On the culpable ignorance of group agents: the group justification thesis

ABSTRACT

People are often responsible for what they do, but they also often possess an excuse. One of the most common excuses is ignorance. Not all ignorance constitutes an excuse, however, for some ignorance is culpable and culpable ignorance is no excuse. But what about group agents? In our everyday practices, we blame group agents constantly. But if groups can be blameworthy, they plausibly can also be excused. Surely one such excuse is ignorance. But, as with individual agents, some group ignorance is also surely culpable. How should we understand the culpable ignorance of group agents? In this paper, I argue for two conditional claims that together constitute what I call the Group Justification Thesis, which is a group-agent adaptation of the Justification Thesis put forth by Biebel [(2018). ‘Epistemic Justification and the Ignorance Excuse.’ Philosophical Studies 175 (12): 3005–3028]: (i) A group agent is excused because of ignorance only if that ignorance is justified, and (ii) a group agent is culpable despite being ignorant only if that ignorance is not justified. One interesting upshot of this view is that a group agent can be culpably ignorant even though none of the members are culpable for their ignorance.

KEYWORDS:

• Culpable ignorance
• group agents
• group moral responsibility
• epistemology of groups
• group agency

1. Introduction

Individual moral agents are often responsible for what they do, but they also often possess an excuse. One of the most common excuses is ignorance. This excuse takes on many forms: ‘I didn’t see it,’ ‘I couldn’t hear over the noise,’ ‘How was I supposed to know that?!’ All of these indicate that one lacks a crucial piece of information necessary to avoid wrongdoing. In other words, the agent fails to meet the widely accepted epistemic condition for moral responsibility – the condition that a morally responsible agent must not be ignorant that her action is wrong. Not all ignorance constitutes an excuse, however, for some ignorance is culpable and culpable ignorance is no excuse.¹ How to make sense of the difference between exculpating and culpable ignorance for individual moral agents has been a source of recent, lively debate.²

But what about group agents? In our everyday language we speak of some well-organized groups as though they possess capabilities akin to us individual agents: we say they believe things, want things, own things, and perhaps most importantly, they do things intentionally. Moreover, we think they are open to normative evaluation across those domains – we think they shouldn’t act in certain ways, we criticize them for lacking knowledge or for having false beliefs, and we think they ought not have certain goals or desires. Indeed, moral evaluation of group agents is particularly common; we blame them constantly. But if such groups can be agents capable of doing wrong and being blameworthy for it just as individual agents are, then surely they might also occasionally be excused just as individual agents are. Given that we also think of group agents as believing things and knowing things, it seems likely that ignorance is one such excuse. But also just as with individual agents, not all group-based-ignorance will constitute an excuse, for sometimes it will be culpable and the culpable ignorance of a group agent (hereafter just ‘group culpable ignorance’) is surely no excuse.

Much recent attention has been given to making sense of these common ascriptions of moral and epistemic agency as they apply to groups, but nearly nothing has been said about the nature of excuses as they apply to group agents.³ However, culpable ignorance in a group setting has been given some attention. Hormio (2018) discusses the pervasive ways that groups can produce culpable ignorance among their members. Schwenkenbecher (2020, 2022) is similarly concerned with how group membership might lead to individual ignorance and discusses what duties a group might have to avoid producing ignorance among its members. Finally, de Haan (2021) argues that certain types of collectives can be culpably ignorant even though that collective is not an agent. Importantly, none of these authors has been specifically concerned with the culpability of ignorance that is attributable to a group agent – only the culpability of the members, the culpability of the group for the member’s ignorance, or to collectives which are not agents.

This paper attempts to fill that gap by offering the beginnings of an account of group culpable ignorance. I argue that whether a group’s ignorance is culpable depends entirely on whether that group’s belief is justified. The intimate connection between justified group belief and group culpability has been recognized by Lackey (2021) as a crucial motivation for examining group epistemic agency, but thus far no comprehensive explanation of the connection between epistemic justification and culpability has been offered. My aim is to connect the two. I argue for two conditional claims that together constitute what I call the Group Justification Thesis, which is a group-agent adaptation of the Justification Thesis put forth by Biebel (2018): (i) A group agent is excused because of ignorance only if that ignorance is justified, and (ii) a group agent is culpable despite being ignorant only if that ignorance is not justified.

To support this claim, I suggest that an analysis of group culpable ignorance should include three features: (1) the resultant belief must be sufficiently supported by the evidence the group actually possesses, (2) the group has discharged all of its intellectual/procedural obligations (e.g., Peels [2017] and Silva's [2019] ‘believing responsibly’), and finally (3) some reference to the pragmatic stakes – the risks of being wrong – which play an important role in determining the threshold beyond which one’s evidence and efforts are sufficient. In short, if acting on a false belief could have very bad results, then it requires a greater degree of certainty. A failure to adequately address these risks at the group level makes it appropriate to blame the group despite their ignorance. My claim is that these features of culpable ignorance are the same criteria as those for determining when a belief is justified (Biebel 2018; Fantl and McGrath 2009). One interesting upshot of this view is that, because the stakes for the group can come apart from the stakes of the individual members, a group agent can be culpably ignorant (because unjustified) even though none of the members are culpable for their ignorance (because justified). However, interestingly, this does not necessarily commit us to the claim that the group agent can be blameworthy even if none of the members are.

The paper will proceed as follows. In the next section (2), I will briefly explain the necessary background assumptions I am making regarding the nature of group agents, their beliefs, and their blameworthiness. In Section 3, I provide what is my preferred analysis of culpable ignorance and explain how it leads to the Justification Thesis. In Section 4, I explain how this extends to group agents and leads to the Group Justification Thesis and discuss some important differences for group agents that create opportunities for culpable ignorance that do not arise for individual agents. In Section 5, I argue that it is possible for a group agent to be culpably ignorant even though none of the members are culpable for their ignorance, and that this reveals one common cause for the culpable ignorance of group agents. I conclude with Section 6 by gesturing at some implications for further research.

