Collective Belief Explicated: Replies to Wallace and Smith

We are grateful to Rodrick Wallace (2020) and Nicholas D. Smith (2020) for their questions and comments on Bruno and Fritzman (2020). Wallace’s comments largely complement our article. Agreeing with Atlan and Cohen (1998), he defines cognition as the choice of an action among available alternatives that reduces uncertainty. As a consequence, he argues that cognition ubiquitously pervades all biological phenomena. Groups thus cognize because they chose actions. Although Wallace does not explicitly claim that groups can be regarded as stand-alone individuals, at least when considering group cognition, his framework seems to suggest this consequence… [please read below the rest of the article].

Image credit: Tirachard Kumtanom from Pexels

Article Citation:

Bruno, Michael G. and J.M. Fritzman. 2020. “Collective Belief Explicated: Replies to Wallace and Smith” Social Epistemology Review and Reply Collective 9 (9): 15-19.

🔹 The PDF of the article gives specific page numbers.

Articles in this Dialogue:

❧ Bruno, Michael G. & J. M. Fritzman. 2020. “Collective Belief Defended.” Social Epistemology 1-18. doi: 10.1080/02691728.2020.1761479.

❧ Smith, Nicholas D. 2020. “Collective Belief Questioned.” Social Epistemology Review and Reply Collective 9(7): 58-63.

❧ Wallace, Rodrick. 2020. “A Comment on Collective Belief.” Social Epistemology Review and Reply Collective 9(7): 55-57.

Turning to Smith’s questions, Bruno and Fritzman (2020) do not argue that group beliefs actually exist, but that “there is no conceptual or coherency objection to group belief, and that rejectionist arguments and appeals to intuitions are not convincing” (1). Our discussion of the examples in Smith (2016), and our own examples, show that his interpretations are contestable.

Smith (2016) maintains that, metaphysically speaking, there are no group beliefs. He argues against anti-summativism by maintaining that groups cannot have beliefs because they cannot acquire beliefs. Specifically, groups cannot learn. So, for example, Smith claims that a class as a whole has not fully learned the multiplication tables for the whole numbers less than 100 if no student has fully learned them.

We argued that, even if the class in Smith’s example has not learned the multiplication tables, the example would show only that this group did not learn them on that occasion. The example does not prove that it is impossible for groups to learn. We also provided a different example of a class that plausibly has fully learned the multiplication tables, although no student has fully learned them (Bruno and Fritzman 2020, 4).

Responding to our example, Smith (2020) writes: “This may simply be a case where appeals to intuitions do not generate consensus, since Bruno and Fritzman seem to take this case as intuitively going their way. Such “dueling intuitions” also seem to occur in each of the other differences I found between their view and mine” (59).

Dueling Intuitions

We do not take this case, or any other, as intuitively going our way. Rather, how those cases intuitively go is contestable, so dueling intuitions cancel each other, and epistemic individualism cannot be supported by intuitions or arguments that presuppose it (Bruno and Fritzman, 2020, 16, note 18). Confronted with dueling intuitions, philosophers cannot double down and question-beggingly providing further examples and thought experiments that presuppose those very intuitions. Nor can philosophers simply agree to disagree. They should instead abandon those intuitions and employ different modes of argumentation, ones that will frequently involve appealing to empirical evidence.

Smith (2016) also criticizes Goldman’s discussion of a form of summativism which holds that a group can have a mind of its own (Goldman 2015, 226-7). This is so, according to this form of summativism, because a group’s propositional attitudes can weakly and holistically supervene on the propositional attitudes of its members. Goldman explains that this supervenience relation can be mapped using a Belief Aggregation Function (BAF). The mapping of a BAF goes to the belief of a group, from the beliefs of its members. The BAF is majoritarian iff the group believes P when a majority of its members believes P, supermajoritarian iff the group believes P when a supermajority of its members believes P, and dictatorial iff the group believes P when a “single fixed member” believes P (Goldman 2015, 226). A BAF can be implicitly held be a group’s members, manifested through their behavior, or a BAF can be explicitly acknowledged by them.

