Collective culpable ignorance

I argue that culpable ignorance can be irreducibly collective. In some cases, it is not fair to expect any individual to have avoided her ignorance of some fact, but it is fair to expect the agents together to have avoided their ignorance of that fact. Hence, no agent is individually culpable for her ignorance, but they are culpable for their ignorance together. This provides us with good reason to think that any group that is culpably ignorant in this irreducibly collective sense is non-distributively collectively responsible for subsequent unwitting acts and consequences.


| INTRODUCTION
Suppose a doctor is unaware of a patient's allergy and the patient dies due to complications of the treatment. If the doctor is to be responsible for her "unwitting act," the act done out of ignorance, then she must be responsible for her "benighting act," an omission to inform oneself (Smith, 1983). Suppose the doctor's ignorance of the fact that the treatment should not have been administered is her fault, because she misread the patient's chart. If so, then, without a legitimate excuse, the doctor is blameworthy for the patient's death. The doctor is derivatively responsible for the patient's death in virtue of being originally responsible for the "benighting act" of misreading the patient's chart. If, however, the doctor has an excuse for her ignorance, for example, a resident failed to register this information, then the doctor is not blameworthy for her "unwitting act" and the consequences.
Given this, there is good reason to accept.
Derivative Blameworthiness: An agent is blameworthy for (the consequences of) an unwitting act only if she is blameworthy for her ignorance the act was wrong. 1 The blameworthiness for the "unwitting act" is derivative of the agent's blameworthiness for the "benighting act" of failing to inform herself. But when is an agent blameworthy for her ignorance? We typically say something akin to: "She should have known better". This allegation, when fair, implies that the ignorance was avoidable and typically stems from some failure of the agent. A plausible principle that spells this out is Reasonable Expectation: An agent is blameworthy for her ignorance of a moral or non-moral fact if and only if she could reasonably have been expected to take measures that would have avoided or remedied this ignorance. 2 (FitzPatrick, 2017;Levy, 2009;Rosen, 2004;Zimmerman, 1997) If the resident failed to register the crucial information, then it is unfair to expect the doctor to have avoided her ignorance. Hence, she is not blameworthy for her circumstantial ignorance. Let us assume that Derivative Blameworthiness and Reasonable Expectation are correct.
In this paper I argue that agents' culpability for their ignorance can be irreducibly collective. They are culpable for their ignorance, but no individual is culpable for her ignorance. This is a purely collective predicate as in "we surround the building." The predicate does not distribute over the members individually. Turning to the consequences, I argue that the group's blameworthiness for the (consequences of the) unwitting act is non-distributive. The group is responsible, but no individual is responsible. Now, the claim that groups are apt targets for blame is not new. Some argue that certain groups qualify as (moral) agents and are fit to be held responsible (Copp, 2007;French, 1984;Gilbert, 2014;List & Pettit, 2011). Importantly, here I argue that a non-agential group can be non-distributively collectively responsible (Björnsson, 2020;Chant, 2015;Feinberg, 1968;Held, 1970). What is novel is that I explain how collective culpable ignorance can generate this result.
Before I start, two caveats. I only focus on circumstantial ignorance. I am skeptical that culpable normative ignorance can be irreducibly collective, but I will not address this. Moreover, I focus on culpable ignorance related to moral responsibility as accountability, which involves the imposition of demands and a readiness to treat an agent adversely when she fails to comply with these demands (Watson, 1996). I understand an agent being blameworthy for X as the agent, on account of X, being a proper target for reactive attitudes such as resentment, indignation, and guilt (Strawson, 1962). I am not concerned with moral responsibility as attributability (Arpaly, 2003;Talbert, 2013).

| OUTBREAK
Culpable ignorance can be irreducibly collective. Consider Outbreak: Two scientists, John and Eve, are members of a global health thinktank tasked with protecting the world from diseases X, Y, Z. John studies the nature of a particular infectious disease X. Eve's study is related to environmental and sociological aspects of Y and Z. John's ground-breaking research proves conclusively fact A: virus X mutates into a highly contagious and deadly hybrid X* under very specific circumstances C. Eve's ground-breaking research proves conclusively fact B: in environment E these very specific circumstances C are generated by diseases Y and Z because of sociological and environmental factors. From the combined results it can be inferred that O: an outbreak of X* in E is imminent. In the think-tank there are norms of transparency and confidentiality. Members are expected to share relevant knowledge but keep all shared information confidential. Although John and Eve are aware of both norms and know full-well they ought to share their research, neither shares their results on various occasions the thinktank gathers because they fear someone will steal their idea and publish the results before them. After some time, there is a disastrous outbreak of X* in E. An outbreak they could have prevented.
