

[22] Legally permissible, I imagine so. There are plenty of historical examples of laws one morally should not encourage by helping to enforce.

[23] Again, you can shoot someone to prevent them from drowning.

[24] This word may be omitted for an external formulation of the principle.

[25] There’s a potential slippery slope argument against my position, insofar as minor crime committers could, in virtue of constant involvement in such contests escape justice forever. I’ll bite the bullet on this one.

[26] Or perhaps before you have entered the contest. However, Antonio acting in the same manner to remove a potential competitor would be just as vicious.

[27] We can reimagine the case as a competition for academic funding. One candidate has a past of sexist behaviour, and the school will not award funding to anyone who they believe will act in a discriminatory manner.

[28] Though if they were, Parfit’s action would still look like bribery.

[29] I understand here the language I appeal to is difficult to meaningfully gel into a Consequentialist framework without presupposing much. The language I have set out IIP in is guilty of this also.

[30]  The first widely agreed sentence was for an issue of attempted, yet unsuccessful arson; Rex v. Schofield, Cald.397 1784

[31] Walen, p75

[32] Which to me almost invokes a contradiction.

[33] Yaffe, p7

[34] Yaffe p39

[35] This example came from a long line of variant cases on “Parfit’s Gangster”. Her name reflects the origin.

[36] It’s not a morally permissible assassination or anything.

[37] Except her perceived failure/success behaviour, but as we are privy to her intentions these can be ignored.

[38] Yaffe, p5

[39] Husak, p63

[40] Ibid

[41] Like most crimes, Arson is defined with reference to intent. Assume that the building is not their own and they do not have consent.

[42] That these acts take place at the same time is not important. This is obviously not a case of overdetermination.

[43] You would have to be quite delusional to genuinely attempt something you believe to be impossible.

[44] Commonwealth vs Johnson

[45] People v. Lee Kong 95 Cal. 666, (1892)

[46] United States v. Thomas 13 U.S.C.M.A. 278 (1962)

[47] Hasnas, p9

[48] Such as attempt to have sex with a 13 year old (who is not thirteen, and above the actual age of consent).

[49] The use of ‘could’ is, as stated, restricted to within a reasonable timeframe of the action. You attempt to get a strike as you release the bowling ball from your grip, though whether or not you strike is determined a few seconds later.

[50] There is a whole class of lottery cases that might be thought similar.

[51] Though, there is little external difference with somebody on unemployment benefits who genuinely attempts to get the jobs.

[52] Yaffe, p47

[53] There are issues such as ‘not being guilty of property damage only because the bodies stopped the bullets’

[54] This is similar to a scene from Home Alone. Sort of.

[55] And the gang member is attempting (successfully) to endanger.

[56] Ignoring worries of persons occupying the adjacent houses.

[57] The ones that do and simply don’t care seem to evoke a greater contempt reaction than those who are simply drunk.

[58] Only, they wouldn’t decapitate you afterwards.

[59] But wrongdoing nonetheless.

[60] You probably have good reason for possessing a bag of ice cubes, but you do not have a good reason for possessing an offensive weapon.

[61] I remain neutral on whether they are as bad, I only claim that they both share the status of impermissibility. Two lights on dimmer switches are equally on or off, but they needn’t be equally bright.

[62] If we know it was murder and not accidental death, then we already know about the intent anyway by the kind of act performed.

[63] It may seem ‘less impermissible’ if we have a moral requirement to give to marginal utility, in the sense that we would be failing less. I would still have this drawn over the impermissibility line.

[64] Scanlon, p39

[65] A good business strategy can be morally objectionable.

[66] Some of my discussion on attempts might offer a way to motivate DDE, though this is not intended.

Leave a Reply

Your email address will not be published. Required fields are marked *