These notes were written by Brian Scholl. The current version of this document is 1.0 (7/10/99). Several people contributed helpful ideas and interesting discussion to these notes, including: Zenon Pylyshyn.  
 
Please forward suggested additions to Brian.Scholl@yale.edu.

Writing and publishing papers, in most academic disciplines, is perhaps the single most important key to advancing the individual authors' careers, to generating interesting and useful discussion of one's ideas, and to advancing the field in general. It is thus also one of the primary sources of conflict. Authorship on a paper can signal that a person carefully designed a project; that they wrote part or all of the paper; that they had no role in designing or writing but that they nevertheless spent a significant amount of time and effort working on it; that they simply had some particularly good ideas relating to the project; that they belong to the lab in which the project was conducted, but did not participate directly; that they helped out with the data analysis; that they owned the equipment on which the experiments were run; that they simply hold a senior position in the lab in which the project was carried out. In those projects which involve more than a single person -- which is to say, most of them -- there are often disputes, misunderstandings, and lingering feelings of unfairness concerning which contributions do and do not merit authorship.  
 
I've been fortunate myself never to have had any problems along these lines, but I know plenty of other people who have. I wrote up these notes as a result of some recent discussions about these issues. I do not yet run my own lab, but I do often employ and collaborate with undergraduate research assistants, and I am starting to collaborate with more and more people in general; as such I thought these notes might make for some interesting and useful discussion. As noted above, I'd appreciate any extensions to or feedback on these notes.  
 
Below are some thoughts on what types of contributions do and should merit authorship, the corresponding types of authorship-related disagreements that can often arise, and a few thoughts on how to avoid and defuse such disagreements. (Note that these thoughts are written from the perspective of an experimental cognitive psychologist; they may not be apropos for other fields, many of which have much different conventions concerning authorship.)

Before talking about specific types of contributions and disagreements, I thought it might be useful to note the importance of the explicit discussion of authorship issues. The single most important thing that can be done to avoid and defuse disagreements concerning authorship, of course, is just to explicitly discuss these issues early and often. Doing so can sometimes seem awkward, petty, and unnecessary, but I think it's worth it. Conflicts concerning authorship, it seems to me, almost always arise out of fairly mundane misunderstandings that could have been easily avoided with a little pre-emptive discussion. The awkwardness involved in discussing these things is easily matched by the worth of avoiding or attenuating the resentment and feelings of unfairness that could otherwise ensue.  
 
For each well-defined project that one is working on, the lead investigator(s) should just quickly discuss authorship with the people associated with the project. This discussion needn't set anything in stone, but should provide at least a rough foundation for future expectations. (As discussed below, I think that authorship should often remain somewhat fluid, and that it's a mistake to simply legislate at the beginning of a project who will and will not be an eventual author.)  
 
Also, I think it's a good idea for a lab groups as a whole to have at least one explicit general discussion of these issues. In most well-run labs, many projects will quickly become highly collaborative, with many people contributing helpful ideas here and there. It is often impractical to include the entire lab on every publication, so it sometimes becomes useful to have people assigned to various primary projects for purposes of authorship, but to have a standing expectation that the entire lab will often contribute various tweaks and ideas to all projects. Of course, especially good ideas may still merit authorship, but this is an inherently subjective judgment -- which is why this type of discussion is a good idea! (More on all of this below.)  
 
When disputes about authorship do arise, I think it is usually a good idea to be liberal with authorship (that is, to give authorship to those who think they deserve it), and then to just be sure to explicitly discuss things so as to avoid the conflict the next time around.  
 
By the way, I think that the real solution to almost all issues of authorship ambiguity would be for the major journals to adopt a mandatory convention wherein each author's actual contributions are succinctly and explicitly stated in the publication, perhaps in the acknowledgments section. I heartily support this sort of thing, for many reasons, though I don't expect such a sea change to ever really take place. I've seen Steven Harnad voicing support for this proposal, and it's also suggested by the author of the following paper (which is well worth reading in general):  

Thompson, K. S. (1994). Scientific publishing: An embarrassment of riches. American Scientist, 82, 508 - 511.

