by George Barimah, Ina Gawel, David Stoellger, and Fabio Tollon*
When thinking about the claims made by scientists you would be forgiven for assuming that such claims ought to be true, justified, or at the very least believed by the scientists themselves. When scientists make assertions about the way they think the world is, we expect these assertions to be, on the balance of things, backed up by the local evidence in that field.
The general aim of scientific investigations is that we uncover the truth of the matter: in physics, this might involve discovering a new particle, or realizing that what we once thought was a particle is in fact a wave, for example. This process, however, is a collective one. Scientists are not lone wolves who isolate themselves from other researchers. Rather, they work in coordinated teams, which are embedded in institutions, which have a specific operative logic. Thus, when an individual scientist “puts forward” a claim, they are making this claim to a collection of scientists, those being other experts in their field. These are the kinds of assertions that Haixin Dang and Liam Kofi Bright deal with in a recent publication: what are the norms that govern inter-scientific claims (that is, claims between scientists). When scientists assert that they have made a discovery they are making a public avowal: these are “utterances made by scientists aimed at informing the wider scientific community of some results obtained”. The “rules of the game” when it comes to these public avowals (such as the process of peer-review) presuppose that there is indeed a fact of the matter concerning which kinds of claims are worthy of being brought to the collective attention of scientists. Some assertions are proper and others improper, and there are various norms within scientific discourse that help us make such a determination.
According to Dang and Bright we can distinguish three clusters of norms when it comes to norms of assertions more generally. First, we have factive norms, the most famous of which is the knowledge norm, which essentially holds that assertions are only proper if they are true. Second, we have justification norms, which focus on the reason-responsiveness of agents. That is, can the agent provide reasons for believing their assertion. Last, there are belief norms. Belief norms suggest that for an assertion to be proper it simply has to be the case that the speaker sincerely believes in their assertion. Each norm corresponds to one of the conditions introduced at the beginning of this article and it seems to naturally support the view that scientists should maintain at least one (if not all) of these norms when making assertions in their research papers. The purpose of Dang and Bright’s paper, however, is to show that each of these norms are inappropriate in the case of inter-scientific claims.
Nevertheless, we will argue that the authors have strong readings of each of the norms, and that such interpretations are neither necessary nor desirable. We claim there are good reasons not to read the norms in their strongest senses. Specifically, that the examples they use for each norm do not seem to capture a more benevolent meaning of that norm. Below we will attempt to reconstruct the authors’ argument, following which we will present our criticism.
To make their case the authors use the example of William Henry Bragg, an early 20th century researcher in the field of radioactivity. In the early days of research into radioactivity, physicists had not yet settled on whether X-rays were in fact “rays” at all. That is, there was a debate as to the fundamental nature of X-rays: were they waves or particles? Bragg developed a particle account of radioactivity where he argued against the dominant wave theory at the time. His proposal was that X-rays are in fact made up of two oppositely charged particles, a positive α-particle, and a negative β-particle, which formed a neutral pair. He disputed the popular wave theory (or “aether pulse theory”), and it was his contention that such theories could not fully account for the results he had observed.
Bragg was therefore advancing a particle theory of X-rays, which he made clear, in the form of public avowals, in many publications and letters to Nature from 1907 to 1912. What happens when we evaluate this example in light of the norms of assertion outlined earlier? First, factive norms: were his assertions true? As it turns out, Bragg was wrong: X-rays are not material particles and there are no neutral pairs, and today we know that X-rays have wave-particle duality and are composed of photons.
What about justification norms? Well, it seems that Bragg was aware that his theories were unlikely to be corroborated. Even at the time Bragg was writing, his theory was far from the scientific consensus in the physics community, and many believed his theories to be improbable. There was strong evidence for the “pulse theory”, and Bragg understood that he was going against what was widely accepted in physics at the time.
Lastly, according to Dang and Bright, Bragg did not even believe his own theory. In personal letters it appears that Bragg was quite circumspect regarding the tenability of his hypotheses. For example, in a letter to Arnold Sommerfeld, Bragg stated that “I do not wish to press this unduly or be dogmatic about it. It seems to me to be the best model to be devised at present, and I have no right to claim more”.
