Author Information: András Szigeti, Linköping University, email@example.com
Szigeti, András.”Seumas Miller: Joint Epistemic Action and Collective Moral Responsibility—A Reply.” Social Epistemology Review and Reply Collective 4, no. 5 (2015): 14-19.
Please refer to:
- Miller, Seamus. “Joint Epistemic Action and Collective Moral Responsibility.” Social Epistemology (2015): 1-23. doi: 10.1080/02691728.2014.971908.
Image credit: Ken Douglas, via flickr
In a series of books and articles, Miller has developed a refreshingly original and complex account of joint action and collective responsibility. This approach constitutes an interesting alternative to the current orthodoxy that seeks to explain shared agency in terms of joint intentions. Miller also offers a novel, moderately individualist conception of group responsibility steering clear of both robust collectivism, according to which group-responsibility does not reduce to the responsibility of individual group members, as well as more radical forms of individualism, according to which collective responsibility is always just the sum of the responsibility of individual group members.
His present paper extends this account to the area of collective epistemic action.  I believe the approach is promising overall and its application to epistemology fruitful. In what follows, I will explore how the general account and its application could be further strengthened by making some of the central conceptual distinctions of the paper clearer.
Varieties of Responsibility
Miller’s distinctions among different kinds of responsibility, i.e., partial vs. full; partial vs. diminished; individual vs. collective, etc., are not sufficiently well-defined. I do not see the objections presented below as lethal. In fact, some of the definitional infelicities could be quite easily resolved, especially the unclarity surrounding Miller’s concept of partial responsibility. Let me go through the problematic distinctions one by one.
Partial vs full responsibility: Miller argues that individuals can have full (moral) responsibility for a joint action and/or its outcome even if the individual contribution to the joint action or its outcome was neither necessary nor sufficient for the action or its outcome to happen (12). By contrast, responsibility is partial if the individual “only makes a very small and incremental contribution as a member of a very large set of agents performing their actions over a long period of time” (13). In other words, neither full nor partial responsibility presupposes that the contribution is necessary or sufficient. So what exactly is the difference between them? Is it just the number of agents involved? Note that, as I will explain below, this is not just a worry about vagueness. The failure to clearly distinguish partial from full responsibility raises more serious conceptual worries as well.
The first problem, then, is this. Is there not a categorical difference between contributory actions which are necessary and/or sufficient for a collective action or outcome to occur, on the one hand, and those which are neither necessary nor sufficient for a collective action or outcome? Surely, contributory actions that are causally necessary (tricky cases of causal overdetermination notwithstanding) and/or sufficient for something to occur are salient both from an explanatory and from a moral point of view. After all, ascriptions of responsibility, and especially ascriptions of moral responsibility, are supposed to be action-guiding.
This is not to say that it is always permissible to perform an action that is neither necessary nor sufficient for some harmful outcome to occur.  However, when I deliberate about what I should do, I am morally required to pay particular attention to prospective courses of action of mine which could be necessary and/or sufficient causes of harmful outcomes. If, as Miller suggests, agents can have full responsibility for harmful outcomes even though their contributions to that outcome are neither necessary nor sufficient for it, then I am not sure whether we will have the resources to mark this categorical explanatory and moral difference.
The second reason for insisting on this categorical difference is that if we could be fully responsible for contributions to outcomes that are neither necessary nor sufficient for that outcome, then it is unclear why we need to bother with collective responsibility in the first place. Perhaps the strongest argument for ascribing responsibility to groups points to cases where each participating individual only makes a small and marginal contribution in a large group over a long period of time (think of global warming, for example). In such cases, when we add up the responsibility of each individual agent for his/her respective contributory action, it seems that the sum will not fully absorb responsibility for the harm produced by the collective action.
Therefore, there are grounds for arguing that we have to proceed in a top-down manner, and ascribe to individuals joint or shared responsibility for being part of a very harmful whole. (Such a top-down approach could be pursued even if we reject the idea of what Miller (9) calls the “corporatist”, i.e., non-distributive, collective responsibility.) However, if we can ascribe full responsibility for contributory actions by individuals even when these contributions are severally neither necessary nor sufficient for the relevant outcome to take place, then it is unclear what need there is for the concept of joint responsibility of individuals. And, a fortiori, it is unclear what need there is for the concept of collective responsibility since Miller defines collective responsibility as the joint responsibility of individuals for collectively brought about outcomes (9).
