Cooperation, also with evildoers
Alexander R. Pruss
Baylor University
Working together with another on a good project can increase the chance of the project’s success or improve the quality of the product, decrease the amount of effort each needs to put in, help each achieve new knowledge, and even lead to a friendship. All of these goods can result from cooperation, but they are not intrinsic to the cooperation itself. These goods merely make the cooperation instrumentally good. Moreover, while these goods are not exactly incidental to the cooperation—to increase the chance of a project’s success is often though not always one of the points of cooperation—cooperation can also lead to a poorer or non-existent product, multiply the amount of work needed, teach us things that are false, and produce enmity. Many hands make light work but too many cooks spoil the broth.
However, there are also ways in which the cooperation is typically non-instrumentally good. In Nicomachean Ethics 9, Aristotle notes that cooperation with a friend allows us to be more continuously active in virtuous, and hence pleasant, activity. The idea seems to be that the activity of a friend is in a way my own activity—my friend is another self—and so if I am having a break and you are working on our common project, your activity is in a way mine by virtuous of the friendship, and hence I am in a way active even while taking a break, on account of your perseverance.
Although Aristotle is thinking of a case of pre-existent friendship, this can apply even absent prior friendship. You are working at a worthy project. I join you. Now your work becomes our work, and hence can be attributed to both of us. Previously it was non-instrumentally good for you to do the work, and now it is non-instrumentally good for us. I think the best example of appreciation of this kind of good is given by my three-year-old son, one of the greatest joys in whose life is to be a helper in the household tasks done by his parents.
One might here also query whether there isn’t a loss to the person who is cooperated with—is it perhaps a greater value to accomplish a task by oneself than with another? My son at times seems to think so! But even if there is a value in having a task “to one’s sole credit”, there is also a non-instrumental value in the cooperation itself: in the fact that people are working together. To share in a task is a form of interpersonal union. Being united with others is a basic human good, and in cooperation we are united with others precisely in respect of one of the features distinctively normative for our kind—agency. Insofar as one is cooperating with one or more others, one is part of a community joined by the task.
In fact, when the cooperation in the worthy task is sincere, mutual and intentional it by itself constitutes a kind of friendship or active love, and this fact justifies my use of Aristotle’s description of the cooperation in friendship. For the central features of love are an appreciation of the other, a pursuit of the goods and goals of the other, and a tendency towards union. And in sincere, mutual and intentional cooperation in a worthy task one appreciates the other as a fellow seeker after the same good goals, one pursues those goals together, and one is united with the other precisely in that cooperation. (One could build a whole sexual ethics out of considerations like this in the case of reproductive types of activity.) So cooperation in a good is itself a form of community and even friendship.
Moreover, in engaging in cases of cooperation, it is normal for various other human goods to be exhibited: communication, specialization, give-and-take, mutual assistance, and so on.
So far what I’m talking about sounds like a romantic paean to the instrumental and non-instrumental goods involved in cooperation. But bear with me: there is a dark side. So far I assumed that the task cooperated in is a worthy one. What if not? Is there a value in cooperation in murder, robbery, exploitation, slander, avoidance of facing the issues of life, and other evils? On the one hand, it seems like some of the values I have listed earlier are still found when the task cooperated in is morally reprehensible. We enjoy heist movies in large part precisely because of this.
On the other hand, most of us think that even though it may not be desirable always to keep our hands clean, fastidiously avoiding the least contact with evildoers (an attitude plainly impossible for those of us who are ourselves sinners), still there is a presumption against cooperating with evildoers’ wicked deeds. We thus have a certain tension here: on the one hand, there are always values in cooperation, and on the other, there is a presumption against such cooperation when evil is involved.
In the first part of the paper, I want to discuss both the values of cooperation in an evil task, as well as the presumption against this cooperation. I then want to move from these general ideas to a discussion of what counts as cooperation, and the kinds of considerations that a virtuous agent needs to take into account.
During the course of the discussion, I would ask you to bear in mind six different cases of involvement in evil, most of them taken from real life:
(1) Selections: Consider the activity of a physician who participates in concentration camp selections, choosing who is too sick to work effectively and hence should be gassed, and who should stay alive. It is worth noting that this physician is may later justify his activity at Nuremberg by saying that if he didn’t do it, someone else would have, and this “someone else” would have chosen more people to die.
(2) Hypothermia Research: Here, consider the activity of a late 20th century researcher who is considering making use of data obtainAed by Nazi researchers who induced often lethal hypothermia in non-consenting prisoners in order to generate data that would be helpful to Nazi military practice. The data might well be useful to us now: for instance, it might allow one to figure out what methods of reviving patients are more likely to be effective.
