Criminal Law and Philosophy

, Volume 1, Issue 2, pp 143–149

Commentary to “Complicity and Causality”


    • Juristische FakultätRuhr-Universität Bochum
Original Paper

DOI: 10.1007/s11572-006-9014-x

Cite this article as:
Hörnle, T. Criminal Law, Philosophy (2007) 1: 143. doi:10.1007/s11572-006-9014-x

Preliminary remark

A theoretical analysis of complicity in moral norms and in the criminal law should address the following questions: What is the wrongdoing of complicity? Why distinguish between principals and accomplices? How can they be differentiated? In dealing with these questions, common cases should be considered. In ordinary cases, the accomplice is actively involved. In order to become an accomplice, one typically has to do something, like aiding the principal or procuring a crime.

John Gardner organizes his essay differently. He starts in a rather unusual place by analyzing a scenario of duress: person A orders person B to commit a horrible crime like flying an airplane into a skyscraper. If B does not obey the order, A himself will commit an even more horrible crime like having fifty airplanes flown into fifty skyscrapers. Gardner argues that B would be an accomplice to the crime committed by A if he chooses to remain passive. This conclusion needs discussion, which I will provide at the end of my paper under the heading “the passive accomplice”. Before examining special circumstances, however, it seems useful to address the “what”, “why” and “how” questions for common cases that involve actions rather than omissions.

The relationship between complicity and causality

John Gardner’s answer to the first question (what is the wrongdoing of complicity?) is straightforward and convincing: the accomplice is to be punished because he—like the principal—causally contributed to the harm done.1 Gardner concludes that one should not think about “wrong in the world” separately from those creating it as principals and accomplices. He is right about this. If the actor is to be punished either as a principal or as an accomplice, his act must have causally contributed to the prohibited result.

In the last section of his essay, Gardner discusses the possibility of “complicity beyond causality”. He mentions the well-known example of the tool (for example, a picklock) given to a burglar who uses it to get into the house, although he could have easily entered by other means (by breaking a window). One could maintain that lending the tool was not a cause of the offence. To address this problem, the issue of causality needs to be examined in more detail. A typical causality test asks us to give an account of a hypothetical course of events, that is, what would have happened without the act in question.2 Applying this test to the foregoing example seems to rule out causality. The thief would have entered the house without the picklock, and would have taken away the same goods. Pointing to the similar string of actions in the hypothetical “no picklock available”-version, one could argue that the provider of the tool cannot be punished as an accomplice. Gardner challenges this conclusion with the following argument: the “subtraction method”, the procedure to “think away” the lending of the tool, is not permissible. According to Gardner, the lender may not rely on hypothetical, alternative stories because in her own practical reasoning the lender did not think about the wrongs that would have happened without her own contribution. His thinking runs parallel to scepticism expressed in the German literature about the appropriateness of the “subtraction method”. A growing number of authors agree that the typical causality test has serious flaws (Roxin, 2006, Sect. 11, nr. 12–13, nr. 19; Jescheck, 1996, p. 282). These flaws become apparent in another example: a firing squad shoots at the same moment; two bullets hit the victim and cause mortal injuries at the same time. Applying the “subtraction method”, each of the two riflemen could say: the victim would have died anyway, through the other one’s fatal bullet, but it would obviously be wrong to say that neither of the two caused death. With an eye to such cases, it is preferable not to rely on subtractions but to look at the events as they happened.3 The same pertains to the picklock-case: in the real course of events, providing the tool was part of a chain which lead to the taking away of goods, and this is all that is needed to establish causation.

So far, it seems that John Gardner’s solution is identical with contemporary modifications of the old “causality test by subtraction”. However, there are differences in detail. First, I wonder whether Gardner’s reliance on the accomplice’s own practical reasoning is convincing. Subtractions and a hypothetical narrative would only be permissible for the aiding person, he argues, if she desires to avoid another course of events. In real life, very few accomplices will succeed according to this criterion. The motive for lending a picklock usually is to help the principal, not to save the victim’s windows. Gardner’s solution, however, leaves the puzzle of how to deal with someone who sincerely decides to hand out tools to burglars in order to spare the windows of houses. I would argue that such a motive is not relevant. The standards for attributing results to actions should be objective and not based on the motives of the actor.