Authorship on papers can often be fairly judged, it seems to me, in terms of the 'replaceability principle'. As a rough heuristic: anyone without whom the project could not have unfolded as it did should be an author; easily replaceable people, on the other hand, may not deserve authorship. In a typical case, this might mean, for instance, that the senior researcher and thesis student will be authors (since they collaboratively designed the project in the first place, and without them the project would never have seen the light of day); that another colleague might be an author (by virtue of having suggested an especially important and novel idea which radically altered the project); that the research assistants who programmed the experiment and ran the subjects would not be authors (because if they hadn't done that work, somebody else would have, without significantly altering the project); that a fellow lab member who suggested a minor improvement might not deserve authorship (because such minor improvements are part and parcel of any lab meeting, with any lab members); that the senior director of the research center where the research was conducted would not be an author (because his being in that position would probably not have at all affected the project, which could have been conducted in the same way elsewhere); that a colleague who consulted on the statistical analysis would not be an author (because somebody else could have provided the same consultation). Note that this is properly seen as a heuristic, to which there may be exceptions (regarding thesis students in particular; see below).

Here are some thoughts, scattered among discussions of the various roles one can play in a research project, on what sorts of contributions should and should not merit authorship (in my opinion, of course).  

• A solitary researcher 
  In increasingly rare cases, there may be an individual (at any level) who, by herself, conceives, designs, and implements a project, runs subjects, and writes a paper completely independently, or with only very minor input from others. There would be little opportunity for authorship disputes here (though see 'Senior researcher/Lab director' below). The prevalence of this type of research of course varies tremendously by field. In some fields, e.g. philosophy, solitary work is the norm. In many other sciences -- e.g. planetary geology -- it's not unheard of to have hundreds of authors on a single paper, and papers with 10 or so authors are extremely common. Cognitive psychology is somewhere in the middle of this spectrum, with most research being conducted in pairs (often mentors and advisees) or in small lab groups of 3 to 5 people. (Cognitive neuroscientists increasingly seem to work and write in larger groups.)  
   
  
• Senior researcher/Lab director 
  Some senior researchers and lab directors in cognitive psychology are still very actively involved in programming experiments, running subjects, analyzing data, and the other day-to-day business of experimental psychology. Most, however, are not. They will be intimately involved in the selection of research directions and in the design of experiments, but the actual implementation is often left to students and other more junior researchers. Individual differences here are often central, of course, to the ill-defined project of choosing of a research advisor in the first place. Some advisors are happy to view the mentor/student relationship as one of didactically-oriented collaboration; others use their students as little more than cheap labor. Some advisors are happy to work on whatever areas or projects interest their students; others may explicitly require, e.g., that post-docs spend 56.25% of their time working on their advisor's projects. Conventions regarding authorship for senior researchers varies widely, and can be especially sensitive given that it is they who often decide how and when to write papers, and who will participate. Many will claim authorship for themselves using the same criteria as they do for anybody, gauging whether they have had a significant impact on the project. In most cases, of course, authorship is definitely deserved, since it is often they who conceived and designed the project initially. Other times, however, senior researchers will claim authorship on papers simply by virtue of the work having been done in their labs; this seems unfair, because of the replaceability principle: if the senior researcher is really just an advisor on a project which the student (or other junior researcher) devised and implemented independently, then they are performing an eminently replaceable function. (Some people would argue that authorship can be deserved in such situations for reasons of financial support, and the donation of lab space, equipment, etc. I suppose that this type of reason is arguable, though I personally find it distasteful, and wouldn't choose to work in such an environment.)  
   
  
• Thesis students (both graduate and undergraduate) 
  Many research projects are conducted as a collaboration between a senior researcher and a student working on the project as a PhD thesis, a Masters thesis, or an undergraduate honors thesis. In most such cases, there is no opportunity for authorship disputes because both will have contributed major parts of the project's design. In some cases, however, there will arise situations wherein a thesis student essentially does all the work for a project which was largely or even completely designed by their advisor. This situation is ripe for misunderstanding. The replaceability principle might here suggest that the student should not be an author on the resulting paper, but there is often a strict (though unwritten) convention that thesis students will be authors on the publication of their thesis work. A very few mythical advisors are alleged to hold the view that students needn't ever be authors on their work, since they are being compensated instead by their training experience and their degree, etc.; this (possibly nonexistent) situation seems reprehensible, and I would avoid such advisors like the plague.  
   
  Many of the disputes between advisors and thesis students that I know about have concerned the resulting order of authorship: should the thesis student automatically be the first author on the resulting paper(s)? Although there are often explicit expectations along these lines, this seems to be a subtle question, the answer to which will depend upon the specifics of the situation. In a situation wherein the student played the major role in conceiving and designing the experiment, they should probably be the first author. In a situation wherein the student is simply carrying out a project conceived and designed by the advisor, however, it seems fair to let the senior researcher be the first author. In the many cases in the middle, it might be fair to judge the order of authorship in terms of relative degrees of 'replaceability': if the advisor and student contributed roughly equally to the conception and design of the project, but the student actually carried out the experiments, then the student might deserve to be the first author, having made a larger overall contribution to the project.  
   