On the basis of this, Dang and Bright conclude that while Bragg’s public avowals “were false, unjustified on total evidence, and were not believed, we want to maintain that these avowals did not violate the norms of inter-scientific communication”. Instead of viewing the Bragg case as indicative of various epistemic issues in scientific communication, the authors argue that this is in fact a feature and not a bug of scientific public avowals. While we believe that Dang and Bright’s argument is an intellectually stimulating one, we have a general issue with regards to their operationalization of norms of assertion, and then some specific issues with the three norms they focus on. We first present our more general criticism and then follow this up with critiques of the specific norms.
To recap: The authors claim that Bragg, in his continuous defence of the neutral material particle theory, had made public avowals that he did not believe, turned out to be false, and were not properly justified given the available evidence.
At the general level, however, the authors seem to apply significantly different norms in the example than they have laid out in their introduction. For example, if one was trying to apply the strong versions of each of the norms to any inter-scientific public avowal (or any assertion made by scientists, for that matter) few scientists, if any, would actually be adhering to them. Nonadherence here would not mean one adopts a position of dogmatic scepticism, but arguably reflects a well-tempered epistemic humility. Such humility would acknowledge that current theories and assertions may turn out to be wrong in the light of further evidence in the future. On the other hand, if one takes a milder framing of the norm, e.g., that one ought to retract assertions that turn out to be false, Bragg’s gradual abandoning of a dogmatic defence of his position may be read in his favour. Especially given the fact that the alternative aether pulse theory also turned out to be false, to simply capture whole theories as either true or false, misses the point. Bragg could reasonably assert, if anything, that his theory was arguably as true or as false as the available alternatives until 1912. Interestingly, after 1912 his defence seemingly stopped. We will now supplement this general critique with some specific remarks about each norm, starting with factive norms.
In talking about factive norms of assertion Dang and Bright fail to acknowledge a distinction between saying something that one knows to be false and saying something that is proven to be false without knowing it is false. In the former case one lies, but in the latter one doesn’t. It seems what the norms of assertion guard against is lying or asserting misleading information, if that is the case, then it is possible to accommodate norms of assertion in inter-scientific communication, at the very least, to guard against lying to colleagues. In most cases when scientific claims are proven to be false we do not conclude that their proponents were liars, because we believe they set out to communicate something they believed to be true or justifiably supported by the breadth of evidence at their disposal.
Moreover, one may argue that even if the primary target of scientific publications is the wider scientific community, it cannot be denied that non-scientists also have access to these publications, which serve as a means of public science education. In fact, according to Elizabeth Anderson, one way laypersons can determine whether there is scientific consensus on a certain topic is to assess surveys, reviews or meta-analyses of the peer-reviewed literature. If laypersons have to consult scientific publications to determine scientific consensus on an issue of societal concern, then it becomes important that scientists adhere to some norms of assertion adequate for expert-layperson communication.
Second, we have belief norms. We argue that the authors either presuppose an exaggerated understanding of the norm or have misunderstood it. The authors opt for a quote from the ICMJE’s Recommendations for the Conduct, Reporting, Editing and Publication of Scholarly Work in Medical Journals, which in the original document serves to clarify the question of (co-)authorship. The paragraph in the ICMJE addresses responsibilities that come with (co-)authorship. Dang and Bright, however, suggest that the paragraph’s original meaning contributes to defining authorship within a group of multiple scholars. The authors refer exclusively to the part of the sentence that states that the researchers involved should have confidence in the work done by their colleagues (in the accuracy and integrity of their work). Dang and Bright use this detail to criticize a disproportionate expectation of compliance with the belief norm. What Dang and Bright leave out at this point, however, is that trust in the accuracy and integrity of colleagues’ work is not identical to faith in the resulting claims. There is a subtle but essential difference between these two. The citation chosen by the authors does not preclude researchers involved in a paper from believing the claims. Instead, the point here is to emphasize that researchers should have confidence that their colleagues have worked according to good scientific standards. This is what is emphasized by the emphasis on accuracy and integrity.
Notwithstanding this, abandonment of the belief norm can lead to two scenarios. In the first instance a researcher could fabricate data and therefore know that their claim cannot be valid (implying that they do not believe it). This point links up with our critique of factive norms above, where we argued for a distinction between instances of lying versus genuine ignorance.