(3) The Kidney: George has murdered Patricia for her kidney. The murder is not prosecuted, because Patricia is a member of a persecuted religious group. You are the doctor, and George asks you to implant Patricia’s kidney in him. You know that Patricia is a good genetic match for George, and moreover there is no one else presently in your locale who needs a kidney.
(4) Alcoholism: Consider a taxi driver who takes a fare she knows to be an alcoholic to the bar, as requested. What is interesting in this case is that the evil involved is one for which the agent cooperated with—the alcoholic—may well be entirely inculpable, and where this evil is primarily done to the alcoholic herself. A variant of this case is where instead of a taxi driver, a friend is asked for a ride to the bar.
(5) Pavement: Here, consider whether it is appropriate to walk on a Roman pavement originally laid by slaves centuries ago.
(6) Police Officer: The livelihood of the typical police officer is tied up with crime and evil. Consider, then, the case of a police officer who derives a net benefit from her work for herself and her family. Here, criminals and the police officer are in some sense jointly bringing this benefit about.
The values I discussed in the case of cooperation in a good activity can be divided into the instrumental and non-instrumental. There is an obvious instrumental bad about many cases of cooperation with an evil activity, which is the reverse of the instrumental good of cooperation with a good activity: the cooperation with evil increases the likelihood of the success of the evil activity.
One might think that this consequentialist consideration is by itself sufficient to account for our intuition that we should avoid cooperation with evil activities. But, perhaps surprisingly, that is not so. Take the Selections case, and suppose that the physician indeed knows that someone else in his place would be worse. If the source of our intuition that we should try to avoid cooperation with evil activities were the consequentialist consideration, then our worry about the activity of the physician would disappear as soon as we found out that the consequences of his participation in the murderous activity are actually good. But that is, of course, not so. While some of us might in the end think that the Nazi physician is justified in cooperating in selections, nonetheless it is clear that there is a presumption against such cooperation. If there is another course of action with equally good results but not involving such cooperation, then even if that course of action is less convenient, it is clear that the physician ought to take it. So the instrumental considerations against cooperation are not sufficient to explain the presumption.
On the other hand, a number of the non-instrumental goods of cooperating in a worthy task flip over into non-instrumental bads when the task becomes bad. Just as, given an Aristotelian view of welfare, to act well is surely a part of one’s welfare, so to act badly is a part of one’s illfare. Cooperation lets me share in the task, partake of the value of participation in it; this is good when the value is positive and bad when the value is negative. If the task is good, then it is a good thing for me to enjoy the fact that co-participants are active even when I am unable to be; but if the task is bad, then even when I am idle, a co-participant may be working to increase the evil I am responsible for. This is bad, at least for me.
I suggested that cooperation in a good activity was a kind of active love. This may still be true in the case of a bad activity when one is ignorant of the badness. But the practical question of cooperation with evildoers only comes up when one takes the activity to be a cooperation with evildoers.
So let us assume that I know the activity is evil, but nonetheless cooperate. Then a number of the aspects of friendship entailed by cooperation in good disappear. I no longer appreciate my co-participants as pursuers of the good, but instead take them to be pursuers of the bad, though I may hope that they are not culpable by reason of ignorance.
I do promote the goals of the co-participants, but I take the goals to be bad. If my co-participants also believe the goals to be bad, then I am helping them to be false to themselves—and that, surely, is an act opposed to friendship and love. But if my co-participants erroneously take the goals to be good, then while I do cooperate with them, I do so disunited from them. Moreover, there is some sense in which I am actually working against them. For they are trying to pursue the good, albeit by mistake ending up doing something wrong. But I am consciously pursuing the bad. Thus the agreement as to goals is in a way only apparent. Maybe they as sincere Nazis are pursuing “the good of racial purity”, but I cannot be promoting any such thing, since I know that racial purity is not a good.
In helping my co-participants in bad activity, I harm them. For it is surely a bad thing to be responsible for evil. A quick thought experiment: You wake up with temporary amnesia. You know it’s temporary because of a note on your desk. Based on your image in the mirror and the documents on your desk, you realize you are one of two twins—one of whom has a chronic illness causing severe pain in the evenings and the other of whom has murdered five people and gotten away with it (maybe the documents include a confession written after an acquittal on technical grounds). Soon your memory will return. Surely you hope you’re not the murderer—it is non-instrumentally bad to have murder on one’s conscience, and indeed it is worse for one to be responsible for a murder than to be responsible for an attempted murder (this claim is independent of culpability-related moral luck considerations). Likewise, friends don’t let friends become murderers, and thus insofar as I help my co-participants, I harm them.