I now come to the third worry about the partial vs full responsibility distinction: Miller adds that agents can incur full responsibility even for small contributions provided certain institutional arrangements are in place: “Indeed, as a consequence of institutional arrangements put in place to deal with some collective action problem, each agent might have full moral responsibility (jointly with others) for some adverse outcome O—notwithstanding the fact that each only made a very small causal contribution to the outcome” (12). But the reference to institutional arrangements is misleading here: institutional arrangements have nothing to do with the distinction between partial and full responsibility.
Suppose (i) that an individual “only makes a very small and incremental contribution as a member of a very large set of agents performing their actions over a long period of time”; but also (ii) that the small contribution is forbidden by some institutional arrangement that applies to it (e.g., performing that sort of action is specifically proscribed by law). Now in this case there will indeed be something that agent is fully responsible for, namely doing something forbidden. But she will still be only partially responsible for contributing to the joint action or its outcome. Thus in the “screw-thief” example (13), the thief who stole a screw will still only be partially responsible for the sinking of the ship, while he may be fully responsible for breaking the law. For note that he would still be fully responsible (and blameworthy) for breaking the law even if doing so did not lead to any loss of life because nobody else stole any screws.
In addition, Miller’s reference to institutional arrangements contains another confusion as well. He claims that since the law is morally justified, “it is evidently morally justified that each screw-thief is also fully legally responsible for the loss of life, notwithstanding that his causal contribution to the sinking might be minute. This being so, it might be further argued that each such screw-thief is also fully morally responsible for any loss of life” (13). But even if it is true that the law is morally justified and that because of this it is morally justified to hold each screw-thief fully legally responsible for the loss of life, it does not follow that each screw-thief is fully morally responsible for the loss of life. Again, what each screw-thief is fully morally responsible for is breaking the law (if the law is morally justified), but not for the loss of life since each thief’s causal contribution to the loss of life is by assumption minute. (Note that there is nothing joint about the relevant action—i.e., breaking the law—for which the agent is fully responsible in this case.)
Partial vs diminished responsibility: As noted, Miller says that partial responsibility is incurred when the agent “only makes a very small and incremental contribution as a member of a very large set of agents performing their actions over a long period of time” (12). But he also states elsewhere that “if there are additional intervening (individual or joint) actions then those jointly responsible for the joint action in question, and its outcome, may have diminished moral responsibility” (11).
Even if we ignore all the worries mentioned in the previous section about partial responsibility, it is difficult to see the need to distinguish between diminished responsibility and partial responsibility. The only difference I can make out is that in the case of diminished responsibility, the involvement of other agents is not simultaneous, while in the case of partial responsibility it is. But should the mere fact that contributory actions take place diachronically rather than synchronically make any difference as regards the type of responsibility involved? And in practice how frequent are genuinely simultaneous contributory actions?
Collective vs individual responsibility: Miller argues as follows: “The agents are collectively (naturally) responsible for the realization of the (collective) end which results from their contributory actions. Consider two agents jointly lifting a large computer onto a truck. Each is individually (naturally) responsible for lifting his side of the computer, and the two agents are collectively (naturally) responsible for bringing it about that the computer is situated on the truck” (10).
However, the reference to collective responsibility is redundant because a straightforwardly individualist analysis is possible here. Agent1 has the end to put the computer on the truck, or perhaps she has the end to put the computer on the truck provided Agent2 also has the same end. And conversely, Agent2 has the end to put the computer on the truck, or perhaps he has the end to put the computer on the truck provided Agent1 also has the same end. If Agent1 fails to cooperate and as a result the computer does not end up on the truck, then Agent1 is individually responsible for this, and there is no occasion to ascribe responsibility to Agent1 + Agent2 together. By the same token, if Agent2 fails to cooperate and as a result the computer does not end up on the truck, then Agent2 is individually responsible for this, and there is once again no occasion to ascribe responsibility to Agent1 + Agent2 together.
If both fail to do their part, then Agent1 is individually responsible for doing her part, and Agent2 is also responsible for doing his part. In this case, ascribing responsibility to Agent1 + Agent2 together is redundant because once we have ascribed responsibility to Agent1 for failing to do her part, and have also ascribed responsibility to Agent2 for failing to do his part, then there is nothing left to ascribe responsibility for.