Note that while they are members of an organized group (the think-tank), John and Eve themselves, as a plurality, constitute "merely" a non-agential group. This plurality lacks any features (e.g., a decision-making procedure or joint commitment) in virtue of which it could possibly qualify as an agent (Gilbert, 2014;List & Pettit, 2011). John and Eve are ignorant of the nonmoral fact O that an outbreak of X* in E is imminent. There is nothing special about their ignorance itself. Of course, had one of them shared their research, this would have made it possible for the other agent to infer that O. Had Eve shared her knowledge that B, then John would not have had an excuse, because John could have avoided his ignorance of O by making the inference from A and B. But, given that Eve did not do so, it is unfair to expect John to have avoided his ignorance of O. If Reasonable Expectation is correct, then John is not blameworthy for his ignorance of O, because John could not reasonably have been expected to have taken measures that remedied his ignorance of O. The same holds for Eve. Therefore, each has a legitimate excuse for their own ignorance. The wrongdoing of each, failing to share important information, serves as an excuse for the other's ignorance of O.
However, it seems fair to allege that they should have known that O. So, who, if any, is culpable for their ignorance of O? Both John and Eve have ready-made opportunity on numerous occasions to share their knowledge when the think-tank convenes. Furthermore, given the think-tank's aim, the potential relevance of their knowledge is salient. Suppose further that the norms of transparency and confidentiality have never been violated. Moreover, the inference that O is not too difficult once facts A and B are known. Given this, while it is unfair to expect John or Eve individually to have avoided their ignorance of O, it is fair to expect John and Eve to have known that an outbreak of X* was imminent. They could easily have taken measures sufficient for avoiding their ignorance. Therefore, they are culpable for their ignorance of O. John and Eve's culpability for their ignorance is irreducibly collective.
de HAAN An interesting feature of Outbreak is that neither agent has violated an individual duty of inquiry. Typically, when an agent is blameworthy for their ignorance concerning the unwitting act, the agent violated a duty to inform herself whether the unwitting act is wrong, meaning the agent should have informed herself regarding the moral and/or non-moral facts (cf. Smith, 2014). It is hard to see how the agents could have had an all-things-considered duty to inform themselves whether the think-tank's inaction was wrong, because neither individually could have been aware of the fact that O.
Instead, each agent violated what I call a duty of exchange. In group contexts, sharing information is crucial for the pursuit of collective goals. A moral epistemic duty of exchange is an individual duty to share potentially relevant information in certain circumstances for the pursuit of morally weighty goals. Irreducibly collective culpable ignorance may arise because some inferences can only be made based on shared information. When agents fail to share information, making fairly simple inferences becomes practically speaking impossible. The "benighting act," then, consists of two acts: John not sharing his information and Eve not sharing her information. Each is directly blameworthy for their part of the "benighting act." But neither is individually blameworthy for the ensuing ignorance of O, because each agent has an excuse: she, reasonably speaking, could not have avoided her ignorance of O. Yet this excuse does not hold for John and Eve together, as they together could reasonably have been expected to have taken measures sufficient for avoiding their ignorance of O. Hence, they are culpably ignorant.
Another important feature of Outbreak is that the agents are akratic. Some think that a person cannot reasonably have been expected to have taken measures to avoid or remedy her ignorance if she was unaware of the fact that she should have taken such measures (Rosen, 2004;Zimmerman, 2008). The agent is culpable for their ignorance only if the ignorance traces back to an episode of akrasia. In a case of wrong action due to clear-eyed akrasia, a person performs a wrong action while aware that she has most reason to act otherwise (Harman, 2011). John and Eve are aware that they ought to share their knowledge, especially given the prevalence of norms of transparency and confidentiality, but on multiple occasions decide not to do so against their better judgment. Both are weak-willed and intentionally withhold their research because they are led astray by their selfishness. Thus, even if akrasia is necessary for Reasonable Expectation to obtain, culpable ignorance can be irreducibly collective.