It is an illusion that one helps create a valuable community by deciding to help evildoers. If the evildoers know that they are doing wrong, then we have a community of people helping one another to become worse. If they do not know it, then the cooperator who does know that the activity is bad is not really a member of the community in the full sense, since the other members are acting for a good that she is not pursuing.
Thus, cooperation in an activity one takes to be evil is not an act of friendship in the way that cooperation in a good activity was. In some ways it is the opposite. For those of us who want to base ethics on love, and who take friendship to be a kind of love, when we see that something is opposite to love, we therefore conclude that we have reason to avoid it.
Nonetheless, such cooperation may still involve some non-instrumental goods, such as the exercise of executive virtues and skills like perseverance and communication. But just as the fact that a given act would be an instance of perseverance gives one no reason to do it when one knows the act to be wrong[1]—the wrongness of the act excludes such reasons—so too the wrongness of the act cooperated in makes goods like communication not serve as reasons for the action.[2]
But what I said above applies fully only in full-blown cases of cooperation. There could be kinds of cooperation where the evildoer’s responsibility for evil is not increased by our contribution, and where one does not seem to acquire responsibility for particular evils. Cooperation after the fact is often like that—the Kidney case in particular exhibits this feature. The surgeon is helping George achieve his goal of getting a new kidney, but George does not seem become responsible for more evils as a result of achieving this goal, nor does the surgeon become responsible for any additional evils—she simply saves someone’s life (admittedly a guilty person’s, but surely a physician should be willing to treat a person sentenced to life in jail). However, I think we all share the intuition that one should at least be reluctant to act in the Kidney case, and not just for consequentialist reasons like that putting in the kidney encourages others to commit murder.
We now see that there are multiple reasons not to engage in full-blown cooperation with an evil activity, independent of the consequentialist consideration that doing so may increase the probability of evil: we can see the flip side of many of the goods in cooperation with a good activity.
But now I want to focus on one consideration that has not yet shown up. We feel that there is something fair and just when someone who has done good receives good as a result and when someone who has done something bad receives bad from it. We are particularly pleased in cases where the bad received matches the crime and is caused by the crime. Patricia the house robber breaks a window, goes into the house, accidentally smashes the expensive TV, and then realizes that by mistaken she had broken into her own house. We say this is poetic justice. Dr. Jones plagiarizes a paper, and then the original author ends up appointed as her tenure referee. Again, the punishment flows out of the crime, turning the expected benefits on their heads.
We enjoy poetic justice, and the enjoyment is not mere schadenfreude. In poetic justice, the designs of the evildoer are not merely frustrated but turned against themselves. Poetic justice frustrates bad plans and rewards good ones, in ways that match the punishment or reward to the crime, and that make the punishment or reward flow from the crime. On the other hand, there is something particularly bad—an injustice—when an evildoer gets away with exactly what she wanted and when this turns out to be everything she expected.
Sometimes justice happens without our intervention. But in some cases society is a causal intermediary between the deed and its fruit. It is important that punishment not merely be deserved, but that it be caused by the crime—if an unjust judge chooses a random member of some unfortunate ethnic group to be imprisoned, and the person unbeknownst to the judge is a criminal, justice has not been done to the criminal. What justice calls for is some kind of a reversal of what the criminal had wished for, and calls for that reversal to be a result of the criminal’s actions: the mere non-fulfillment of her plan is insufficient for justice.
But the non-fulfillment of her plan is a start towards justice. If we have a justice-based reason to produce some kind of reversal of what the criminal wanted—she wanted a comfortable home as the fruit of her bank robbery and gets a sparse cell, worse than her conditions before her crime—a fortiori we have a justice-based reason to ensure that an evildoer’s plans are at least unsuccessful. I wish I had an argument beyond the intuitions I cite here, but right now I just rest on the intuitive plausibility of the claim that justice calls on us at least to frustrate an evildoer’s plans.
In cooperating with an evildoer, however, we further her goals, and thus we are working in a way that hinders what justice calls for. The cooperation helps the evildoer get a reward—the fulfillment of one’s goals is indeed a kind of reward—where instead a punishment is due. Our earlier considerations against cooperation with an evil plan were based on the badness for the cooperator and the primary agent of the evil activity, and were based on charity for self and others. The present considerations strengthen the previous by adding a properly justice-based consideration. And as the flip side of this justice-based reason not to cooperate with an evil plan, we have a justice-based reason to cooperate with a good plan: we help reward the good action by bringing it to the fulfillment that the primary agent intends for it.