Secondly, to adequately answer questions of attribution, it is not sufficient to determine causality. Causality is a necessary condition, but not a sufficient condition to attribute harm to accomplices and principals. Further restrictions must be satisfied. Let me point to a different case to illustrate this point. The burglar’s partner wants to cheer him up on the day of the break-in by preparing a sumptuous breakfast. The burglar indeed feels encouraged and departs in a good mood. According to Gardner’s line of thinking, the partner should not be allowed to subtract her contribution to the burglary (Herzberg, 1971, p. 7). As in some other parts of Gardner’s paper, it is not evident whether he argues as a moral philosopher or as a criminal lawyer. His rather wide view of responsibility might be adequate within a moral dialogue. For the purposes of criminal law, however, attribution requires additional filtering stages beyond causality. The burglar’s partner might be responsible in a moral sense, but it cannot be the task of criminal law to cover all factors that deserve moral blame. Rather, criminal law should be “ultima ratio”, restricted to the core of actions which are most closely related to the harm done. This requires a more discriminating assessment of attribution—an assessment that ought to be made from an objective point of view.4 The question is whether we (the collective made up by all citizens) want to use criminal justice resources or ignore those acts on the periphery of events.

To decide what acts are relevant for the criminal law, a comparison with hypothetical events seems useful. The test would be: is the hypothetical narrative (without the act in question) in its relevant elements the same as the real story, or does the hypothetical narrative differ? Relevant elements are those with significance in the criminal process. In a broad sense, not restricted to legal assessments, the partner’s support was relevant: telling “the whole story” would mention the lavish breakfast. But if the story is edited for the purposes of the criminal law, it is no longer “the whole story” and it does not matter what the burglar had for breakfast. Therefore, the partner’s encouragement was not relevant and she is not an accomplice to the crime. In the case of the borrowed tool, the hypothetical narrative would include smashing the window, whereas the real burglar used a picklock. This difference is relevant for the story that interests the criminal law. This story is not just “he got into the house and left with a TV”. The story is somewhat richer: how a burglar enters the building is a significant detail in police reports and in the description of facts in a criminal verdict. Therefore, the break-in was not only caused by, but can also be attributed to the lender of the tool in a criminal context.

Despite the foregoing remarks, I agree with John Gardner in his main argument: a causal contribution to the harm done is a necessary condition for wrongdoing, both for principals and accomplices (even though, in my view, an additional filter is needed to determine what can be attributed). Lindsay Farmer is sceptical for a different reason. He challenges the view that causality is a necessary condition for labelling someone as an “accomplice”. According to Farmer, it is not promising to focus on causality because the real problems cannot be grasped from this perspective. Rather, one should think in terms of “collective responsibility”. Responsibility, Farmer argues, is more than causality. For example, acts after the commission of a crime might warrant the judgement that the actor had participated in an earlier wrong. Although Farmer does not specify exactly how such “participation” works, it is evident that by acting today one cannot participate in an event that was completed yesterday. Farmer’s notion of “participation”, however, seems problematic. In penal politics, relying on vague concepts such as “collective responsibility” tends to fuel irrational reactions. Let me demonstrate my position with an example. During the last decade, we have witnessed constant increases in sentence ranges for offenders who distribute or collect child pornography. Behind this policy stands the notion that “somehow” such persons “participate” in the sexual abuse of children. But arguing like this obscures substantial differences: the point is that those who handle child pornography typically did not harm the child shown in the photo. Collectively they create dangers for other children because they support a market which creates demands to produce new child pornography. Similar arguments apply to receivers of stolen goods. The popular German saying: “Der Hehler ist schlimmer als der Stehler” means that the fence is worse than the thief. Again, the danger of imprecise thinking arises. The receiver of stolen goods is not an accomplice to the theft which is completed before he buys the goods. What he does is the same as buyers of child pornography: he becomes part of a market which stimulates futures crimes. There may be reasons why post crime behaviour should be punished, but one needs to examine carefully why this is so, and the result of this examination will be that danger is different from harm and that responsibility for future crimes is diluted. Penal sanctions should reflect that the wrongdoing is less serious than that of principals and accomplices who cause immediate harm. Putting everything under the banner of “collective responsibility” could veil these crucial differences.