2. Background

There are many somewhat controversial issues at the periphery of this project. What do I mean when I refer to a ‘group agent’? What does it mean to say they ‘believe’ or are ‘ignorant’? What is a theory of epistemic justification meant to be a theory of, exactly – what is a ‘justified belief’? And finally, what do I mean when I say an agent is culpable or blameworthy? In this section I will briefly answer these questions.

What is a group agent? When I speak of a group agent, I am talking about a specific type of group that is well organized and created with a specific purpose. Such groups generally have some established decision procedures (List and Pettit 2011) or some ethos defining their collective purpose or goals (Tuomela 2013). Most often these types of groups constitute organizations, institutions, or corporations, though they might also include courts, states, or even simple committees. What unites them is that we naturally think of them as capable of deliberation, making decisions, and intentional action. And, in virtue of this, they are the bearers of duties/obligations (Collins 2019b, 2023; Gilbert 2004; List and Pettit 2011; Schwenkenbecher 2020; Tuomela 2013). Jennifer Lackey (2021) says it best: ‘Put succinctly, if we can properly hold a group, G, responsible for φ-ing, then this is sufficient for regarding G as a group in the sense relevant for this project’ (7). So, when I speak of a group agent, I will mean a group that is the appropriate target of normative evaluation in light of the group’s capacities to deliberate, make judgments (form beliefs), and act intentionally.

What is group belief and group ignorance? In everyday language, ignorance is often equated with a lack of knowledge,⁴ but when it comes to normative assessments, the concept of ignorance tends to have a slightly different scope. One can lack knowledge, for example, by forming an unjustified true belief, but as Gideon Rosen (2008) observes, we would blame a person who acted despite having an unjustified true belief that what they are doing is wrong every bit as much as we would a person who knows it.⁵ So, when I speak of ignorance I will mean ‘a failure to believe of some true proposition, that it is true’ (Zimmerman 2017, 79), either because one believes falsely or one withholds belief.

Of course, claiming that groups can be ignorant implies that groups can have beliefs, but what is the nature of group beliefs? As will become apparent below I endorse a kind of inflationary view of justified group belief. Views like this claim that it is possible for the group (qua group) to fail be justified in believing a proposition even though all (or enough) of the members are justified in believing it, and vice versa.⁶ However, I remain neutral about the ontology of such beliefs. I leave it to readers to employ their favored account, whether that be a kind of functionalist approach like List and Pettit (2011), Bird (2014), and Strohmaier (2020); a kind of joint acceptance model like Gilbert (2002), Hakli (2011), or Mathiesen (2011)⁷; or interpretationism like Tollefsen (2015).⁸ On any plausible ontology, group beliefs will be open to normative evaluation of the sort at issue, so what I have to say will apply.

What does it mean to say a belief is justified? When I say that a belief is justified, I mean it in the binary sense that a belief is either justified or it is not. More specifically, the notion of justification I have in mind is normative in this sense: a belief is not justified if and only if it is appropriate to claim that the agent ought not believe it, and conversely, a belief is justified if and only if it is no longer appropriate to claim that the agent ought not believe it. It should be noted that this is a form of epistemic permissivism, a belief is justified if it is epistemically permissible to believe it. In other words, it is possible that one’s evidence might be strong enough to permit belief while not being so strong that withholding belief would itself be unjustified (though I also think that sometimes withholding belief is unreasonable given the strength of the evidence).

What do I mean when I say an agent is blameworthy? In my view, an agent is blameworthy under two conditions: (1) The agent performed an objectively morally wrong action and (2) the agent was morally responsible for having committed that action. These very general conditions are widely endorsed in moral philosophy, and in what follows I shall assume that group agents are capable of both. I take it that (1) is something widely accepted; group agents can be wrongdoers (Collins 2019b, 2023; de Haan 2020, 2023; Hess 2014; Lackey 2021; List and Pettit 2011; Silver 2022; among many others). Indeed, group agents can do wrong in some ways and at some scales that no individual agent ever could. British Petroleum – the company – is the entity to which we attribute an oil spill. No individual human agent could feasibly drill for, and subsequently spill millions of barrels of oil into the ocean. Only a group agent could do that. (2) is more controversial but still, I think, widely accepted. Although accounts of the phenomenon vary: perhaps to be morally responsible the group agent must be capable of recognizing and responding to moral reasons (de Haan 2023; List and Pettit 2011), or capable of being duty bearers (Collins 2019b, 2023). Perhaps they must be the appropriate target of reactive attitudes such as resentment or indignation and capable of having some functional version of the appropriate response attitudes such as guilt or remorse (Björnsson and Hess 2017; Tollefsen 2007), or perhaps they must possess free will (Hess 2014). I will remain neutral on which conditions are required for group moral agency, but I will proceed as though, whatever those conditions are, at least some groups meet them. It should be noted, though, that the notion of moral responsibility here employed differs from other forms of responsibility such as strict liability in the law, or forward-looking responsibility.

With the stage now set, the remainder of the paper will present a theory on the difference between group culpable and exculpating ignorance. However, since my claim is that we can adapt a theory of individual culpable ignorance to suit group agents, I should first explain the concept of culpable ignorance as it applies to individual agents. In Section 3, I will provide a version of my favored analysis of the difference between culpable and exculpating ignorance for individual agents. In Section 4 I will explain how this extends to group agents and point out a few key differences. Finally, in Section 5 I argue that one of these differences can result in group culpable ignorance even though none of the members are culpably ignorant.