Second, a researcher who has worked to the best of her knowledge and is confident that her colleagues have done the same may yet not believe in her claim because she is well aware it might be wrong. If this is the kind of scenario Dang and Bright had in mind with their critique of the belief norm, then their paper would come across as less provocative than its current reading suggests. However, their wholesale criticism of the belief norm, instead of narrowing its scope, leaves out considerations of scientific misconduct. It would have been easy to avoid this pitfall by interpreting the ICMJE recommendations in its entirety.
Finally, we come to the cluster of justification norms. In contrast to the authors, we choose a different order of the clusters. Our order sorts the norms according to their degree of indispensability, starting with the cluster of most minor relevance. The factual norms represent the weakest cluster because both lack of knowledge and truth must not be (especially in the humanities) an exclusion criterion for public avowals (excluding the case of agnotology). Provided that the research was conducted properly, the norms of belief can apply in such cases. Therefore, the cluster of belief norms follows in second place. Even if both clusters of norms cannot be fulfilled, their absence may be compensated by the cluster of justification norms. One can assert X without knowing whether it is accurate or believing it to be true – as long as one can justifiably explain (according to the rules of scientific reasoning) why one is asserting X. Or can’t one? Here a weakness is revealed, which will show that all clusters (but especially the justification norms) should be maintained. The assumption above may be sound under the conditions that all participants in scientific discourse are trying to produce knowledge. However, it ignores the fact that supposedly inter-scientific statements can also be made by representatives engaging in pseudoscientific, non-scientific or fraudulent conduct (who could forget Andrew Wakefield?). A contribution to the scientific discourse that does not take any of the cluster of norms into account loses, as it were, essential demarcation criteria that distinguish the contribution from pseudoscientific claims. If looked at another way: is it not the case that pseudoscientific claims at least play to the norms of belief? This makes it all the more important for scientists to draw a clear line between their contributions and non-scientific or pseudoscientific claims through norms of justification (if not through factive norms).
With respect to Dang and Bright’s specific argument for justification norms, they claim that Bragg’s theory was not justified by the total evidence available. Here they choose to apply arguably the strongest framing of the norm possible (as noted in our general critique above), instead of the version they introduced earlier: that assertions must be discursively justified. It seems Bragg did just that, when he defended his theory against his rivals, by arguing how his account was allegedly better able to explain certain observations than the prevalent alternative. If instead one sets out to apply a stronger framing of justification norms (qua requiring them to be in line with the total evidence available), arguably few, if any, scientific theory and assertions thereof would ever satisfy the norm. The authors seem to be somewhat aware of this as they occasionally restrict the requirement to available evidence to the person asserting at a specific point in time. Not doing so, would mean, that one may adhere to the norm at time t, but could retroactively fail to have adhered to it, once what the norm refers to as “the available evidence” changes over time.
Based on the aforementioned, it is indeed possible to claim that Bragg violated the norms of assertion. This is only possible, however, if one frames them in a way, that few, if any, scientific theories and assertions thereof would be able to abide by. If, on the other hand, one takes much less radical framings of said norms, Bragg did not only adhere to them, his adherence also does much to explain why his input, even if strictly false and unjustified, was valuable.
The norms of assertion in their weaker versions, may arguably make the division of labour in science more efficient, contrary to the claim by the authors. If Bragg would have chosen to abide by no norms of assertion, not even these weaker ones, he may have not been honest about the limitations of his theory and dogmatically stuck by it, thereby risking others to follow suit on a hopeless avenue of inquiry. One does well to remember that division of labour explains why a competition of ideas helps organize inquiry efficiently by helping divide up scientists into a number of potentially fruitful approaches, but coordination does not run contrary to that, if it allows scientists to help others avoid dead-ends. That is where moderate versions of norms of assertion may come into play. At the very least, then, scientific claims should be justified. Dang and Bright seem to be aware of this and therefore propose a contextualist justificatory norm for inter-scientific communication which is distinct from the norm of justification in the analytic epistemology literature. Contrary to Dang and Bright, we hold that a more charitable reading of the norms of assertion (more importantly, the norm of justification) in analytic epistemology leaves room for the kind of justification which Dang and Bright propose for inter-scientific communication.
*All the authors are funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project 254954344/GRK2073, “Integrating Ethics and Epistemology of Scientific Research“.