But Outbreak need not necessarily involve akrasia. There is good reason to doubt akrasia is necessary for culpable ignorance (FitzPatrick, 2008(FitzPatrick, , 2017cf. Levy, 2009;Robichaud, 2014). William FitzPatrick argues that Reasonable Expectation may also obtain due to "the culpable, nonakratic exercise of such vices as overconfidence, arrogance, dismissiveness, laziness, dogmatism, incuriosity, self-indulgence, contempt, and so on" (FitzPatrick, 2008, p. 609). We can reasonably expect agents not to exercise such vices. Suppose that John and Eve's roles in Outbreak are different. John is tasked to keep up with relevant virology studies, and Eve with relevant environmental and sociological studies. John's available evidence includes a study that proves fact A, but John is lazy and has failed to heed all available evidence before the meeting. Eve is also lazy and has failed to heed her available evidence that includes fact B. From facts A and B, it could again be inferred that O. The agents do not violate their duties of exchange but their duties of inquiry concerning the facts from which the inference could be made. This creates another layer of ignorance. The agents are not akratic, but instead their faults stem from a nonakratic exercise of their vice of laziness.
Again, each agent's wrongdoing serves as an excuse for the other agent's ignorance of O. Even if John had done his job accordingly, he could reasonably speaking not have avoided his ignorance of O. However, this excuse does not hold for them. It is fair to expect them to have avoided their ignorance of O, because each of them was tasked with gathering evidence of the relevant domains, they had reasonable opportunity to ascertain the relevant facts, but simply failed to do so because they are lazy. John and Eve's culpability for their ignorance of O is again irreducibly collective.
Some may think that John and Eve are individually culpable for violating their respective individual duties of exchange (or, alternatively, their duties of inquiry), but that no one is culpable for their ensuing ignorance of O. The agents are culpable, perhaps extremely culpable, for failing to share their information, but nothing else. One might think that this distributive reading has the advantage of avoiding burdensome commitments about the obligations of nonagential groups, assuming that if John and Eve's culpability is irreducibly collective, then they must have violated a group-level obligation. 3 "Revisionists" about obligations of non-agential groups introduce new moral vocabulary and concepts to explain the obligations of pluralities. Some argue that the unorganized group is the bearer of the group-level obligation (Aas, 2015;Isaacs, 2011;May, 1992;Schmid, 2018;Wringe, 2010), whereas others argue that members hold obligations jointly (Pinkert, 2014;Schwenkenbecher, 2019Schwenkenbecher, , 2020 or share them (Björnsson, 2014(Björnsson, , 2020. Such revisionary proposals conflict either with the Agency Principle (only moral agents can have moral duties) or the Ability Principle (A can have a moral duty to X only if A is able to X). "Conservatives" claim that we can explicate the obligations of non-agential groups in terms of (complex conjunctions of) individual obligations and avoid revisions to our moral concepts and principles (Collins, 2019;Goodin, 2012;Lawford-Smith, 2015). 4 First, in response, the suggested distributive analysis does not suffice. Eve and John may be extremely culpable for their individual-level violations of their duty of exchange (or inquiry). The degree of their responsibility is very high. But there is still a gap between what they intuitively are culpable for and what each is individually culpable for. This concerns the scope of responsibility rather than degree. Intuitively, it is true that they are culpable for their ignorance of O. After all, John and Eve's individual excuses do not hold for them together. They could reasonably have been expected to have avoided their ignorance. They violated their duties of exchange. This resulted in ignorance. But to fully account for the moral relevance of their unexcused ignorance, we need to invoke group-level culpability. This becomes especially evident when we consider what their ignorance leads to, namely unexcused morally wrongful "unwitting" inaction. I return to this in Section 3.