3. The justification thesis

Recall from the introduction that one of the most common excuses is ignorance. The most obvious instances of this are cases where an agent simply has no reason to believe that what they are doing is wrong. Suppose Zaria is visiting Clive and Clive invites Zaria to make herself some tea saying only ‘It’s in the cupboard above the sink.’ Zaria really likes black tea, and since the most natural interpretation of Clive’s invitation is that any tea is fair game, she makes herself some black tea. However, Clive forgot to tell Zaria that he was planning on saving the black tea to give to his mother for her birthday. Intuitively, Zaria is not blameworthy for using the black tea Clive meant to save for his mother, and the reason for this is because she did not know the tea was off-limits. Since she did not know, she is not blameworthy for the error – she is excused in virtue of her ignorance.

Ignorance is not always an excuse, however. Suppose a doctor named Steve is in a bit of a rush because he is late for his golf tee-time, and he gives his patient’s chart only a cursory glance. As a result, he fails to see that the patient is allergic to penicillin. As it happens, penicillin is the medication usually prescribed for the patient’s symptoms. Since Steve does not know about the allergy, he prescribes penicillin causing the patient to have an allergic reaction and nearly die. Steve is ignorant of the patient’s allergy, yet he obviously should not be excused. But why not? What is the difference between Zaria’s ignorance and Steve’s?

The answer is that Steve (but not Zaria) should have known. Put more precisely, we cannot appropriately tell Zaria that she should not have been ignorant. She falsely believed that the black tea was fair game, but this false belief is reasonable under the circumstances. Conversely, we certainly can appropriately tell Steve that he should not have been ignorant because carefully checking a patient’s chart for allergies is exactly the sort of investigation a doctor is obliged to do.⁹

So, the key difference is one of normative evaluation: an agent is blameworthy despite their ignorance when they should not have been ignorant. But how do we evaluate this? Under what conditions is an agent open to moral criticism for their ignorance? I suggest that there are three factors that are relevant to this normative evaluation: (1) the evidence one possesses, (2) what Gideon Rosen (2003, 2008) calls ‘procedural obligations’¹⁰, and (3) the stakes – how badly things could turn out if one is wrong. Let me explain each of these in more detail.

The first criterion is the most straightforward. Clearly, one’s belief should be sensitive to one’s evidence. If it is not, then we may point out the flaw in that belief immediately – it is not even supported by one’s own evidence. However, there must be more to the story than just this. Notice, Steve does not technically possess evidence that his patient is allergic to penicillin because he is not aware that the chart indicates as much. Steve doesn’t read the chart properly. Hence, he misses that bit of evidence, and so Steve’s ignorance of his patient’s allergy is consistent with the evidence he possesses. This is why, as Rosen (2003, 2008) and Biebel (2018) point out, this strictest evidentialist criteria, while necessary, is not sufficient to ground culpability, for if all that were required for one’s false beliefs to be free of criticism is that it aligns with the evidence one possesses, then Steve would be off the hook.¹¹ Steve clearly should not be excused, so we need some other criteria.

The natural next step is to point out that Steve’s belief is flawed because there is available evidence that can be acquired with little effort. Moreover, it seems like Steve really ought to make that effort. Indeed, just as Rosen (2003, 2008) once more points out, there are certain ‘procedural obligations’ that must be discharged before an agent’s believing is free from criticism. This is not just true of Steve’s beliefs about his patient’s allergies, it is also true of many ordinary beliefs. We often need to double check, investigate further, spend a bit more time thinking about things, or take some other reasonable steps to, as Holly Smith (1983) puts it, improve our cognitive position. This is because such activities are belief influencing in that they improve one’s pool of evidence, either by adding more to the pool, or by improving one’s evaluation of the evidence already in it, thus ensuring that we believe responsibly (Peels 2017). This seems to be the problem with Steve’s belief; he didn’t check the chart thoroughly, he easily could have, and doing so is necessary to discharge his procedural obligations.

However, there is still one complication remaining. Procedural obligations are, at the most general level, obligations to take reasonable measures to improve one’s evidence, but determining which measures are reasonable is not settled by merely pointing out how easy evidence is to acquire. Zaria also could easily have discovered that the black tea was off-limits; she need but ask Clive. The difference is that Steve is obliged to make the effort and Zaria is not, but what makes it so that Steve must make this extra effort, but Zaria need not? Our analysis must include some way to determine how much effort is enough to discharge one’s procedural obligations – something that explains the difference between Zaria and Steve.

One obvious difference is that Steve is acting in his capacity as a doctor, and doctors have obligations that do not apply to other people. It should come as no surprise that a person’s profession can carry with it obligations related to the specifics of their job. Doctors, lawyers, accountants, and really anyone who provides a specialized service to others. We expect that they will make more effort to get things right than someone who isn’t a doctor, lawyer, accountant, etc. But it would be a mistake to think that these obligations are limited to one’s profession. Parents who have a child with a peanut allergy, for example, have a special obligation to carefully investigate the ingredients of the food they give their child. So, it isn’t limited to one’s profession. Instead, what gives rise to these special procedural obligations seems to be how important it is to get things right, or alternatively, how badly things could turn out if one is wrong (Biebel 2018). A doctor ignorantly prescribing penicillin to a patient with a known allergy will probably turn out very badly. The same can be said of a parent ignorantly feeding peanuts to their allergic child. However, the same cannot be said of Zaria ignorantly using the black tea Clive intended for his mother.

Of course, there is the potential for things to go extremely badly even for patients whose charts do not list an allergy to medication, or for children who do not currently have a known peanut allergy. After all, many patients never have the occasion to discover an allergy to penicillin, and peanut allergies can develop later in life. Should we require a doctor to do expensive tests on every patient just to be safe? Should we demand that all parents diligently avoid feeding their children peanut butter because there’s always a chance, however remote, that the child has developed a peanut allergy? After all, the stakes are extremely high!