  The important thing in this situation, as always, is to explicitly discuss expectations concerning authorship. A situation wherein an advisor designed (most of) the project, but the thesis student did all the work, for instance, seems especially ripe for problems arising from conflicting expectations.  
   
  
• Research assistants 
  Many labs often employ research assistants to help out with (or even take charge of) scheduling and running subjects, constructing stimuli, and other similar tasks. In the majority of such cases, these RAs should probably not be authors on the resulting papers, simply because of the replaceability principle: if they hadn't done these tasks, somebody else would have done the same work, in largely the same way. RAs, in other words, often do not contribute anything irreplaceable to a project. (In addition, of course, RAs are often employed literally, and are compensated via work-study credit or even a salary.) However, even this can be a gray area. Many future grad students get their start as RAs who regularly have particularly good ideas, and in our lab, for instance, RAs have sometimes been authors on projects to which they contributed an inordinate number of eventually-crucial, irreplaceable ideas. In some extreme cases, I suppose that an RA might merit authorship simply by dint of having put an extraordinary amount of work into a project, but then again, there are many labs wherein RAs put in significant amounts of time and effort for nothing more than 'research experience'. More often, RAs will start out by helping out on projects for which they will not be authors, but will gradually assume more responsibility for the actual design of the experiments, and may then take on a project on which they will play a primary role and on which they will be an author; often this project will comprise a thesis or other type of formal independent research. Similar comments here apply to situations of statistical consultation, etc. I know of students who have been made authors on papers simply for doing the straightforward statistics for a probability-impaired senior researcher. This seems inappropriate to me, and cries out for the type of solution suggested by the Thompson papers cited and discussed above.  
   
  
• Other students, colleagues, lab members, and random passers-by who offer ideas 
  Another common situation that results in authorship-related conflict concerns anybody who offers a clever idea that promises to improve a project. Often, this will arise from lab meetings, where a lab member who is not directly involved in the project offers a helpful suggestion. This can also happen as a result of a colleague commenting on a manuscript, or perhaps making a suggestion at a scientific meeting. Most such suggestions, in my opinion, do not merit authorship, simply for practical reasons. As a matter of course, any lab meeting or scientific talk will result in many such ideas, and it is not feasible to grant authorship to so many people. In the context of lab meetings, especially, such suggestions are often 'repayed', as it were, in similar suggestions when the next project is discussed. There can then be a standing rule of sorts that everybody is expected to contribute ideas to all projects during lab meetings, but that such input does not typically confer authorship, since it evens out in the long run. (This scheme works best when each lab member has at least one project on which they're a primary researcher, and on which they will be an author by default.)  
   
  Again, however, there are often exceptions to this rule, and authorship guidelines in a lab should be fluid enough to accommodate them. I know some lab leaders who try to simply legislate the authors at the beginning of a project, and then stick to that list with no exceptions, to avoid any conflicts or ambiguity. Personally, I don't think this is the right thing to do. I think that some 'random' suggestions -- the really good ones! -- are deserving of authorship. But the decision of what counts as a really good or essential or novel idea is obviously subjective, which is why all of this requires some explicit discussion. Often a colleague will feel slighted because they were not offered authorship for their extraordinary groundbreaking suggestion, whereas the current authors view this particular suggestion as obvious, mediocre, or as one which they have heard independently from several sources. A bit of explicit discussion can attenuate such misunderstandings.  
  

For other useful discussions of the sociology and ethics of academic authorship, see the following articles:  

[To be added when I return home to my files. I know I've clipped various articles along these lines from the APA Monitor, APS Observer, etc. Other pointers to relevant discussions would be appreciated.]

In sum, it can often be useful to discuss these types of issues in collaborative projects and lab meetings, in a pre-emptive attempt to synchronize everyone's expectations, and thus to help attenuate or avoid future authorship-related conflict.  
 
Of course, no amount of discussion can avoid all such disagreements. Some people will always be jockeying for position, trying to amass as many publications as possible -- 'he who dies with the most toys wins' -- and some authorship disagreements will simply boil down in the end to irreconcilable differences as to the worth of a suggestion or improvement, etc. It seems best in such situations to be liberal with authorship, and then to choose one's collaborators more carefully next time. The best collaborations are always between people who are already established and confident themselves, such that authorship issues fade into unimportance, and you can really focus on the ideas and the research.  
 
However, the sociology of academia (and the satisfaction of mentoring!) necessitates that many collaborations will be between people of radically differing degrees of experience with (and knowledge about) the research process, and people can often advance surprisingly far in their careers while still having skewed expectations concerning authorship. It thus seems worthwhile to discuss such things explicitly once in a while.