Second, turning to John and Eve's epistemic obligations, obligation-revisionists are likely to take Outbreak as evidence for their position. I cannot evaluate these positions here but let me point out that Outbreak is not necessarily an argument in favor of obligation-revisionism. If John and Eve are together culpable for their failure to know O, then it must indeed be morally wrong that they fail to know that O, which implies a violation of procedural epistemic obligations (Rosen, 2004). However, "conservatives" can simply reject that collective responsibility necessarily entails the violation of a group-level obligation (de Haan, 2021;cf. Lawford-Smith, 2015). They can argue that their wrongful ignorance of O results from the violation of their two individual duties of exchange (or inquiry) concerning A and B. This suffices to account for the procedural epistemic obligations. 5 One could accept that a plurality can be (non-distributively) collectively responsible for their ignorance but reject that this entails the violation of a group-level obligation. Hence, Outbreak speaks in favor of responsibility-revisionism, but not necessarily for obligation-revisionism. I leave open here how to account for the obligations of non-agential groups, but this is no reason to deny that John and Eve are irreducibly collectively culpable for their ignorance of O. de HAAN Another possible response to Outbreak is to argue that each is culpable for the other agent's ignorance of O and that this accounts for all culpability. This may initially appear plausible. Note that the claim "John and Eve should have known that O" is distributive, because plausibly it is also correct that John should have known that O. Sandy Goldberg (2017) argues that there can be a gap between what an agent should have known and for what ignorance an agent is culpable. It is possible the "Should have known" allegation holds but someone else is culpable. For example, a rookie high-school baseball player may be ignorant of the fact that a hit ball is still in play if it hits the third base bag. The back-up player should have known this. There is a practice-generated expectation of such knowledge, as any baseball player should know the rules of baseball. But the rookie's ignorance (arguably) does not stem from a fault on his part. He could not reasonably be expected to have avoided his ignorance. His coach should have taught him this rule. Hence, the rookie is not culpable for his ignorance that results in the opposite team winning the game. Instead, the coach is to blame. (Goldberg, 2017(Goldberg, , p. 2887.
If correct, then we may accept Interpersonal Reasonable Expectation: X is blameworthy for Y's ignorance of P if and only if X could reasonably have been expected to take measures that would have avoided or remedied Y's ignorance of P.
Given this, in Outbreak, could John be blameworthy for Eve's ignorance of O and Eve be blameworthy for John's ignorance of O? After all, John's ignorance is partly due to Eve's fault, and Eve's ignorance is partly due to John's fault. This would mean the culpability for their ignorance is not irreducibly collective but distributed inversely. But this is implausible. Although both the coach and the agents in Outbreak lack an excuse for failing to take the respective measures, there are important differences between the cases. First, unlike any agent in Outbreak, the coach knows the fact of which the player is ignorant. Second, unlike in Outbreak, if the coach does not know the rule, then the coach would be culpable for his ignorance of this fact. Third, most importantly, the coach could have taken measures sufficient for remedying the player's ignorance, namely telling him about the rule. In Outbreak, there is nothing John could have done that would have been sufficient for remedying Eve's ignorance. As mentioned, John could have informed Eve about A. But this does not guarantee that Eve knows that O. She might nonetheless fail to make the inference. So, each agent could fairly have been expected to take measures that were necessary for avoiding or remedying the other agent's ignorance of O. 6 To account for this, Interpersonal Reasonable Expectation must concern measures that are sufficient or necessary for avoiding Y's ignorance. But so formulated, Interpersonal Reasonable Expectation is no longer plausible. Suppose a secretary forgets to schedule a lab for a scientist without a legitimate excuse. All labs are booked for the next semester. Access to the lab is necessary for the scientist to do his research. Suppose the scientist is on the brink of finding a cure for some disease. Is the secretary now on the hook for the scientist's ignorance of the cure? This is counterintuitive. The relation between the scientist's ignorance and the secretary's failure is not clear or robust enough to warrant such an ascription. It is reasonable to expect the secretary to have taken measures necessary for avoiding the scientist's ignorance, but it is counterintuitive to think that this is sufficient for the secretary to be blameworthy for the scientist's ignorance.
Moreover, Outbreak can be modified such that John is not the only scientist who knows that A. If so, then John's contribution is neither necessary nor sufficient for avoiding Eve's ignorance of O. So, Interpersonal Reasonable Expectation must now include measures that merely contribute to avoiding Y's ignorance. This principle is very implausible. It would lead to an overabundance of ascriptions of interpersonal culpable ignorance and call into question the very notion itself.
Therefore, Interpersonal Reasonable Expectation cannot account for Outbreak. John and Eve are not blameworthy for each other's ignorance of O, instead their culpability is irreducibly collective. What does this mean for responsibility for the "unwitting act" and consequences?

| COLLECTIVE RESPONSIBILITY
Suppose that the outbreak results in the deaths of hundreds. Suppose further that had John and Eve shared their results, and inferred that an outbreak was imminent, there would have been enough time to prevent the outbreak. Who, if any, is blameworthy for failing to prevent the deadly outbreak?
