The proper social epistemic functioning of science requires that public avowals in science fail to satisfy the norms we surveyed at the beginning. We take it to be uncontroversial that if there will be sustained and organised inquiry, any collective inquiry of the sort that science in fact is, there must be successful communication of the latest results and ideas about them. We will assume that the norms governing utterances communicating those results and ideas must reflect and endorse this state of affairs. Any norm for making avowals which rendered it impermissible to put forward just those claims which constitute necessary features of collective inquiry would be misaligned with the purpose of the activity it is supposed to govern. The scientific community has developed norms, both explicit and implicit, which govern utterances that are appropriate for communicating scientific findings to other scientists. We shall argue that these do not look like any of the purported norms of assertion, and for good reason.
First, note the role of the division of cognitive labour. It is well recognised in philosophy of science that such a division is an important strategy for collective inquiry (Kitcher, 1990; Strevens, 2006). Scientists invest time and resources on different approaches to a research domain. For example, during early investigations into γ and X-ray behavior, different researchers used different techniques to study γ and X-ray behavior. This division of labor is often not explicitly planned; rather the limited resources, credit and incentive structures of science encourage scientists to pursue new avenues of research and use different methodologies (see Kummerfeld & Zollman, 2015; Zollman, 2018).
This division of cognitive labor means that during periods of active inquiry, scientists will often be publishing discoveries which are seemingly in conflict with each other. This in fact occurred during the γ and X-ray episode. Bragg investigated the particle-wave problem using, what we call today, hard X-rays, which are higher in energy. Charles Barkla (Bragg’s leading critic and fellow Nobel laureate) investigated using, what we call today, soft X-rays, which are lower in energy. This is because Bragg and Barkla used slightly different techniques for producing X-rays, due to the limited availability of radioactive source materials. They also used different experimental techniques to look at X-ray behavior (see Wheaton, 1983). This led them to come to radically different conclusions about the nature of X-rays. Bragg’s experiments seemed to show that hard X-rays behaved like particles, because he observed large transfers of energies which is indicative of a particle hitting others. Since γ rays are even higher in energy than X-rays, Bragg was able to observe particle-like behaviors. Barkla, on the other hand, observed his soft X-rays to behave more consistently like waves.
At most one of these models of X-rays could have been correct (as it turned out neither are correct), yet both scientists had to publicise their work if scientific progress were to be made. Public avowals which break factive norms are thus epistemically beneficial and ought to be encouraged. Without them we would have never found out eventually that how X-rays are produced changes the energy of the X-rays—this was a fact we only discovered because Bragg and Barkla came to different results due to how they were generating X-rays in their labs.
Moving beyond the necessity of the division of labour, the case study can be placed in the context of a broader historical argument. This builds on the kind of observations that inspired the pessimistic meta-induction (which typically focused on physics) and Stanford’s (2006) problem of unconceived alternatives (which focused on the history of biology). This can be supplemented by recent work on replicability in psychology (Open Science Collaboration, 2015) and cancer research (Begley & Ellis, 2012). Frequent false public avowals are a necessary part of scientific progress, especially in areas of active inquiry. While scientific realists have argued that the history of science is also full of epistemic successes (for example, see Fahrbach, 2011, 2017), it is nevertheless the case that individual contributions to science in areas of active inquiry are prone to error and are continuously replaced by new errors.
These historical arguments can be supplemented with theoretical arguments. All our inquiry inevitably takes place in a situation of scarce resources and competing demands on our attention. Where this is the case, we have theoretical reason to believe that scientists will not gather as much data as a disinterested epistemic planner would have them gather (Heesen, 2015, 2018). Further, some of our inquiry takes place in contexts where there are weak signals, sparse data, and considerable difficulty in running replications. When this is so, limitations to what is possible through statistical inference give us reason to suppose that most of our published findings will be false (Ioannidis, 2005; Romero, 2016).
These facts together suggest we have not found a better way of communicating findings than one in which there are frequent false avowals. As the X-ray case shows, it is crucial for false theories to be put forward, so to encourage more research; Bragg’s neutral pair theory of X-ray was directly responsible for sparking interest in studying radioactivity among many other physicists. If we agree with versions of the pessimistic meta-induction wherein new scientific theories tend to demonstrate pervasive reference failure in their previous iterations, it means that almost all scientific public avowals turn out to be false.
What does the necessity of dividing labour, and the pessimistic conclusions about present and past science, tell us about our norms for scientific public avowals? We think they quickly rule out factive norms. For factive norms would judge inappropriate a great part of those avowals that are necessary for the progress of any collective inquiry which must divide labour and proceed by correcting errors. It is not just that we in fact often will say false things in the course of inquiry, but rather that inquiry could not proceed in a way that was even remotely successful if we did not do so. Given the ubiquity and necessity of such things to collective inquiry this renders factive norms simply inappropriate as a means of deciding what conclusions ought to be put forward from individual contributions to the wider project. The norms of inquiry should not rule out necessary parts of the process.
In fact, we note, there have been many who have argued that factive norms would not even be appropriate for reporting the results of “completed science”, whatever that might be. Consideration of the nature of physical laws (Cartwright, 1983), scientific models (Frigg & Nguyen, 2016), or what it would be to achieve understanding through science (Dellsén, 2017; Elgin, 2007), have all led philosophers to conclude that the terminus of scientific inquiry need not be taken to be possession of true beliefs. Avoiding requiring adherence to factive norms in the midst of inquiry thus looks even more attractive when one considers that it may not even be the desired end state.
