There are many differences between the scholarly cultures of the humanities and the natural sciences. One seemingly superficial but striking difference is in the number of authors per paper. A significant number of papers in the humanities are single-authored whereas in the natural sciences multi-authored papers are the norm; for some articles, the author list is longer than the article itself. For example, this paper on an experiment performed at the CERN Large Hadron Collider sports an army 2,926 authors, enough to fill two concert halls.
In contrast to the natural sciences, publications in the humanities are typically single-authored; examples are essays in philosophy, linguistics, and literary criticism. Such scholarly endeavors are by nature individualistic. The author’s style of writing and argumentation play an important role. References to these essays are therefore often accompanied by quotes rather than by a dry summary of findings. It is apparently not only important what the author said but also how he or she said it.
In the humanities, the author can be held responsible for the entire content of a paper; all remaining errors are my own is a common expression in the acknowledgements of such scholarly contributions. In the natural sciences, complementary types of expertise are essential to carry out a project. (I haven’t tried it yet but I’m pretty sure you cannot single-handedly conduct an experiment in a particle accelerator.) As a result, among the thousands of authors there probably isn't a single one who oversees the entire paper.
So what is the lay of the land in the social sciences? In psychology—the field I am focusing on—multi-authored papers have become the norm. Often co-authorships are student-mentor partnerships but especially with the advent of neuroimaging techniques, the complementary-expertise model of the natural sciences has become common.
Multi-authored papers raise all kinds of issues regarding credit. How much credit should go to the first author relative to the other authors and what is the status of the last author? Various journals, including Psychological Science as of this year, are now requiring authors to specify their respective contributions. Who designed the experiment? Who analyzed the data? Who wrote the paper? Who went along for the ride? And so on. This is a good idea. Moreover, it is not only a good idea for assigning credit but also for assigning responsibility.
To what extent should a co-author be held accountable for the entirety of a scientific article? A—what I would call—shared-responsibility view holds that by signing on as co-author, a researcher is responsible for the entire paper. The rationale for this assumption is that if you want the credit then you should also accept the responsibility. And if there is blame to throw around, it should fall on everyone. If you burn your behind, you’re going to have to sit on the blisters as the Dutch expression goes.
Another view assumes divided responsibility. Authors are only responsible for the part that is covered by their domain of expertise. The rationale for this view is that you cannot hold people responsible for things they have no control over. This would seem obvious in the physics example I just gave. Forgive my profound lack of knowledge on the topic, but I would assume that the guy who cranks up the particle accelerator and the guy who touches up the images in Photoshop have no overlapping expertise, so it would seem unfair to rake the former over the coals if the images are artistically subpar.
Which view should we adopt in psychology, shared or divided responsibility? The case of Barbara Fredrickson, which I described in my previous post, provides a poignant illustration of the issue. Fredrickson had co-authored a paper with Marcial Losada in which they presented, among other things, a mathematical model of emotional dynamics based on fluid dynamics. A recent paper convincingly and eloquently showed this model to be a mathematical shambles.
In a response to this critique Fredrickson radically disowned the model. She argued that the modeling was entirely Losado’s work and that it, all things considered, was not even relevant to the rest of the research, so that it could be safely expunged from the record.
Many people find Fredrickson’s response inadequate (see for example the comments on this Neuroskeptic post) and it is easy to see why. On multiple occasions Fredrickson has embraced the model and touted its virtues, for example in a popular book and in this talk (starting at 12:35). Her own website until very recently displayed the butterfly image produced by the mathematical model. The image is gone now, but at the bottom of this post is a screen shot.
By washing her hands of the model now, Fredrickson has shifted from a shared model of credit to a divided model of responsibility. All gain, no pain in other words.
There may not be a good solution to the problem of assigning credit and responsibility but all gain no pain doesn’t seem the right model. It might be a start to require authors to indicate not only which components of the work they want to receive credit for but also which ones they want to be held responsible for. And wouldn’t we want to have a perfect match between credit and responsibility?