Surely this spills over from reasonable precaution into paranoia. If what determines how far one must go to discover the truth is the mere possibility that things could turn out very badly, then we would be paralyzed by the constant need for more thorough investigation. Fortunately, it isn’t the mere possibility that matters, what matters is the likelihood that such a possibility might occur. There is always a non-zero probability that something very bad could happen, but our epistemic and practical lives need not be fraught with worries about bad outcomes that have little perceivable chance of happening. This is why parents can happily let their children eat peanut butter. There is certainly some chance that the child has recently developed an allergy to peanuts, but that possibility is so remote that one need not make special effort to rule it out.¹² Especially considering that there is seemingly no method of investigation available to discover it, even if it did turn out true.

Let’s take stock. We have seen the obvious point that an agent’s beliefs must match their evidence. We have also seen that this is not enough because a person might fail to make a reasonable effort to gather additional evidence. However, we saw that we needed some explanation for what counts as ‘reasonable’. The explanation offered is that the amount of effort required to ensure one is not ignorant must match the degree of foreseeable risk involved in being wrong.

There is one final point to make. Notice that the question we have been considering is what might account for why it seems appropriate to say that Steve should not be ignorant that his patient has an allergy to penicillin, and yet it does not seem appropriate to say that Zaria should not be ignorant that the black tea is off-limits. In other words, the object of our analysis is what makes Steve’s ignorance impermissible and yet Zaria’s permissible. But the question of when belief (or ignorance) is permissible is, as we saw in Section 2, the primary concern of theories of epistemic justification. Indeed, the above analysis gives us at least prima facie reason to think that the difference between culpable and exculpating ignorance is just a difference in justificatory status – justified ignorance excuses while unjustified ignorance does not.

This is controversial, however. On the one hand, most epistemologists have long been stubbornly opposed to the idea that moral factors matter to one’s epistemic status, although recently a growing contingent endorse the claim.¹³ On the other hand, moral philosophers do not seem concerned to even address the question (who cares if it’s epistemically justified? What matters is whether it’s morally culpable!).¹⁴ Controversial or not, the thesis that moral and pragmatic features play a role in determining whether a belief is justified is the basis for the theory of pragmatic encroachment (Fantl and McGrath 2009; Fritz 2017).¹⁵ If this is right, then a full theory of epistemic justification will need to make room for these pragmatic features, and Biebel (2018) argues that such a theory also explains the difference between culpable and exculpating ignorance. This has considerable intuitive appeal. After all, it seems very natural in folk language to describe the difference between Steve and Zaria’s beliefs as one of justification – Zaria’s ignorance is justified but Steve’s is not, which is precisely why Steve is blameworthy despite his ignorance and Zaria is not. If this is right, then it yields the following thesis taken from Biebel (2018).

Justification Thesis

i) An agent is excused in virtue of ignorance only if that ignorance is justified and

ii) An agent is culpable despite being ignorant only if that ignorance is not justified.

Given its intuitive appeal, it is worth exploring whether this analysis also applies to group agents. In the next section, I argue that it does.

4. The group justification thesis

Interestingly, while the connection between epistemic justification and the moral evaluation of individual agents is controversial, the connection between epistemic justification and the moral evaluation of group agents is more established. Indeed, Jennifer Lackey (2021) explicitly states this as the main motivation for her highly influential account of group epistemic justification:

If we do not understand the justification of group beliefs, then we cannot make sense of our widespread epistemic attributions to collective entities – of evidence they have, or should have, and of propositions that they know, or should have known. Moreover, the justificatory status of such beliefs matters a great deal to whether groups are morally and legally responsible for certain actions and, accordingly, the extent to which they ought to be held accountable. (55, emphasis mine)

So then, in the context of group agency we may proceed as though the question of whether a group agent ought or ought not to believe a proposition is at the heart of both their epistemic status and their culpability, and the connection is strong enough that answering whether they are justified in believing will play an important and necessary role in answering whether they are blameworthy.

If Biebel (2018) is correct that a pragmatic theory of justification provides an explanation of the difference between culpable and exculpating ignorance for individual agents, and if Lackey (2021) is correct that there is the same intimate connection between a group agent’s culpability and the justificatory status of its beliefs, then the next question ought to be whether the pragmatic theory of justification employed in Biebel (2018) applies to group agents.

Fortunately, this question has already been answered. In more recent work, Biebel (2023) argues that the theory of pragmatic encroachment (the basis for what he earlier calls the pragmatic theory of justification) also applies to group agents. The main thrust of pragmatic encroachment is that two agents, call them X and Y, may share the same set of evidence, but the risks involved if X is wrong are much greater than the risks involved if Y is wrong. Hence X must do more to discharge its procedural obligations than Y. If the theory of pragmatic encroachment is true, then this difference in stakes constitutes a difference in justificatory status, hence, assuming X and Y both come to believe that P, X’s belief that P is not be justified, while Y’s belief that P is justified. Biebel’s claim is just that this is true for any agent, group or individual. After all, the stakes for one group agent can surely be different than for another group agent.

So then, we have a theory of epistemic justification that incorporates all the features that play a role in determining whether a person’s ignorance is exculpating or culpable, which is the basis for Biebel's (2018) Justification Thesis. The dialectic surrounding group epistemology, and group epistemic justification agrees, at least in principle, with the Justification Thesis’s assertion that epistemic justification is intimately related to the culpability of a group agent, and we also have good evidence that the theory of pragmatic encroachment as incorporated into a theory of epistemic justification also applies to group agents. All this combined supports the following group agent adaptation of the Justification Thesis.

The Group Justification Thesis

(i) A group agent is excused because of ignorance only if that ignorance is justified, and

(ii) a group agent is culpable despite being ignorant only if that ignorance is not justified.