To illustrate a group’s propositional attitudes supervening on those of its members, Goldman provides an example of 100 guards at the art museum. Each of three groups of 20 guards suspects that a different guard plans to steal a famous painting. The first group of 20 guards suspects Albert of planning the inside heist, a second group suspects Barnard, and the third group suspects Cecil. In his example, Goldman stipulates that each of the 60 suspicious guards deductively infers the existential proposition T: “There is a guard who is planning such a theft” (Goldman 2015, 227). The group of those 60 guards collectively believes that T, since a BAF would map the propositional attitude regarding T of the group of 60 suspicious guards onto the propositional attitudes concerning T of each of those guards.

Although Smith recognizes that “each member also makes the obvious inference that there is a guard who is planning such a theft” (Smith 2016, 5), he maintains that the group of 60 suspicious guards does not collectively believe that T, since each group of twenty guards suspects a different individual.

We count that, when plausible assumptions regarding social dynamics are added to Goldman’s example, the suspicious guards would recognize three things, i.e. that they might be suspicious of the wrong guard, that the guard who is planning the heist might have an accomplice, and that the guard who is planning the heist might intend to steal other artworks too. Such heightened suspicions would likely result in an enhanced vigilance in the 60 guards as a whole: the first group of 20 guards would not only suspect Albert, a second group would not only suspect Barnard, and the third group would not only suspect Cecil. The group of 60 suspicious guards would collectively believe that T. We conclude that “it is plausible to think that the interpersonal practices and mechanisms of psychological influence of Goldman’s guards lead to patterns describable as following a BAF” (Bruno and Fritzman 2020, 5).

Smith (2020) writes: “Nothing in the example (even once we include recognition of the social dynamics) has to be taken in a way that attributes psychological attributes to the entire group, rather than to (some) members of that group. In brief, it seems to me that however we conceive of the social dynamics, it will be in a summative way—and that is not what is needed for collective belief of the sort that Bruno and Fritzman (or Goldman) wished to defend” (60).

Smith (2020)—and, in retrospect, likely Smith (2016)—does not fully recognize that Goldman is discussing summativism. Goldman’s example illustrates how the propositional attitudes—not psychological attributes, as Smith writes—of the group summatively supervene on the propositional attitudes of its members.

To support his brief against group belief, Smith (2020) introduces a new example: he suspects that a student plans to steal one of his books (60). He claims that this suspicion would not lead to his enhanced vigilance regarding other students. This would be so, though, only if Smith knows that this student does not have an accomplice who might steal the book. Otherwise, he would have enhanced vigilance regarding the student’s friends and classmates. If the student does not have an accomplice, however, this example cannot support Smith’s brief, as it involves only himself and the individual student. The example has no group whose propositional attitudes could supervene—through a majoritarian, supermajoritarian, or dictatorial BAF—on the propositional attitudes of its members.

Identity and Kinds

Smith (2020) concedes that arguing from functional equivalence to kind is not always fallacious, whereas Smith (2016) maintains that such inferences are always fallacious. He nonetheless maintains that arguing from functional equivalence to kind is not always correct: “Even though my cats are remarkably good at catching and killing mice, it nonetheless seems to me that it would be a dreadful metaphysical error to conclude from that fact that cats and mousetraps and the same kinds of thing” (Smith 2020, 61).

Cats and mousetraps can be the same kind, however, distinct from many other kinds to which they do not both belong. A thing will belong to multiple kinds, in general, and some of those kinds will vary over time. In addition to a mousetrap, a cat can belong to the following kinds: companion, Zoom meeting interrupter, eater of tuna on the unattended table, lap warmer, waker-upper, and purr puppy. Similarly, ox carts, horse-drawn carriages, cars, trains, and airplanes belong to the modes of conveyance kind, but only the latter three belong to the self-propelled vehicles kind.

Smith writes that “as my example of cats and mousetraps shows, the inference from the fact that two entities seem to perform the same function to the conclusion that they are the same kind of thing does not work when the nature or essence or what-it-is-ness of one of the relevant entities is not reasonably taken to be conceivable or analyzable in purely functional terms” (Smith 2020, 61).

