How does the premature death of eminent researchers in the life sciences influence the renewal and progress of their research fields? While great scientific authorities are alive, any questioning of their leadership is complicated, but when they depart prematurely, their specialist field experiences an evolution in new directions that broadens the frontiers of knowledge. The number of articles published by the collaborators of a scientific figure falls when that individual dies, but the number of articles published by non-collaborators undergoes a considerable increase. The publications that have the greatest impact are those produced by people positioned outside of the scientist’s specific speciality, that tackle essential questions in that field, although from another perspective. 

It has long been accepted that the process of knowledge accumulation – i.e., how new research builds on previous research work – is of central importance for scientific progress and economic growth, but how this process works is not well understood. On what do researchers base themselves when choosing between different ways of studying a specific problem? Is the evolution of science based on autonomous laws, or is its direction influenced by individuals, incentives and institutions? This question regarding the evolution of science has undergone a long theoretical debate.

If we are to believe the famous quote attributed to Max Planck, science advances, indisputably, one funeral at a time: “A new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it". This quote from Planck expresses the idea that a scientist’s attitudes and beliefs can largely determine scientific advances or, at least, delay them. These attitudes and beliefs will be stronger in the case of consolidated scientists. 

However, barely any empirical data exists to back the hypothesis that veteran scientists take longer to accept novel theories than their younger colleagues (Hull et al., 1978; Gorham, 1991; Levin et al., 1995). For this reason, we believe that this article can lay down groundwork to help gain a better idea of how knowledge is transmitted across different generations.

This research study aims to analyse the effect of the premature death of prominent scientists on the dynamics of the knowledge areas to which they contributed during their lifetimes. For this, we started off with a list of 12,935 elite scientists in fields of study related with the life sciences in the USA. The respective fields were delimited with tools constructed for this purpose. These scientists represent a sample of nearly 5% of scientists in their specialities. They are defined as elite professionals based on criteria such as the amount of research funds received, citations of their work in publications, number of patents, their belonging to organisations of prestige and prizes and awards received. A key aspect of our study involved identifying how many of them died prematurely, for which we used information from the United States press and academic obituaries. 

Next, the CVs were analysed of 452 elite scientists (54 women, 398 men) who died between 1975 and 2003 before retiring or accessing management posts. In other words, they died when they were still fully competent and active in their profession. 

Apart from the fact that it is much easier to identify the cause of death for elite scientists, as opposed to the average scientist in the profession, the study also focuses on the scientific elite because the big stars leave in their wake a corpus of work and a group of followers interested in preserving their legacy, which enables their professional itineraries to be accurately tracked. The primary (and most original) point about this focus is that the time of death of elite scientists is used to calculate to what point, following their death, the production of knowledge changes in their fields of specialisation. 

In short, sufficient data exist to analyse the path that has taken them from their origins to situations of generalised recognition. Reconstructing the academic careers of all these scientists enables us to find out their publication history, which means we can then define who is a “research superstar”. There are data on employment histories, degrees and qualifications earned, date of completion of studies, researcher gender and the departments to which they belonged, as well as a complete list of articles published, and the patents and research funds obtained annually by each individual.

Moreover, to work out how a specific field evolves, we have a list of all the scientists who at some stage have published in the field of research of the scientist who died prematurely, independently of whether they collaborated with the deceased individuals or not. We define a collaborating scientist as somebody who appears as co-author of a publication together with the deceased elite scientist. Similarly, non-collaborators are those members of the scientific community that never wrote an article with the deceased scientist; within this category are individuals working in fields that are both scientifically close and distant from the elite researcher. 

To delimit the fields of research, it was observed that the 452 deceased scientists had published 3,074 articles in the five years prior to their death. For these articles, similarity based on the overlapping of key words helped us to identify the articles that are closest in thematic terms. Each of these 3,074 articles, together with the most similar articles in the scientific bibliography, are situated in a field of research. For example, geneticist Ira Herskowitz died in May 2003. Our algorithm identifies 72 scientific articles related with the article “The transcriptional program of sporulation in budding yeast”, published by the journal Sciencein the year 1998 and with Herskowitz as the main author. Our definition of the field of research encompasses this article and the 72 that our algorithm additionally identifies as related articles. Examples of other fields of research are the brain mechanisms that affect colour perception (Russell L.De Valois) or the pathology of smoking and emphysema (Aaron Janoff).