This adaptation is straightforward, but there are some very important differences between individuals and group agents.

First, it is widely recognized that group agents do not act independently of their members. British Petroleum, for example, cannot drill for offshore oil in the Gulf of Mexico without some set of its members performing actions that, when done together amount to drilling for oil. This means that the kinds of epistemic activities that would discharge the group agent’s procedural obligations must also be performed by some set of the members. Note, however, that the mere fact that the group relies on its members to perform the actions that would discharge these procedural obligations, does not imply that these obligations reduce to obligations of the members. It might be, for example, that a group agent such as medical equipment manufacturer Johnson and Johnson, has a department for safety inspection, but that the board has fired the whole staff due to several lawsuits resulting from their poor performance. The fact that no members currently exist to do safety inspections does not imply that the company is not obligated to do those inspections before judging that their products are safe. Moreover, even if there are enough members to do those inspections, those members are under no personal obligation to do them. They may instead choose to resign and leave the group without having violated any of their own, nor the group’s procedural obligations.

This example might seem a bit tricky. One might think that, while the individual may resign instead of doing the inspection, other group members might be individually obligated to pick up the slack. They may, for example, have a kind of ‘rescue obligation’ where one is obligated to prevent great harm if one is in a position to do so. Someone who remains a part of the group might be personally obligated to do the inspection so as to prevent the release of defective medical products. This is, one might think, an individual obligation analogous to the obligation to save a drowning child if one is in a position to do so; not doing so seems morally wrong and this wrong belongs to the individual.

I agree that there is a mixing of personal and group-based obligation at issue, but it seems to me that those who remain at the company are only in a position to prevent defective products from being released in virtue of their position within the company. Individuals who are not group members simply cannot be in a position to prevent harm. To see this, suppose Kate is trained as an inspector but is currently retired. Suppose she is jogging by Johnson and Johnson’s factory the day the inspector very loudly quits, and she overhears that no one is left to do the inspections. Kate does not, then, inherit an obligation to do the inspections herself. Kate isn’t an employee, and while she may volunteer to do them in lieu of her training, she need not. In fact, it seems to me that Johnson and Johnson ought not allow her to do them even as a volunteer unless she can provide evidence of her training and she formally joins the company as a member, thereby becoming accountable for performing this task in her role as a member of the group. This supports Stephanie Collins' (2019b) claim that there are group-level duties that are not reducible to individual duties, even if group-level duties must be fulfilled by the individual members acting in their group-based role.¹⁶

Collins’ claim is supported even more clearly by pointing out that some group-level procedural obligations have no obvious translation to the individual level. One of these is the obligation to facilitate the sharing of information. For example, imagine Yuri is a safety inspector for a company that manufactures brake pads for automobiles. In the process of testing, Yuri discovers that a newly developed line of brake pads crack under certain cold weather conditions and therefore fail to meet industry safety standards. Yuri needs to relay this information to the decision makers ‘upstairs’ but the only way to do that is using the company’s archaic electronic messaging system installed on their out-of-date company computers. Yuri attempts to send the memo but as is typical, the system crashes several times before it finally is sent through. The managers are aware that the archaic system barely works, because it sends them error emails each time it crashes. The trouble is that they receive so many meaningless alert emails that they habitually ignore them because they are so rarely important. This time, they ignore the alerts from the system, but one of those alerts is Yuri’s legitimate warning memo. Because they ignore the alert out of habit, they are ignorant of the safety flaw, and do not incorporate it into their reasoning about whether to release the product. Since the rest of their evidence supports the proposition that the product meets safety standards, they collectively decide it is safe to release the product.

This is a case where I think the company is culpably ignorant, even though Yuri, a group member, performs the actions necessary to meet (some of) their procedural obligations and is himself not ignorant. This occurs because the methods by which information gets relayed are defective. This implies that the group agent’s procedural obligations extend further than merely having the right members in place to gather and evaluate evidence. Because a group agent may include multiple members playing differing roles, the group agent must also have a unique group-level procedural obligation to ensure that evidence can be passed smoothly to the decision makers whose role is deliberating and forming the group agent’s beliefs and intentions. I say it is unique because it is a procedural obligation in the forming of the group agent’s belief that is unique to a group setting – without the group agent there would be no obligation to ensure information can be shared effectively among the members.

However, Collins also offers the insight that any violation of a group-level duty (obligation) will always involve the violation of obligation of some set of the members. It is, after all, only possible for a group agent to discharge its procedural obligations if some critical set of members performs actions in their capacity as a group member. Such capacities are partly dictated by the role the group member occupies. Safety inspectors have certain duties as safety inspectors just as doctors have certain duties as doctors, but those duties only exist when one is acting in one’s capacity in that role. So, if Yuri is acting as an inspector on behalf of the auto parts manufacturer, in a role assigned by the company, then Yuri’s discharging his duties just is the company discharging its duties to do safety inspections. If Yuri fails to do the inspections properly, not only will he fail to discharge his own procedural obligations, but so will the auto parts manufacturer.

The obligation to facilitate information sharing might seem to complicate this because it is very often no single person’s specific job to make sure information flows smoothly, yet the company still must ensure this happens. This doesn’t conflict with Collins’ claims, however. In Yuri’s case above, for the company to address the breakdown in information flow some set of members must be assigned to address it, and the company fails to do this despite the decision makers (and therefore the group agent) being aware of the problem. Hence, Collins is correct that we can derive the violation of group-level obligation from some set of members violating their own group-member obligation. In this case, the decision makers are ignorant of the defect that Yuri discovered, and they are culpable for it because they are culpable for failing to do what was necessary to ensure the smooth flow of information. Their culpable ignorance is also the company’s culpable ignorance.