John and Eve cannot be blameworthy for failing to prevent an outbreak in a distributive sense. John and Eve's inaction failed to prevent an outbreak, and their inaction is due John and Eve's ignorance of non-moral fact O. But each has an excuse for their individual inaction. John's "unwitting" inaction results from his ignorance of O. Following Derivative Blameworthiness, if John is blameworthy for (his part of) the inaction, then John must be culpable for his ignorance. However, John is not culpable for his ignorance of O. The same holds for Eve. Therefore, any claim that John or Eve is individually responsible for his/her part contradicts Derivative Blameworthiness. Some may think that for this reason we must reject Derivative Blameworthiness. Perhaps we simply must attribute blameworthiness in this case without tracing it back to some prior blameworthiness. Or perhaps John's blameworthiness for the outbreak is not derivative of culpable ignorance, but derivative of John's blameworthiness for failing to share his information. This does not work. John has a legitimate excuse for his inaction. If John could reasonably speaking not have avoided his ignorance of O, then he could reasonably speaking not have avoided his inaction. This excuse holds even if John is blameworthy for failing to share information.
Could no one be blameworthy for failing to prevent the deadly outbreak? Had it not been for the opportunities provided by the context in which the scientists find themselves, they likely would not have been responsible for the consequences of keeping the results to themselves. But both scientists agreed to a role in a think-tank that was specifically tasked with monitoring and combatting the disease. An essential part of this role is sharing their expertise for policy-making. Each had ample opportunity to share their results. Instead each decided to place an unwarranted fear above their role-responsibilities. It is reasonably foreseeable that failing to share potentially vital information could lead to disastrous consequences in this context. Most importantly, John and Eve are together culpably ignorant of the fact that an outbreak was imminent. Typically, if the consequences of an unwitting act are foreseeable to a reasonable extent, and one is culpable for one's ignorance about the unwitting act, then one is blameworthy to some degree for the consequences. The fact that John and Eve are collectively responsible for their ignorance of O is crucial because this removes any excuse for their inaction. They could fairly be expected to have known that O, therefore they can fairly be expected to have prevented the deadly outbreak. Their failure to prevent a deadly outbreak constitutes unexcused moral wrongdoing. This must not be left unaccounted for. Therefore, John and Eve's blameworthiness for the unwitting act of failing to prevent the outbreak is non-distributive. Neither John nor Eve de HAAN is individually blameworthy, but they are collectively responsible for failing to prevent the outbreak.
I have not said anything about whether the think-tank should be considered a moral agent (French, 1984;Gilbert, 2014;List & Pettit, 2011). Given the organization of the think-tank, this is plausible. Some may think this must mean that the think-tank is to blame. But Outbreak shows that, just as a single member can be blameworthy within a group agent, a plurality of members can be collectively responsible within a group agent without the group agent itself being to blame. Suppose the think-tank consists of 10 members including John and Eve. If a member appointed as an enactor fails to perform an action, the group agent might likewise be culpable, because it should have responded to the member's inaction and appointed someone else. But in Outbreak it is not obvious the group agent could have avoided this, precisely because it concerns ignorance. The group agent has created a culture with the right norms of transparency and confidentiality. The group agent does not systematically prevent the possibility of sharing knowledge. Hence, the group agent seems to be organized in the right way. And so, the group agent did everything it can to avoid this scenario, but it had two recalcitrant members performing their tasks poorly. Therefore, John and Eve are collectively responsible, but the think-tank is not.
It is important to see that the structural features of Outbreak do not necessarily depend on the plurality being part of a group agent. What makes the culpability for John and Eve's ignorance irreducibly collective is that each agent's violation of a duty of exchange or inquiry provides the other member individually with an excuse for their ignorance of some further morally relevant fact. But these excuses do not hold for the plurality. Whether it is fair to expect them to have avoided their ignorance depends on the circumstances, but this pattern can show up within any group. It would be interesting to explore in which scenarios these epistemic norms show up. But for now, what I hope to have shown is that irreducibly collective culpable ignorance potentially creates numerous cases where non-agential groups are blameworthy as such for their "unwitting acts".
ORCID Niels de Haan https://orcid.org/0000-0002-4021-7410 ENDNOTES 1 As formulated here, Derivative Blameworthiness is consistent with views that deny blameworthiness for the unwitting act derives from blameworthiness for the ignorance (Harman, 2011). Quality-of-the-will accounts (Arpaly, 2003;Björnsson, 2017) claim blameworthiness for unwitting acts is to be explained by a lack of moral concern. But quality-of-the-will accounts (should) also hold that blameless ignorance excuses "unwitting acts" (because the agent does not perform the act with a lack of moral concern). Therefore, the case is also relevant for quality-of-the-will accounts. 2 I follow FitzPatrick (2017) here.