With one more assumption we think the arguments just reviewed also rule out justification norms as appropriate for governing individual contributions to collective inquiry. Our additional assumption is that whatever notion of justification is at play here it meets the total evidence requirement (Good, 1967). Our total evidence includes the information just provided about what kind of process science is. Not just how generally reliable our inquiry is, but also how reliable it is for hard problems in particular, and how reliable we are ourselves as individual or communal epistemic agents. For any one paper, the reasons one could produce in favour of its central claim could well be outweighed by these competing second-order considerations. Hence if one is to communicate the results of sustained inquiry on hard problems then one cannot limit one’s avowals to those a total-evidence respecting epistemic norm would permit. In the X-ray case, Bragg knew that the total evidence, on balance, did not support his corpuscular theory, but he nonetheless published and defended it publicly. This was not condemned by his contemporaries, rather they took his theory seriously despite everyone being aware that there exists compelling research which did not support it. So the surveyed justification norms of assertion cannot govern scientific avowals if they are to foster and permit publishing surprising findings or results, and we believe would not appropriately do so for any collective inquiry into a difficult or obscure matter.
Finally, one might think that none the less the belief norms may apply to public avowals. For all that has just been said, scientists may believe their claims to be true or justified, as long as scientists are ignorant of this history or these theoretical results. Or, perhaps, as long as they retain their faith in their own particular claims even while being aware that in general studies like theirs are not reliable. So long as scientists can maintain belief in themselves, we could require they only avow when they believe all they have said. Indeed, the International Committee of Medical Journal Editors might be requiring this when it says “[a]ll members of the group named as authors… should have full confidence in the accuracy and integrity of the work of other group authors” (ICMJE, 2013: 3).
We think such a retreat to mere belief or second order belief would be a mistake. For one thing, there is a large and persuasive literature in the philosophy of science detailing situations wherein inquiry may proceed well without scientists believing their claims (e.g. Kapitan, 1992; Dawes, 2013; Cabrera, 2018; Dellsén, 2018; Palmira, 2020, Fleisher, 2020). But over and above this we do not think that bad faith can be required of people. Quite generally we do not think a good social system can require participants to be ignorant as to the nature of their activities and their history, or require that they reason irrationally upon being informed about these things. False ideology or absurd arrogance should not be a prerequisite for inquiry. In this context it means that we cannot require as a condition of successful public avowal in science that scientists may not learn the various historical or theoretical facts that would undermine their faith in their assertions. Scientists do not need to believe that they are epistemically special in order to successfully participate in science.
If one agrees that bad faith cannot be mandatory then one can move from the above arguments to a rejection of both the justification norms and the belief norms. Scientists may well know of themselves that they are engaged in an activity which is not reliable for the kind of results they report. They thus may well not think their investigations are sufficient reason to believe their conclusions, or would suffice for justification all things considered. Their total evidence (inclusive of second order evidence) could not justify this, and they form their credences appropriately. Hence they may not believe their results at all, or believe them to be justified, and yet still properly avow. We ought not rule out as inapt these avowals of scientists who take the full measure of their epistemic situation. Hence building on the social epistemology of dividing cognitive labour, the pessimistic meta-induction and related theoretical results, we have been able to generalise our case study. This ruled out the factive, justification, and belief norms as the proper means of deciding what sort of conclusions are appropriately put forward as individual contributions to collective inquiry.
In some sense, of course, one could avoid this by just making weaker claims. Rather than saying X causes Y, one could say that according to one’s data X causes Y. The weaker claim might well be true, or at least justified even in light of total evidence, and perhaps ought to be believed in any case. Certainly it is the case that many scientific result claims are made in a hedged way in something like this fashion. Of course, at present we do not think that scientists always obey this norm of making only such weaker statements. However, if one was a strong advocate of one of the norms surveyed one could insist that strictly speaking only such weaker avowals would be proper, and fault those scientists who do not live up to this.
We note this possibility simply to note that it is a path we shall not explore here. We are concerned here with what should be expected of individual contributions to a broader inquiry, not really the exact forms they take. We are only making an analogy to the norms of assertion literature, not trying to weigh in on speech act theory. In the context of scientific research, we cannot only concern ourselves with events in a particular laboratory, or evidential relationships between particular data sets and a proposition. We also must put forward candidates for scientific communal uptake and potential targets for future inquiry. This goes all the more when one considers, as we have largely set aside, that one wishes ultimately to make use of scientific claims as the basis of public policy, where their external validity is what is of paramount importance (c.f. Cartwright, 2012 §4).
We do not doubt that it would be possible to reform scientific communication behaviour such that one sticks to scientific public avowals that are proper according to one of the surveyed norms by insisting on appropriate hedging. One could then just say explicitly (perhaps in so many words): “we also suggest such and such as a candidate for further investigation.” Or one could understand it to be an implicature that if Lab 1 reports that its inquiry suggested that X causes Y, Lab 2 might profitably investigate whether its own inquiry would suggest as much also. But our concern is not really with the precise linguistic form such claims would take so much as the social uptake amongst scientists. However results are conveyed, scientists must decide what claims are worthy of further tests (Friedman, 2020; Thorstad, 2021). Our point is that it would be inappropriate for scientists to insist that (in the absence of fraud or mistake or misfortune) these pursuit worthy claims must be true, or justified, or believed to be as much by their proponents. Further, we shall briefly outline a proposal for an actual norm that could appropriately govern, and is perhaps governing, scientific avowals in the next section, based on an analysis of the social epistemology of science. If one understands what norms avowals must satisfy in the good case, one can better adjust one’s attitude to them, without needing scientists to change their manner of communication.