The analysis shows that, at least in the life sciences, the premature death of elite scientists influences the global dynamics of research: after these deaths, scientists who did not collaborate with their deceased colleagues become more visible in two dimensions. Firstly, because they present new publications with greater frequency in the field in which the deceased figure worked. Secondly, because this increase in publications will also receive more citations in the future, showing that they are contributions and ideas with a high impact. We have observed that these “emerging stars” are usually scientists who previously did not work in the same field, since their research articles were not among those most similar to the articles written by the deceased elite scientist. This leads us to think that nor did they work on very similar projects or questions.

The results are consistent with the fact that the scientists outside of a certain field seem to be reluctant to question the work of a great figure who is considered leader in that specialist area, only entering the field once it was left orphaned by the “superstar”. 

Furthermore, as also seen reflected in graph 1, the collaborators of the star scientist publish less following his or her death – around 40% less after 5 years – as we already saw in Azoulay et al. (2010). The new result here is that the number of publications from non-collaborators increased by an average of 8% after 5 years. Given that the number of non-collaborators is much greater than that of collaborators, the activity of these non-collaborators ends up fully compensating for the lower productivity of the collaborators. There are no signs that this situation is merely transitory, as if anything, the effect increases over the course of the years. We observe a similar behaviour if, instead of scientific publications, we focus on the destination of scientific funding; following the death of an eminent researcher, scientists from outside the field in which the deceased scientist worked not only publish with greater impetus, they also receive more funding.

In summary: while the co-authors of a major figure suffer their loss, it cannot be said that major scientists from another lab or a competing research group take on their leadership role without delay. Instead, the impetus seems to come largely from people outside of that immediate environment who are not necessarily established there, and who appear to study the essential problems of a certain field by leveraging new ideas that have emerged in other areas. 

In addition to the changes in scientific production, examined in the previous section, there is another question of great interest: Do the increasing contributions of non-collaborators produce any real change in the intellectual direction of the articles published? To explore this issue, two types of analysis are used.

Firstly, within a certain field of research, we distinguish between contributions that intellectually are more and less close to the research questions in that field. We do this by basing ourselves on the semantic distance between the key words of each pair of articles. The data show that articles with similar themes have approximately double the impact (measured through the citations they receive in works published subsequently) than those on more distant themes. For this reason, we conclude that, despite the death of the superstar, when we examine the new scientific contributions within a field of research, they continue analysing the same scientific problems, but from another viewpoint, which in general is fresher and more novel. 

Secondly, the new articles that appear following the death of the eminent scientist have been classified into two groups. The first group contains publications that cite at least one work from the research specific field. The second contains publications that only cite works outside of that subfield. Following the disappearance of a star researcher, only the number of articles in the second group increases. 

Moreover, it is observed that nearly all the development of the subfield affected by a premature death (measured by the publications following that death) can be attributed to articles by non-collaborators, who do not cite the deceased elite scientist. Furthermore, when compared with fields that have not experienced the death of an elite scientist, these new contributions come from more recent scientific works.  

Among the deceased scientists there were few editors of academic publications or members of committees that award research grants; therefore, we can rule out the possibility of them using their influence to determine who could or could not publish their work or receive grants within their field. However, if the barriers that limit entry do not come from direct control by the great stars, what is it that disincentivises access for “outsiders”,i.e., people who up to that point had not participated in the field.

The shadow of Goliath

The possibility exists that researchers from outside the field are discouraged, simply, by the prospect of questioning an eminent figure in the discipline. It could be said that the mere presence of a famous academic is sufficient to put off people who, from outside, aim to enter a specific research field. In these environments, a large part of new entries takes place when an especially relevant figure disappears.

All in all, the outstanding capacity of that star scientist might not be the only factor that influences the entry (or the non-entry, in fact) of new researchers. For this reason, this research study also analyses how the specificities of each research field with regard to intellectual and social coherence, modulate the relations between new and old researchers, and what role is played by collaborators of the deceased star scientist. 

The intellectual coherence of the field of research

Even following the disappearance of a leader in a research field, it may be that there are no incentives to access this field if, from outside, it is perceived as an intellectually coherent environment. We understand that an environment is considered intellectually coherent when the researchers that participate in it share the set of considerations, approaches and methodologies that drive that field. In other words, if there are no great controversies or cracks that allow the entrance of new theories subsequent to the death of the star scientist. To probe this concept more deeply, two measures of intellectual coherence are proposed.

The first index of intellectual coherence reflects to what extent the articles from a single field use similar key words to each other. The more the key words overlap each other, the more coherent the field is. The second index uses the list of references quoted in each article from the field before the death of its leader. The higher the proportion of studies from the field itself cited in its own publications, the more coherent that field is and, therefore, the greater the disincentive felt by external researchers regarding entering that field.