There is, however, another clear difference between group agents and their members pointed out by Biebel (2023), namely a difference in stakes. The stakes for the group agent can clearly come apart from the stakes for the individual members, particularly when those individuals are not acting in their role as a member of the group. If this is true, and if the pragmatic theory of justification is right, then it is possible for the members of a group agent to be justified in believing a proposition that the group agent is not, and vice versa. If this is right, and if the justification thesis and the group justification thesis are both correct, then the conjunction of these three claims implies that it is possible for a group agent to be culpably ignorant even though none of the members are culpable for their ignorance. Interestingly, as I will argue in the next section, this is compatible with Collins’ claims that a group’s failure to discharge its procedural obligations will always involve a failure of at least one of the members to discharge at least one of her obligations. This is important because it highlights one very common cause of group culpable ignorance.

5. Divergence in stakes

Consider the following case. Acme is a pharmaceutical company specializing in manufacturing prosthetics. They are nearing the end of the long and very expensive development of a new state-of-the-art artificial hip.¹⁷ Naturally, they have spent a great deal of time and money testing the product to ensure it is safe. A team of medical scientists in charge of these tests report to the project head, let’s call him Calvin, who has been assigned to oversee the production and sales. These scientists have come to believe that the replacement hip is safe to release to the public based on their rigorous and extensive internal tests. The evidence from these tests does indeed make it very likely that the replacement hip is safe, however, Calvin knows that company policy demands additional evidence from testing done by independent parties not funded by Acme to avoid conflicts of interest. This extremely high standard of evidence is in place precisely because of the high stakes, both legally and morally, of releasing a replacement hip that has not been thoroughly tested for safety.

If we stop the case here, we can see what Biebel (2023) claims. Plausibly, Calvin is justified in believing the replacement hip is safe from his individual perspective, even without the external testing, but Acme would not be justified. This is because the stakes are different for each. We can imagine Calvin saying to his team of scientists, ‘Look, I believe you. After all, you’ve given me excellent reason to think the hip is safe. Even so, Acme demands a higher standard, and your evidence, while an important and necessary contribution, isn’t good enough for Acme.’ And Calvin would be correct. Importantly, if the group justification thesis is correct, and Calvin is justified in believing but Acme is not, then it is possible for Acme (the group agent) to be culpably ignorant even though none of the members are culpable for their ignorance. To see this, let’s add more to the story.

Calvin knows such testing can take a lot of time and he has clients ready to go. Waiting could cost the company hundreds of millions of dollars in revenue, not to mention a hefty bonus for himself if he doesn’t get the product out as soon as possible. Moreover, Calvin has been with Acme for 25 years and has been very successful when trusting his gut in the past. This has earned him the respect and trust of the board, so that they grant him the leeway to make a judgment call. Because of this, he knows that his job is secure, and he will be shielded from blame if anything goes wrong.

Even so, Calvin is not insensitive to the moral stakes involved. He is aware that it would be horrible to sell an unsafe replacement hip and wants to avoid doing that. So, he speaks to his team of scientists seeking their assurances. They are all insistent that the data indicate it is extremely likely that the hip is safe. Given his personal stake in delivering the goods to the company’s clients coupled with the strength of the evidence, he makes the judgement call to not wait for the independent tests. He has his team write up sales pamphlets under Acme’s brand highlighting the excellent strength of the internal test results, thereby attributing the belief that the hip is safe to Acme. Moreover, we may also stipulate that the board accepts Calvin’s judgment, because Calvin has earned their trust in virtue of his past success. Acme, through Calvin, sells the replacement hips to hospitals prior to receiving the results of the independent tests. Unfortunately, when those tests come back, they reveal a problem with the replacement hip that leads to a high risk of severe and possibly deadly infection. In other words, both Calvin and Acme are ignorant – they falsely believe that the hip is safe.

I think this is a case where Calvin’s belief is justified, but Acme’s is not, even though Acme’s belief is the result of Calvin’s decision. The reason is that at the individual level – that is, outside of his role in the group – Calvin’s stakes are extremely low. After all, there is very little of consequence that will happen if he acts on his false belief that the hip is safe outside of his role in the group. As such, Calvin qua Calvin is justified in believing that the hip is safe. However, Calvin qua Acme decision maker should not base his decision on Calvin’s personal beliefs. Instead, when acting in his capacity as a decision maker for Acme, Calvin should take up the standards set by company policy. As such, he ought to set aside his own point of view and make decisions from what Kendy Hess (2014) refers to as the group’s rational point of view (RPV). Acme’s RPV includes the space of reasons that apply to the company as such, and Biebel (2023) points out that the stakes are included in an agent’s RPV (group and individual alike). Hence the stakes from Calvin’s personal RPV are much lower than they are from Acme’s RPV. This is why it makes perfect sense for Calvin to say something like ‘I believe that the hip is safe, but the company needs better evidence.’

However, in the case described above, Calvin makes an exception to company policy, and this decision is based on, or at least heavily influenced by his own RPV, rather than solely that of Acme. The problem is that the moral stakes only exist from Acme’s RPV – Calvin could not, outside of his role in the group, cause any harm to anyone by acting on his personal belief that the replacement hip is safe. His belief only has the potential to cause harm when used to make decisions on behalf of Acme. The trouble arises because he evaluates the strength of the evidence mostly from his own RPV and judges that it is good enough to address the stakes from Acme’s point of view. This is a mistake, and hence culpability for the company’s ignorance can ultimately be traced back to Calvin’s mistake. So, we might think Acme is blameworthy for selling unsafe replacement hips only because Calvin is blameworthy for not following company policy.

There is, however, some tension here that needs to be resolved. The Justification Thesis states that ignorance counts as an excuse only if it is justified, and Calvin’s personal ignorance is justified. Shouldn’t he be excused for violating company policy then? It was the result of his personal justified-but-false belief that the hip was safe, after all.