In full-blown cases of cooperation, there are both charity- and properly justice-based reasons why one shouldn’t cooperate with evildoers in the completion of their plans. The justice-based consideration, if you accept my intuition, is present whenever our action contributes to the fulfillment of any part of the evildoer’s evil plan, even when the part of the plan that we are contributing to is good. In the Hypothermia case, by using the results of the Nazi research one is contributing to the Nazi doctors’ plan of using non-consenting patients for the progress of science, even though one is not contributing in any way to the harm done to the patients. Nonetheless, one is acting in a way that hinders justice since justice calls for the frustrating of the evildoer’s plan.
[Slide] In a very insightful article, Kathleen Kaveny has distinguished cases where one draws a benefit from a wrongful action, which she calls “appropriation”, from cases where one actually helps in the wrongful action, which she calls “cooperation”. I think this way of cutting up the logical territory is mistaken, however. It lumps together the Police Officer, Hypothermia Research, Kidney and Pavement cases, all of which involve the drawing of a good from an evil. But the case of the Police Officer is plainly different from the other three: the fact that the livelihood of a police officer depends on the existence of crime does not produce any presumption against being a police officer. And in the Hypothermia Research, Kidney and perhaps Pavement cases, one is not only appropriating, but one is acting in the way that the evildoer meant one to—one’s action is a part of the evildoer’s plan, one is a pawn of the evildoer. In these cases, one is actually cooperating.
[Slide] Thus, a better way to cut up the cases seems to be as follows. First, we distinguish cases where one interacts with the action plan of an evildoer, perhaps with a benefit being drawn from that, but one does not actually contribute to any aspect of the evildoer’s plan. The Police Officer case is paradigmatically like that—one interacts with the criminal contrary to the criminal’s plans—and there is intrinsically no presumption against this. Admittedly, there may be certain moral dangers in being too close to evildoers. Perhaps their example will morally deleterious. (But one shouldn’t overestimate this. We are all sinners anyway.) Perhaps the fact that one is benefiting from the evil will make one less assiduous about trying to eradicate the evil or may lead one to even contribute to it (in kind of the way that fire fighters have been known to start forest fires in order to give themselves employment). But just about every morally good action has moral dangers (at least, that of self-righteousness!)
So the only cases which we need to see as intrinsically morally problematic are ones where one contributes to the execution of the evildoer’s plan. In such cases, there is always a consideration of justice against such contribution, and often other considerations as well. These are cases of cooperation, as opposed to the mere appropriation of the Police Officer case.
Whether the Pavement case is one where one contributes to the execution of the slave owners’ plan or not will depend on what one thinks the slave owners’ plan was—was it just to lay up a sidewalk for centuries, or was it a shorter term plan? This fact shows that in cases of drawing a benefit from an evil, one’s temporal distance from the evil can matter, because it is less likely that one’s action falls in line with the evildoer’s plan if one is more temporally distant from the evildoer. It is typically the case that action plans trail off into indefiniteness as one goes further into the future. Using items made by slave labor thousands of years ago is, thus, less problematic than using such items made yesterday, because one’s use of the items made yesterday is much more likely to have been a part of the evildoer’s plan. (An additional consideration in the Pavement case is that if the slavery is in the past, by refusing to make use of the product of slave labor, one is making the labor meaningless, and hence one is in some way contributing to the devaluation of the slaves.)
We can now subdivide the kinds of cooperation in three different ways. One kind of subdivision is in terms of the part of the evildoer’s plan that one contributes to. One might contribute in such a way that one is furthering the evils in the plan. Or one might contribute in such a way that one is furthering the non-evil parts of the plan. In the Selections and Alcoholism cases, one is furthering the evils—one is choosing who is to be murdered or driving the alcoholic to where the drunkenness will take place. In the Hypothermia Research and Kidney cases, one is furthering goods—it is good for treatments for hypothermia to be found and it is good for lives to be saved (even, I say, lives of murderers).
Obviously, it is more problematic to cooperate with the evil parts of the evildoer’s plan. In such cooperation the considerations of charity towards the evildoer and of charity and justice towards the victim will be paramount. However, if I am right, cooperation in respect of the good or neutral parts of the plan is also problematic in respect of justice.