This incorrectly regards nature (or essence or what-it-is-ness) and kind as identical. It also incorrectly assumes that an entity can be only one kind of thing. As noted above, though, an entity usually belongs to multiple kinds. While some kinds may not be fully analyzable in functional terms, they nevertheless may be partially analyzable. This partial analysis can defeasibly support the inference that an entity belongs to a particular kind, while allowing that this entity may also belong to other kinds.

Smith further maintains: “I take their argument about functional equivalence simply to beg the metaphysical question of whether or not beliefs, desires, and the other sorts of mental states are to be conceived as essentially functional states. It is one thing to accept that mental states have functions; but it is another thing altogether to assume that they are what they are simply because of the functions they perform (such that anything else that performed such functions would be the same kinds of things)” (Smith 2020, 61-2).

Towards a Consensus

Our argument about functional equivalence begs no question. It assumes neither that mental states are essentially functional states, nor that mental states are mental states solely because of their functions. Given that inferences from functional equivalence to kind are generally correct, our argument is that the burden of proof is on those who claim that any specific inference is incorrect. Specifically, the burden of proof is on those who maintain that the functional states of groups are not mental states, even when those states are functionally equivalent to the mental states of individuals. That is, the burden of proof is on those claiming that the inference from functional equivalence to kind is incorrect for mental states. If mental states are not mental states only because of their functions, what else is required? If something other than function is required, does that preclude groups from having mental states? Perhaps groups as well as individuals can have whatever else is required to have mental states.

Smith (2020) responds that the discussion of cognitive singularity in Smith (2016) does show this. Cognitive singularity would occur when a group, one that exists for only a short duration, has only one belief during its brief existence. This is not possible for individuals. Smith considers the possibility of cognitive singularity to be so bizarre that it undermines the entire notion that groups could have beliefs. He notes that we propose that the ability of a group, unlike an individual, to only ever have a single belief might be one way in which group belief differs from individual belief, and that the requirement that the features of group belief must be the same as those of individual belief begs the question by presupposing epistemic individualism (Bruno and Fritzman 2020, 13).

However, we also argue that considerations of cognitive holism would lead to the rejection of the possibility of cognitive singularity. A search committee could believe, for example, that an offer of employment should be extended to a specific individual. In order to have that belief, however, the committee must have many other beliefs. The committee must also have beliefs regarding what the position is, what the terms of the job offer are, which of those terms are negotiable, what the qualifications for the position are, how that individual’s qualifications surpass those of the other candidates, what the committee will do if that individual declines the offer, and whether the committee itself may extend the offer. Each of those beliefs, moreover, would require many others.

We share Smith’s hope for a future consensus. Such a consensus is possible, however, only if epistemic individualism is not presupposed, and if it is recognized that dueling or clashing intuitions cancel each other, requiring other arguments. Smith seems to agree, as he concludes his comments by suggesting that “it is perhaps best simply to identify where such clashes are, in the hopes that others can help to find ways through them to achieve consensus in some other way” (Smith 2020, 62).


Atlan, Henri, and Irun R. Cohen. 1998. “Immune Information, Self-Organization, and Meaning.” International Immunology 10 (6): 711-717.

Bruno, Michael G., and J.M. Fritzman. 2020. “Collective Belief Defended.” Social Epistemology.

Goldman, Alvin. I. 2015. “Collective and Institutional Epistemology.” In A. I. Goldman and M. McGrath (Eds.), Epistemology: A Contemporary Introduction. 224-248. New York: Oxford University Press.

Smith, Nicholas D. 2020. “Collective Belief Questioned.” Social Epistemology Review and Reply Collective 9 (7): 58-63.

Smith, Nicholas D. 2016. “The New Collective Social Epistemology and the Old Fallacy of Functional Equivalence.” The International Journal of Assessment and Evaluation 23 (1): 1-9.

Wallace, Rodrick. 2020. “A Comment on Collective Belief.” Social Epistemology Review and Reply Collective 9 (7): 55-57.

Categories: Critical Replies

Tags: , , , , , , , , , , ,

Leave a Reply