The answer, I think, is no. This is because Calvin’s ignorance about the unsafe nature of the prosthetic hip is not relevant to his blameworthiness for his decision. As was pointed out above, when Calvin acts as a decision maker for Acme he should take up the RPV of Acme, and when he decides to make an exception to company policy he is perforce, deciding to ignore Acme’s RPV. In so doing he knowingly does something he is obliged not to do. So, Calvin is blameworthy, but this is only because his wrong, ultimately, is in ignoring company policy which then leads to the company’s ignorance. His own ignorance about the safety of the replacement hip is not relevant.

Even so, the fact that Calvin knowingly does wrong, and is blameworthy for it, does not capture the full sense of blameworthiness involved in the case. Acme the company also seems blameworthy beyond just the blame we can appropriately direct at Calvin. Recall that the board has essentially granted Calvin the power to make judgments and act on behalf of the company with very little oversight or consequences. In giving Calvin permission to make judgment calls, Acme (through the board) significantly mitigates the degree of blame that Calvin is due.

This is another big mistake. The reason it is such a mistake is that it is not easy for an individual agent to take up the RPV of the group agent even in the best of circumstances. As such, very often group members lack the psychological capacities to appreciate the stakes, and this is precisely because those stakes often are very opaque from the perspective of individual agents, and many group-level stakes do not apply to the individuals. For example, the stakes involved in a US criminal trial are quite significant, and this is precisely why jurors are instructed to evaluate the evidence according to the highest standard of ‘beyond a reasonable doubt’. But it is often very difficult for the jury members to appreciate just how high that standard truly is. This is at least partly because they often struggle to appreciate the stakes – most of them have never been arrested and are unaware of the horrors of prison life or the social consequences involved in being convicted of a crime. As such, they simply lack the perspective necessary to understand just how bad it would be to inflict that on an innocent person. Hence they can struggle to take up the group’s RPV.

Fortunately, in a corporate setting company policies can often explicitly state what standards are appropriate so that the group members do not need to struggle when taking up the RPV of the group. In the example above, Acme has a policy of demanding external testing absent any conflicts of interest. This policy is explicit and specific precisely to ensure that there is no difficulty in taking up the group’s point of view – knowing whether the evidence is good enough is as simple as putting checks in boxes and not approving a product until all the boxes have been checked. But when Acme (through its board) gives Calvin permission to declare a product safe without checking all those boxes, they have essentially undermined the purpose of the policy in the first place. Hence, the lion’s share of the blame is owed to the board that represents Acme itself.

So, we can trace the company’s violation of its procedural obligations to the violation of the obligations of a set of the members of the group agent: Calvin and the board. But note that these violations of obligation were not done out of ignorance. Calvin knows he should stick to company policy, and the board knows that they should not allow that policy to be violated. Hence the company is culpably ignorant even though none of the members are themselves culpable for their ignorance of the same proposition, but this does not conflict with Collins’ claim that every violation of a group-level duty will involve the violation of some set of individual group-member duties. It is just that violations of group-level duties that cause the ignorance are not, themselves, done out of ignorance.

This is a significant result. Note that many who write on the topic of group agency and responsibility claim that groups can have mental states (or their functional equivalent) and even be morally responsible over and above the mental states and responsibility of any of the members. In fact, these days it is the more orthodox view. Groups are moral agents in their own right and thus, it is claimed, they can be blameworthy even if none of the members are.¹⁸ Indeed, this claim is often motivated by the alleged existence of so-called ‘responsibility gaps’ where there is some morally relevant harm for which we would typically blame someone, but when we look to find the blameworthy party, there is no human moral agent to be found. Some explain away this apparent responsibility gap problem by claiming that the group agent is itself a moral agent at which we can direct blame (Collins 2019a; Copp 2006). If one accepts that groups can be blameworthy even if none of the members are, and that groups can believe things (or fail to believe things) that none of the members do then one ought to welcome the idea that a group agent can be culpably ignorant even if none of the members are culpably ignorant as it is yet more gris for their inflationist mill.

However, interestingly if my arguments here are correct then one need not commit to the claim that a group can be blameworthy over and above any of the individual members in order to maintain that a group can be culpably ignorant even if none of the members are culpably ignorant. This is because the group’s culpable ignorance will still result from a failure of individual group-based obligation. It is just that the failure of obligation is a knowing one. As such, one can accept my thesis that a group agent can be culpably ignorant even if none of the members are culpable for their ignorance of the very same proposition even if one rejects the claim that a group agent’s blameworthiness is not reducible to the blameworthiness of any individual member(s).

6. Conclusion

In this paper, I set out to offer an account of the culpable ignorance of group agents. The account I offered is a straightforward extension of Biebel's (2018) Justification Thesis to group agents, yielding the Group Justification Thesis which states that a group agent is excused in virtue of ignorance only if that ignorance is justified, and a group agent is culpable despite their ignorance only if that ignorance is not justified. On this account, justified ignorance is ignorance that is properly sensitive to the agent’s evidence, the agent has met all their procedural obligations, and the stakes from the agent’s RPV determine the lengths one must go to discharge those obligations. Interestingly, some epistemic obligations exist only in a group setting, like the obligation to facilitate the sharing of information. Moreover, because the stakes for the group agent can come apart from the stakes for the members, it is possible for the group agent to be culpably ignorant even though none of the members are culpable for their ignorance, which highlights one important way group agents wind up culpably ignorant: group agents rely on their members to discharge their procedural obligations, but those members may have difficulty setting aside their own rational point of view and taking up the rational point of view of the group agent.