We can, of course, subdivide further based on other features of one’s cooperation in respect of the actual causal contribution to the plan, such as whether it is necessary (would someone else be found if one refused), whether it is significant (maybe only slightly less evil would result if one didn’t help), etc. All of these will be ways of classifying the cooperation based on what one contributes to and the degree to which one contributes to it.
A second way to subdivide cooperation is in terms of the cooperator’s intentions. What is one trying to accomplish? Here, there seem to me to be three primary options.
(a) One might be intending to contribute to the evil goals of the action. Here, one is fully a part of the evil action—one is doing evil just as the other evildoer is, and what one does is wrong for the same reason. One is aligning one’s will with evil, setting oneself on a bad road. The Catholic moral tradition calls this “formal cooperation”.
(b) One might be intending to contribute to the successful fulfillment of the agent’s plan as such but without specifically intending for the evil goals to be fulfilled. This is a somewhat odd possibility, but I am thinking of a case, for instance, where you don’t intend the evil parts of a friend’s bad plan, but you act in the hope that her plan will succeed, perhaps cooperating after the evil has been done, for instance because you like your friend and want the plans of friends to succeed. Intentions are intensional, and so you could cooperate in a plan under the description “murderous plan”, which would be the previous case where one intends the bad part of the plan, or you could instead cooperate in a plan under the description “friend’s plan’, which is the present case. But justice calls for the frustration of evil plans, and in cooperating even under the description “friend’s plan”, one is intentionally doing something that is opposed to justice. This seems to be simply wrong, for much the same reason that it is wrong to intentionally reward an evildoer. One can sometimes gauge whether this is what one is intending by asking appropriate counterfactual questions. Supposing that one can either use data from Nazi hypothermia research or use data from some unproblematic source, would one still opt to use the Nazi data? If so, then it seems likely that fulfilling the Nazi plans is a part of one’s intention. If not, then it’s likely not.
(c) Finally, one might not intend either the evil in the plan or to help the evildoer fulfill the plan. Rather, one might have an intention to do some action, say to implant a kidney, which as it happens contributes to the evildoer’s plan. Likewise, if instead of actually designating who is to die, one were to provide a medically correct negative assessment of the health of a concentration camp inmate, knowing that the inmates who get such assessments are killed, but not intending this killing (for instance, one might simply enjoy doing’s job as a physician and evaluating patients’ health), then one’s cooperation in the killing is of this last sort. If something like the Principle of Double Effect holds [slide], then this kind of cooperation will be permissible if there is sufficiently good reason for it. One is contributing to the evildoer’s plan, but one does not intend to do so, and it could be that the goods one achieves are proportionate to the bad of furthering the plan. The Hypothermia Research, Kidney and Alcoholism cases are like that. Whether the Selections case can be classified here is a difficult question, and may depend on finer details of the situation than I gave. The Catholic tradition calls this “material cooperation”.
The distinction between the kinds of cooperation here depends crucially on intentions, and hence opponents of the Principle of Double Effect are likely to reject it. However, there is intuitively a reason to accept something like this distinction independently of Double Effect. There is a presumption against cooperating in evil plans. It seems plausible that the more intimate the cooperation, the stronger the presumption. And how intimate a cooperation is is surely, at least in part, determined by which if any of the evildoer’s goals one shares.
There is a final distinction which also gets at how intimate one’s cooperation is, but this time in terms of the evildoer’s intentions, or more precisely in terms of how the cooperator figures in the evildoer’s plan. There is a difference between a plan presupposing a state of affairs and the state of affairs being a part of the plan proper. An action plan specifies what one is accomplishing. If something is a part of the plan proper, it is something one is trying to accomplish. On the other hand, there may be prerequisites for a plan that are beyond one’s control. If an assassin lures a victim to an isolated location, then the victim’s going to that location is a part of the plan proper—it is something that the assassin accomplishes if the plan is successful. This example shows that the actions of others can be a part of one’s plan. If, on the other hand, the assassin lies in wait on the victim’s usual route, then the victim’s being there is presupposed but not accomplished by the evildoer.
Being a part of the evildoer’s plan is a more intimate involvement in the plan than simply being presupposed by the plan. It is a part of the alcoholic’s plan that the taxi driver take her to the bar, and the taxi driver’s actions are responsive to the actions of the alcoholic. If, on the other hand, we replaced the taxi by an automated rapid transit system that has to be turned on every morning, then the action of the person turning the system on is presupposed by an alcoholic going to the bar, but the action is not a part of the plan. One doesn’t turn the system on for the sake of the alcoholic—one would have turned it on anyway. Similarly, it may be a part of the Nazi medical researcher’s plan that future researchers make use of his results.