In responding to this worry, Miller provides the following explanation: “If, on the other hand, each [i.e., Agent1 as well as Agent2] has the end [to place the computer on the truck] interdependently with the other and both try to realise it but, nevertheless, fail due, let us assume, to their lack of sufficient combined strength then they are collectively responsible for failing to achieve their collective end” (11). It is unclear why lack of strength is something one can be responsible for, but we could say that Agent1 and Agent2 fail to realize the collective end because they fail to agree on how to lift the computer onto the truck (for a similar example, see Petersson 2008, 258). I still cannot see though (pace Miller and pace Petersson) why we need to invoke the concept of collective responsibility here. Agent1 is individually responsible to the extent the failure to agree was due to his (culpable) stupidity or stubbornness, and the same applies to Agent2. In short, the response merely changes the ground of individual responsibility for the failure of the collective enterprise; it does not transform individual responsibilities into collective responsibility (whereby collective responsibility would be something more than the aggregate of individual responsibilities).
There are of course several ways to respond to this individualist worry. As noted, perhaps the strongest (non-corporatist) response is to argue from cases in which individual contributions are not culpable when considered separately from a bottom-up perspective – not culpable because the contribution is either marginal (therefore neither necessary nor sufficient) or because it is individually rational (as in tragedy-of-the-commons type cases which can but need not involve marginal contributions). Such cases call for a top-down approach that takes the responsibility of the group for the harmful outcome as its point of departure. This kind of responsibility would indeed differ from individual responsibility, and perhaps it is the type of responsibility we should reserve the concept of partial responsibility for. As we have seen above, however, in order to be able to make use of such scenarios it would be important for Miller first to clarify the distinction between partial as opposed to full responsibility.
Epistemic Action and Epistemic Responsibility
Now to the paper’s central thesis: “The significance of the argument in this paper, if it is successful, is that individual participants in collaborative epistemic actions can be held morally responsible for these actions in more or less the manner as they can be held morally responsible for collaborative non-epistemic actions” (2). I do not have any direct objections to this, but would like to home in on a few issues that are raised by this claim and which may fruitfully be addressed in future research.
One possible reply to the thesis is addressed head-on by Miller. This is the worry that (some) epistemic states (e.g., beliefs) are not voluntarily acquired and so we may not be morally responsible for them (14-16). Some would presumably say that the parallel with non-epistemic actions does not hold up because, while we can admittedly undertake certain actions to acquire or not to acquire certain beliefs, as Miller says, once we have undertaken those actions the resultant belief will be inescapable. I may or may not perform certain calculations, but once I have performed them, the belief regarding the result of those calculations is not something I can choose not to have. And in this respect, beliefs are different from actions because even if there are compelling reasons for some action of mine I can choose not to perform that action.
But I would like to focus on another issue here. Some critics may argue that there are two ways in which an epistemic outcome can be morally significant. First, an epistemic action (individual or joint) can be morally significant if it unjustifiably harms someone. The point is that the harm caused by an epistemic action can be unjustifiable even if the epistemic action does not violate any epistemic norm. There are many situations in which telling the truth or not ignoring a piece of evidence causes harm. There is nothing special about epistemic actions here in the sense that we do not have to be concerned with the fact that the action was an epistemic action to assess whether the harm it caused was justifiable.
The case of the scientist who communicates her dangerous discovery to terrorists is an example of this kind of harmful epistemic action. The fact that the action is epistemic is incidental to its moral assessment in this case. The scientist helps the terrorist and, as it happens, she does this by communicating an important piece of information. But the scientist could have helped the terrorists by giving them a lift in her car to the lab or in many other ways.
But there may be other cases in which, arguably, the failure to comply with epistemic norms itself causes harm. (In fact, some may argue that we are morally responsible for complying with epistemic norms even if breaking them does not cause harm. My point here does not depend on this contentious claim.) For example, a teacher who teaches her pupils to believe inconsistent things arguably harms them. The failure of a political party to communicate a coherent message may undermine its otherwise morally worthwhile efforts. The unexamined life is not worth living.
For the record, I am not sure whether this way of describing the moral significance of epistemic norms is correct. Perhaps there is no special value, not even prima facie moral value, to truth, consistency, knowledge, etc. But if there is, epistemic actions might be seen as different from non-epistemic actions in terms of what makes them morally significant. Nor am I arguing that this difference, even if it can be shown to exist, undermines the main argument of the paper. Still, it would be important to address it so that we get a better grasp on what exactly the agent is held morally responsible for when performing epistemic actions.
Miller, Seumas. “Joint Epistemic Action and Collective Moral Responsibility.” Social Epistemology (2015): 1-23. doi: 10.1080/02691728.2014.971908.
Petersson, Björn “Collective Omissions and Responsibility.” Philosophical Papers 37, no. 2 (2008): 243-261. doi: 10.1080/05568640809485221.