If this is all true, then it opens up some avenues for further research. For one thing, if I am right that culpable ignorance for a group agent comes apart from the culpable ignorance of the individual members in virtue of the differences in stakes, and that it can be quite difficult for the members to fully appreciate the stakes for the group, then one way to prevent group culpable ignorance is by addressing this difference in stakes. How can we get the individuals take up the group’s RPV and fully appreciate the stakes? What factors might get in the way? I have suggested some – the group’s internal makeup and the lack of personal connection between the members and the consequences (e.g., a jury may not know how bad the consequences of their decision really are) – but which other factors might be relevant is certainly worthy of investigation both theoretically and empirically.

Additionally, as Holly Smith (1983) points out, culpable ignorance is an important consideration that lies somewhere between fully blameworthy (to the highest degree), and fully excused. That is, someone who knowingly and maliciously causes harm with the intent to do so is clearly blameworthy to a greater degree than someone who is simply not paying adequate attention and unintentionally causes the same harm. Hence determining when group ignorance is culpable vs. when a company, say, knowingly hides relevant information will play a significant role in determining how blameworthy they are, and hence what measures we take to hold them accountable, both in the law and morally speaking.

Moreover it is worth wondering if the possible difference in the stakes that I have pointed out might give us reason to think that groups are culpably ignorant much more often than individuals. Or, if not more often, whether group culpable ignorance is more important to avoid simply because of the degree of possible harm. My sense is that the latter of these is the more likely, because, while group agents tend to have a very limited scope of action, they can potentially cause vastly more harm than any individual within that scope. As such, it might be the case that the opportunities for ignorance are fewer but the potential for harm requires much higher standards. I haven’t the space to argue for such a conclusion here, but there is clearly much more to say on this topic. My hope is only that the theory I have provided here will be useful in addressing these other questions.

Acknowledgements

Early versions of this paper were presented at the Annual Conference of the International Social Ontology Society in 2023 and at the Krakow Seminars in Law, Language, and Philosophy at Jagiellonian University in November 2023. I would like to thank Krzysztof Posłajko and Jan Rostek for helpful discussion.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was funded by ‘National Science Centre, Poland’ [grant number 2021/41/B/HS1/01961].

Notes

1 Though, see Smith (2017) who claims that the agent is culpable for the ignorance but that the ignorance still provides an excuse for the subsequent wrongdoing.

2 Discussion of the epistemic condition of moral responsibility go back at least as far as Aristotle, but I think the recent interest can be traced back to Austin (1956) and Hart (1968). However, the most influential early works were probably Holly Smith (1983) followed by Zimmerman (1997, 2008); and Rosen (2003, 2004, 2008) which spawned numerous responses. Including Angela Smith (2005); Sher (2009); King (2009, 2014); Hieronymi (2008); Talbert (2013); Fitzpatrick (2008, 2017); Levy (2009, 2011, 2017); Montmarquet (1995, 1999); Peels (2011a, 2014); Biebel (2018); Arpalay (2003); Harman (2011, 2015); Wieland (2017); and two anthologies containing works by many of the authors just mentioned and others in Peels (2017) and Wieland and Robichaud (2017).

3 Though, some few in the business ethics literature have discussed excusing corporations for wrongdoing in virtue of the competitive environment in which they operate. See, for example, Kavka (1983), Heath (2018), and Silver and Garofalo (2024).

4 For a lively back and forth on the nature of ignorance see Le Morvan (2011, 2012, 2013) who takes the view that ignorance is a lack of knowledge and Peels (2010, 2011b, 2012) who takes the view that ignorance is a lack of true belief.

5 See Peels (2014).

6 These are often called ‘divergence cases’.

7 Hakli and Mathiesen get their version of ‘acceptance’ vs. ‘belief’ from Cohen (1992). Gilbert does not endorse this version of ‘acceptance’.

8 This interpretationism is derived from Dennett's (1987) intentional stance.

9 It might be argued that Steve and Zaria do not have a false belief so much as they simply lack any beliefs at all. I don’t think this would affect the analysis much. A doctor who has done their due diligence may not be willing to claim the patient is not allergic, but they will say that they are as certain as can reasonably be expected under the circumstances. Steve is clearly not as certain as can reasonably be expected. I will stick with the language of “false belief” for ease of exposition.

10 Rik Peels (2017) later calls these ‘intellectual obligations’

11 Some of the most ardent defenders of evidentialism such as Feldman and Conee (1985) and Feldman (2000) claim that this is all that is required for belief to be justified.

12 Of course, there might be some bad outcomes that are just so terrible that we simply should never risk it even if the possibility is very remote. We should not ‘bet the farm’ so to speak, even if the bet seems like a sure thing (see Hawthorne 2004).

13 See, for example, Fritz (2017); Schroeder (2012); Ross and Schroeder (2014) and Moss (2018).

14 Rosen (2008) is an exception here as he argues explicitly for the claim that epistemic justification cannot be the basis of culpable ignorance because one can be justified while still blameworthy and one can be excused despite failing to be justified. Biebel (2018) responds directly to Rosen’s argument at length and so I will not rehearse the argument here. See also Biebel (2023, 8–9) for further reason to think the epistemic and moral are not so rigidly divided.

15 The theory of pragmatic encroachment is also endorsed as relevant to knowledge ascription by Stanley (2005) and Hawthorne (2004) among others.

16 I would like to thank an anonymous reviewer for pressing me to address this.

17 This case is a fictional imagining based on a real-world case wherein DePuy Orthopedics, a subsidiary of Johnson and Johnson pharmaceuticals, released a defective replacement hip resulting in thousands of lawsuits. https://www.reuters.com/article/us-johnson-johnson-settlement/jj-agrees-to-pay-about-1-billion-to-resolve-hip-implant-lawsuits-bloomberg-idUSKCN1SD1YO/

18 See, for example, List and Pettit (2011); Lackey (2021); Collins (2019b, 2023); de Haan (2020, 2023); Hess (2014); Biebel (2023) and Tollefsen (2015) among many, many others.