The social coherence of the field

It could be considered that a field is socially coherent when the researchers who work in its midst frequently collaborate with each other, and also when they mutually review each other’s texts. Using a more colloquial term, we could talk about the field being endogamous. With the aim of investigating these aspects, two hypotheses of social coherence were developed.

The first hypothesis is that those fields in which star researchers have had many students will be less welcoming to outside scientists than those whose leaders did not train many graduates or post-doctoral researchers. The second hypothesis is that the fields that present the greatest social coherence are those in which it is easier to find co-authors.

What intellectual and social coherence do the data show?

When we observe a high intellectual or social coherence in a field of research, the closest collaborators of the deceased researcher could be continuing to regulate access to that field. It may be that the presence of major figures was a source of dynamic activity while they were alive, but the renewal of leaders originating from other areas is only observed in those areas more prepared to support and accept fresh ideas from outside.

The data show lower entry of outside researchers in intellectually coherent fields that have revolved around a small set of techniques or ideas. Similarly, the influence of new ideas and approaches is attenuated in those fields in which the collaboration networks are especially close-knit. 

Therefore, we can conclude that the essential collaborators of the deceased researcher can still regulate entry into the field through intellectual and social barriers, and those related with resources provision.

This analysis has highlighted that, once an elite scientist disappears prematurely, a proliferation of articles appears in that scientist’s field written by researchers who never collaborated with him or her. Although the data are not shown here, the funding that these “new actors” receive for research grows in parallel with the publications.

This proliferation is not due to a reorganisation of the leadership within the field, but to the entry of new scientists from outside of it. Specifically, the data indicate that this increase in contributions from scientists previously outside of the field is concentrated largely on its essential questions, but includes a higher proportion of ideas originating from other environments to which the dead scientist had never contributed. This intellectual hybridisation turns out to be quite fruitful, as the new articles offer relevant contributions, at least judging by their long-term impact in terms of the citations that they receive.

Data were also analysed on the mechanisms that regulate entry into the field. In general, the entry of new scientists after the unexpected death of an eminent researcher has less impact in places where a compact network of collaborators is capable of maintaining the entry barriers. In other words, the star scientist’s legacy and influence are maintained through their collaborators when the field shows good cohesion, whether through intellectual barriers or social ones.

It could be thought that the loss of an elite scientist who had been of capital importance for a certain field of research would be interpreted by the rest of the scientific community as an opportunity for renewal. In other words: once the research field has been “orphaned”, it seems reasonable to think that any reticence towards considering and incorporating avant-garde ideas will have declined. However, this only occurs when the research field presents an environment willing to accept and back  new ideas.

In summary, the results of the study coincide with what is expressed in the idea attributed to Planck that gives this text its title: science advances funeral by funeral. However, the consequences in terms of social welfare are ambivalent.

Although it can be considered that star scientists restrict the entry of new ideas and new academics in an area, custodian efforts in that field may turn out to be beneficial when still in the early days; they could also permit an accumulation of progress based on common budgets and methodologies and, in itself, the capacity to control the intellectual evolution of a specific scientific area could be a spur to incentivize risk-taking in research.

All in all, the fact that the tutelage of a major figure can freeze the patterns of participation in a specific scientific field increases the attractiveness of measures that encourage the access of researchers who are new or have fewer academic connections and contacts. Those measures include the limits to the funding that a single lab can receive; the awarding of additional points to researchers who apply for funding for the first time; reviewing procedures based on double-blind methods, and the approval of emeritus awards that lead veteran scientists to reduce their laboratory activities.

This article has been adapted from the study:

AZOULAY, PIERRE, CHRISTIAN FONS-ROSEN and JOSHUA S. GRAFF ZIVIN(2015), "Does Science Advance One Funeral at a Time?", NBER Working Paper 21788 
 

More bibliographic references:

AZOULAY, P., J.G. ZIVINY and J. WANG (2010): “Superstar extinction”, Quarterly Journal of Economics, 125(2).

GORHAM, G. (1991): “Planck’s principle and Jeans’s conversion”, Studies in History & Philosophy of Science, 22(3).

HULL , D.L., P.D. TESSNER and A. M. DIAMOND (1978): “Planck’s principle”, Science ,202(4369).

LEVIN, S.G., P.E. STEPHAN y M.B. WALKER (1995): “Planck’s principle revisited: a note”, Social Studies of Science, 25(2).