Reasons for distinguishing between principals and accomplices and how to do it

Do we need to distinguish principals from accomplices? Lindsay Farmer suggests that it might be useful to abandon such attempts. Such proposals have been made before (Kienapfel, 1971), and some legal systems (for example, Austria) do not draw this contrast. From the viewpoint that both principals and accomplices have caused harm, one could consider the search for further distinctions to be futile. John Gardner differentiates between principals and accomplices. He argues that there are two different modes of responsibility: while responsibility for one’s own actions is direct, the responsibility for the actions of others is indirect. He does not cover the issue of sentencing, but one could deduce from his “two modes of responsibility” argument that accomplices should receive lighter sentences than principals. As a rule of thumb, this result would be defensible. In German law, sentences for those helping the principal (“Gehilfen”, that includes encouraging) must be reduced (Sect. 27 Penal Code). Matters become more complex with respect to those who are guilty of abetment. Under German Law, these persons must receive sentences equal to principals (Sect. 26 Penal Code). Following John Gardner’s concept of “indirect responsibility of accomplices”, it would not be convincing to punish an abettor like a principal. Either the rule in the German Penal Code cannot be justified, or the notion of “indirect responsibility” does not fit cases of abetment. I would argue that the issue depends on the exact nature of abetment. Sometimes, the abettor might be less blameworthy than the principal. But there are cases where the description “indirect responsibility” does not correspond with psychological facts. If the principal was hesitant and needed to be pushed towards the crime, the responsibility of the abettor is greater than in cases in which the principal avidly followed a mere suggestion. Thus for sentencing, the concept of indirect responsibility is too formal, and needs to be sensitive to empirical, psychological descriptions of pressure and control. If the details of a specific interaction are taken into account, it can be appropriate to punish the abettor like the principal in some cases but not others. Inflexible sentencing provisions like in the German Penal Code are not useful (Ashworth, 1999, p. 428; Kienapfel, 1971, pp. 31–37). Does that mean that one should (like Lindsay Farmer and others) adopt a broader concept of “offenders” without attempting to distinguish accomplices from principals? I would argue that this distinction, although not always conclusive for sentencing, still has value. Differences between inciters, principals and mere helpers are commonly drawn. What Gardner describes as “direct” and “indirect responsibility” is deeply ingrained in our way of evaluating past wrongdoing, and it would be misleading to abandon these categories in the criminal law (Maiwald, 1979, p. 343, 354). Descriptions of criminal events would be less precise if we had just one single category called “offender” that lumped together the main actors, the abettors, and all those who aided only in a minor way.

I turn now from the question “should we differentiate at all” to the “how”. John Gardner assumes that one can distinguish between principals and accomplices by analyzing different modes of causality. He suggests the following difference in the case of killing: if the causal route from certain human behaviour (acts or omissions) to the death is neither interrupted by a novus actus interveniens nor mediated through killing by another, then the actor (or the person guilty of omission) is a principal in the killing. Someone is merely an accomplice if he makes a causal contribution to the death in other ways. In his comment, Lindsay Farmer expresses doubts about whether the concept of causality can shoulder the load that Gardner places on it. I share these doubts. Not much is gained by invoking causality to decide whether someone is to be punished as a principal or as an accomplice. Causation is a necessary condition for both, but different criteria are needed to discriminate between principals and accomplices. A widely shared approach in the German literature assesses dominion or control over the offence (“Tatherrschaft”) (Roxin, 2003, Sect. 25, nr. 27–28; Kühl, 2005, Sect. 20, nr. 25–27). This notion of “dominion over the offence” is more complex than the question of causality. One needs richer ways of describing the psychological circumstances, and the various differences in the degree of knowledge and willpower that allow control over situations. Gardner’s concept of “refined causal contribution” works for some cases, but not for others. Imagine someone who makes a child or mentally ill person stab a victim. One should have no difficulties calling the man in the background the killer, not because of some “refined causal contribution”, but because he was the one who had control. Another typical problem is how to distinguish a co-principal from an accomplice if the second person involved in the crime helped substantially. Again, this difficulty can only be solved by measuring and comparing degrees of dominion.

The passive accomplice

A large section of John Gardner’s contribution covers passive accomplices. In his scenario, B would be an accomplice to A’s crime if he chooses to remain passive when “either B flies an airplane into a skyscraper, or, if he refuses, A will make fifty airplanes crash”. Underlying his position is a very strong moral claim: B has a moral duty to act. There are three problems with Gardner’s arguments. First, it is at least questionable whether “there is only one justified thing to do in such a case”, that is, to harm the smaller number of victims. Gardner would have received harsh criticism if he had presented his torture and skyscraper examples to a German audience of criminal or constitutional lawyers. Such criticism would not be connected to agent relativist or agent neutralist views, but to another point. Sacrificing the lives of, say, three hundred persons in the skyscraper would have to be justified to them. One would need to explain to each of them: you have to vanish now, because your life is worth less than that of ten or one hundred other potential victims. Such a quantitative comparison of the value of human lives is not obviously correct. Most German authors and the German Supreme Court maintain that these trade-offs are morally reprehensible.5 I will not dwell on this problem, however, as it does not directly involve the issues of complicity that concern us here.