There is good, though defeasible, reason not to allow oneself to be a part of the evildoer’s plan. Besides the fact that one has a justice-based reason to be frustrating that plan, one also has a good, though still defeasible, reason of self-respect for this: to be an evildoer’s pawn is contrary to one’s dignity. But, again, the Principle of Double Effect may allow one to act in a way that makes one be an evildoer’s pawn if one does so without intending to be such a pawn, either as an end or as a means, and if one acts for a sufficiently great good.
The closer a form of cooperation is with an evildoer’s project, and especially with the evil parts of the project, the more reason there is to avoid that cooperation. How close a form of cooperation is depends on several factors such as how essential the cooperator is to the project, how much the cooperator’s intentions match up with those of the evildoer, and how much the cooperation was a part of the evildoer’s plan. Any form of cooperation is morally problematic—there is always a justice-based reason to avoid furthering the plans of an evildoer.
In cases where one intends the evil done, one does wrong just as the evildoer does. And when one intends to help the evildoer succeed in her plan, one goes against justice. But apart from these two cases, it seems that the considerations against cooperation are defeasible. One argument for defeasibility is the Principle of Double Effect—given sufficiently good reasons, it is permissible to act in a way that will produce an unintended bad.
We can also see the defeasibility by thinking about cases. If a murderer points a gun to my head and asks me to drive her to her hide-out, it is not wrong for me to cooperate in this aspect of the plan, although it would have been wrong for me to murder someone on her behalf even if she were holding a gun to my head. Moreover, in any large organization, we will have co-workers who at some point employ means that we consider ethically unacceptable—for instance, I would be surprised if people in Publicity Departments didn’t make epistemically unwarranted statements on occasion or if some co-workers didn’t excessively sacrifice their families for the sake of their work. By working for the same organization, we are cooperating in these co-workers’ larger plans to further the organization’s goals, and this cooperation is not wrong when the evils done by these co-workers are neither too great nor essential to the organization’s functioning, when we do not intend these evils, and when our own cooperation with these particular co-workers is not too close.
A requirement to engage in no cooperation with evil at all would paralyze us. In fact, this would be a requirement which would be pretty much impossible to fulfill. We can be pretty sure that when we purchase anything, some of the money we inject into the economy will eventually be used by somebody for some evil.
So, in the cases where one does not intend the evil or the furtherance of the evil plan, the question becomes one of weighing. As we have seen, there are several factors we can take into account when weighing how strong a presumption there is against cooperation with evil, and these then need to be considered against the goods resulting from the cooperation.
I do not think there is any general rule that would determine which forms of cooperation are permissible and which are not, absent the wrong intentions. Some cases will be easy. There is little wrong with making good use of the results of slave labor from thousands of years ago—quite possibly, this is not a case of cooperation at all, since the exploiters did not really fit us into their plan, but even if the exploiters were so foresightful, our own role in the plan is miniscule. But the Hypothermia Research, Kidney and Alcoholism cases all strike me as difficult. I have no recommendation for how to decide such cases except to become an Aristotelian virtuous agent—a phronimos—and then, when in such situations, to come to a virtuous decision, ideally with the help of the Holy Spirit.
However, this does not render what I said earlier useless. For the account I gave also lists some of the facts the phronimos should take into account when coming to a decision, facts about how closely one is cooperating with the evil plan.
I want to end, however, with a cautionary note. What I have discussed were the intrinsic bads of cooperation. There are also extrinsic dangers. One may be a bad example to others. A courageous stand against the evil plan may sometimes be more appropriate than a careful figuring out of just how much cooperation is permitted. After all, as Kaveny has noted, there is a serious danger of self-deception. One may say that one isn’t endorsing an evil plan one is cooperating with, but over time, one’s protests (if any) may grow faint, and so one may come to endorse the plan in all but word and identify with the evildoer with whom one is cooperating. One can imagine a hypothermia researcher looking at Nazi data, being drawn in, exclaiming over the cleverness of an experiment, while forgetting that the lives that this cleverness cost. It is quite possible to go from disapproval, to praise of technique, to tacit approval.
This consideration of self-deceit and our own moral weakness suggests that we should, as the Talmud says, “make a fence around the Law”: unless there is strong reason to cooperate with evil, we should avoid even those forms of cooperation that are intrinsically defensible.