This is a part of the symposium on socially extended knowledge
In Support of Distributed Scientific Knowledge
By Orestis Palermos
In what follows, I would like to focus on Duncan Pritchard’s novel idea of Socially Extended Knowledge and how it contrasts with the notion of Distributed Knowledge. Here is hopefully an accurate summary of Pritchard’s interesting proposal:
Socially-Extended Knowledge: Knowledge generated by an individual’s cognitive processes, to which the “collaborative element of the social dynamic has become integrated.”
Despite its strongly social nature, Pritchard argues that Socially-Extended Knowledge is individual knowledge, by invoking a prominent virtue-theoretic approach, according to which knowledge is cognitive success that is significantly attributable to one’s cognitive agency. Pritchard explains that, in cases of Socially-Extended Knowledge, it is the individual agent who possesses the socially extended cognitive process, such that any resulting cognitive success can be significantly attributed to the corresponding individual agent. Thus, Socially-Extended Knowledge is individual knowledge.
The kind of case that Pritchard uses to motivate his view involves two scientists who engage in a mutually beneficial collaboration, whereby each may incorporate the other in their cognitive routines as a well-trusted expert to be regularly consulted for accomplishing their respective epistemic aims. While continuously relying on each other, each resulting cognitive success can be readily attributed (to a significant extent) to a specific individual.
Pritchard notes that this kind of socially extended—but still individual—knowledge can be contrasted with the alternative of Distributed Scientific Knowledge. On this alternative proposal, when individuals collaborate, there exists an emergent group agent to whom the resulting knowledge belongs. Given how metaphysically demanding this claim is, Prichard rightly notes that to accommodate knowledge generated by epistemic collaborations one may be better off employing his Socially-Extended Knowledge, which holds instead that the resulting knowledge is individual knowledge. The onus, therefore, is on proponents of Distributed Knowledge to explain why epistemologists may need to retain their radical idea: Why can’t Socially-Extended Knowledge accommodate all collaborative knowledge?
Pritchard is of course careful to point out that this is only a prima facie challenge for proponents of Distributed Knowledge, since, in certain cases, it might still be necessary to speak of a group-level epistemic agent. I think this is correct and while it is impossible to here provide a meaningful metaphysical argument for Distributed Knowledge, it is worth highlighting why it may be useful to retain this challenging idea from a practical perspective.
Towards the end of his paper, Pritchard mentions Karin Knorr-Cetina’s ethnographic study of complex scientific collaborations in High Energy Physics (HEP), because of its strong influence in motivating the existence of Distributed Scientific Knowledge. Though Pritchard suggests that much of what is going on within the complex collaborations that Cetina is interested in may ultimately be accounted in terms of socially facilitated and Socially-Extended Knowledge (hence individual knowledge), he takes notice of Cetina’s remarkable claim that no individual is overall responsible for the management and organization of HEP experiments: According to Cetina, experiments are run instead by discourse—i.e., intricate interactions between scientists (achieved on the basis of multiple and diverse channels of communication) that provide the experiments with “a sort of distributed cognition,” which allows scientists to efficiently self-organise. In fact, Cetina goes so far as to speak of the “erasure of the individual as an epistemic subject” and even mentions a new kind of epistemic subject that is “collective and dispersed.” Given such claims, it is indeed tempting to think of the responsibility for any resulting collaboratively produced success as significantly attributable to the whole experiment and the corresponding success as a distributed kind of knowledge.
Does this mean that, unlike what happens in the collaborations Pritchard has in mind, no responsibility (in the form of either blame or credit) can be attributed to the members of HEP experiments? Sometimes—perhaps often—when significant (either negative or positive) contributions can be traced back to specific individuals, attributing individual responsibility may still be possible. But whatever the case about the possibility of attributing individual responsibility may be, Cetina’s study indicates that we should ensure never to neglect attributing epistemic responsibility at the collective level too.
I think this important and to see why it will help to focus on collaborative failures. When collaborations fail, attributions of individual responsibility should not exhaust our inquiry into the collaboration’s shortcomings. This is because Cetina’s study seems to suggest something along the following lines: Were the communication pathways between scientists optimal, discourse would have detected individual failures and appropriately dealt with them before negatively affecting the collaboration. If something like this is correct, then by failing to attribute responsibility at the collective level (and by attributing it instead to individuals alone), we would be risking addressing all individual shortcomings (perhaps by replacing every individual member), while ignoring important issues with the collaboration’s communication pathways—an omission that would effectively render the collaboration susceptible to similar structural shortcomings in the future.
With this in mind, and going back to Pritchard’s analysis, I do not doubt that Socially Extended Knowledge is an important addition to the possible ways in which knowledge can be social in nature. Indeed, it is likely that much of scientific knowledge—especially that resulting from collaborations where each member can be held significantly responsible for their part in the final product, and the social structure is supposed to play no significant role in quality control—is socially extended much in the way Pritchard suggests. However, it is also likely that many scientific cognitive successes—produced by highly distributed scientific research teams such as those within HEP experiments—are attributable to group-level processes and agents. As metaphysically challenging as this theoretical possibility may be, its practical ramifications for the successful organisation of complex epistemic collaborations suggest that we should continue entertaining it.