Secondly, even if it is morally appropriate to kill three hundred human beings in order to safe many more, it is a long way to the general statement that “the avoidance of wrongdoing by anyone is fundamentally everyone’s concern”. This strong moral claim is very disputable. Even if a duty to act arises in genuine emergencies which threaten the lives of many persons, a different assessment is required in more mundane situations. Using (and disagreeing with) another of Gardner’s examples, I would argue that keeping a promise about the publication of a paper is the author’s business. At stake are both the author’s freedom and the freedom of others. The sloppy author has the right not to have everybody intervene—he has to accept criticism from the publisher and from a personal “moral community” (partners, close friends etc.). Still, it seems a strange proposition that one would be obliged to accept everyone’s interferences. Those knowing the author are morally justified in simply ignoring his minor (and also not-quite-so-minor) personal deficiencies. One is entitled not to pay attention to such wrongdoing simply because one prefers to do something else or because one prefers peace to confronting the other person. Gardner mentions the issue of freedom briefly, but argues that “freedom has little bearing on the morality of complicity” and that the amount of intrusion is limited only by the fact that it might sometimes be inefficient. If one takes this stance seriously, everybody would constantly be trying to be as efficient as possible in morally improving others. For those who do not wish to live in such a world, moral complicity must be limited to failures to prevent veryserious future wrongdoing by others.

Thirdly, even if one were to accept a general moral proposition that not acting in the face of future wrongdoing means to become an accomplice, the inferences for legal matters need to be spelled out. John Gardner does not deal extensively with this question. At one point, he mentions that “freedom is important in deciding how much of the morality of complicity we should institutionalise in the law”. The person Gardner sees as moral accomplice is one who fails to prevent others from doing wrong, which means that she could only be convicted for an omission. To punish omissions like acts, the law requires special circumstances creating special duties. A parent who does not try to avert a murderous attack on his child would be sentenced for his part in the murder.6 But special duties as those of parents aside, one could not be convicted for murder by omission as an accomplice in the “airplanes into skyscrapers”-example.7 And this must be so, as the blame for omissions is different from what is appropriate for acts. The person who swings the knife must be blamed for harming another person, for taking the victim’s life, or for making his life miserable due to the wounds inflicted. The person who does not intervene (although he could easily do so) deserves blame because she did not show sufficient attention, empathy or courage. German law (Sect. 323 c Penal Code) allows such blame to be expressed. This norm does not apply to those with special protective duties towards the victim or the duty to supervise the offender, but to everyone else who remains passive in the face of grave and imminent danger. But sentences are lenient; the maximum penalty in Sect. 323 c German Penal Code is one year imprisonment, and the usual sanction is a fine. In moral discussions about the existence of a duty, the “yes” or “no” answer is important. For purposes of the criminal law, graduations count. It is crucial that the omission to avert serious injuries is blameworthy to a much lesser degree than the behaviour of those who actively harm others. John Gardner does not tell us explicitly whether he opts for treating his passive accomplice as an accomplice not only in moral discourse but also in the criminal law. But it needs to be emphasized that not everybody who fails to prevent the wrongdoing of others should be sentenced as an accomplice to the crime.


In the German literature, this is the prevailing explanation, see Roxin (2003, Sect. 26, nr. 12–15, 26–27). See for a different theory Schumann (1986), who argues that accomplices express solidarity with the principal and thus disconcert the public. This is not convincing as it underrates the degree of responsibility.


In German criminal law, called the „conditio sine qua non“-formula, see Kühl (2005, Sect. 4, nr. 9)


This approach to determine causality is called “Lehre von der gesetzmäßigen Bedingung”. See Roxin (2006, Sect. 11, nr. 15) and Jescheck (1996, p. 283)


See for the concept of objective attribution (“objektive Zurechnung”) Roxin (2006, Sect. 11, nr. 44–145), Jescheck (1996, pp. 286–289) and Kühl (2005, Sect. 4, nr. 36–60).


The Bundesverfassungsgericht recently had to deal with a new legal provision permitting that the airforce shoots down a kidnapped airplane if it appears that the kidnapper will use the plane for an attack. The court declared this law unconstitutional, with the argument that to kill passengers and crew in the kidnapped plane in order to safe the lives of others violates their right to life and their human dignity, NJW 2006, 751, at 758. See also for similar arguments Roxin (2006, Sect. 16, nr. 38–42) and Jescheck (1996, p. 363).


Some even argue that parents’ duties are so important that failing to fulfil them makes the parent always a principal to the crime (Roxin, 2003, Sect. 31, nr. 140; for the contrary view—parent is accomplice—Jescheck, 1996, p. 696). Gardner’s argument is convincing: the parent who does not intervene against actions by others would be an accomplice.


On the contrary: if one would opt to fly the airplane to avoid a much bigger number of victims and survived, one would be most likely be convicted by a German court as conventional doctrine claims that this could neither be justified (see the citations in footnote 5) nor excused (excuse in the criminal law, Sect. 35 of the German Penal Code, calls for the intention to rescue oneself, one’s family or someone to whom one is personally close. Rescuing strangers is not even an excuse according to German law!).


Copyright information

© Springer Science+Business Media B.V. 2006