Check against delivery
Dear Minister HEITOR, dear Manuel,
Dear Rector Joao SÀÁGUA ,
Ladies and Gentlemen,
I thank Rector SÀÁGUA and Vice-Rector Elvira FORTUNATO for their kind invitation to participate in the NOVA Science Day. It is always a pleasure for me to be in Portugal and enjoy its remarkable hospitality.
For my lecture this morning I chose the topic “interdisciplinarity” not only because of its importance and relevance today but also because of the challenges it represents for policy-makers, funding agencies and for universities and research organisations. If I had to give a title to this speech, it would be “Supporting Interdisciplinarity, a Challenging Obligation”.
First, it is very important to be conscious of the considerable variety of situations that involve interdisciplinarity: from the obvious case of complex problems to bring together truly diverse competencies (of which climate change provides an excellent example) to emerging new disciplines and from evaluation issues to training issues, etc.
A lot of discussions have taken place around possible differences between interdisciplinarity, pluridisciplinarity or even transdisciplinarity. I will not be discussing this here, and for the sake of simplicity only refer to interdisciplinarity.
Dealing with complex issues
There are numerous examples of scientific issues that require expertise from different disciplines. In recent years, one of the most evident examples has been provided by the follow up of climate change by the International Panel on Climate Change (IPCC) set up by the United Nations – a unique initiative. The IPCC faced and still faces many challenges:
- to collect a considerable amount of data with a stable methodology;
- to develop models and scenarios validated for the past on the data assembled;
- to produce reports subject to a very thorough cross-evaluation by validated experts from different disciplines and different countries.
To make sense of the information gathered requires an extraordinary combination of knowledge. If there are indeed models for the general atmospheric circulation, it is very demanding to articulate them with the description of local situations.
As a result, it is quite hard to make an extensive list of fields that need to be mastered in this endeavour: Geography, Physics, Chemistry, Statistics, Mathematics, Plant Biology, Sociology and in particular Urban Sociology, Agriculture, Political Science. All these domains are needed in order to come up with scenarios for future decades for which uncertainties can be properly estimated. Indeed, uncertainties are to be expected. They just have to be bounded carefully in order to deliver a clear message about the most relevant measures to take and the impact on many sectors of society and on the lives of ordinary citizens.
This represents a very comprehensive effort, mobilising thousands of scientists with many backgrounds. We all know that the major challenge is probably not even the scientific effort that needs to be done; but it is getting politicians to make good use of the collected evidence and to implement the changes that need to be understood and accepted by everyone.
Accompanying the emergence of new areas
A quite different setting where interdisciplinarity is relevant concerns the emergence of new domains that, later, continue to exist as independent entities with their own organisations and structures.
I would like to point to two examples with the aim of showing that the timeframes for such developments can be quite different: the birth of Molecular Biology, which made the Human Genome Project possible and led to Bioinformatics and the development of Cognitive Sciences and Behavioural Economics.
Molecular Biology results from an attempt in the late 1930s to explain life using the basic laws of Physics and Chemistry. The combination of these two disciplines and the belief that they allow one to consider mechanisms involving very complex, but specific, structures led to the creation of a new field. The key actors of what became a revolution are some specific macromolecules: nucleic acids, and in particular the most famous one, DNA, and proteins, through which living organisms act. It took quite a long time, some 50 to 60 years, to develop this field successfully as it required characterizing the structure of these complicated molecules, their function and their relationships.
As is well known, it was the unravelling of the very peculiar double helix structure of DNA that played a key role in understanding how a genome functions. This led to nothing less than the deciphering of the human genome. Such a project looked completely unreachable for some time because of its complexity and its high costs, reaching hundreds of millions of euros. Thanks to extraordinary progress in technology and computing power such an endeavour, which originally required the mobilisation of many teams around the world for months to share in a planned way their know-how and resources, has now become a routine exercise. Individuals can now be provided with their genome in minutes and for a few hundred euros. All this lies at the heart of the development of Bioinformatics and many activities in ‘omics’, starting with genomics, proteomics, etc. Again, let me stress here that such a perspective, which is now the basis of hope for the advent of personalised medicine, was considered for a long while just a dream.
All this could only happen thanks to extensive exchanges leading to the development of a common language and shared methodologies by scientists who, initially, had quite different approaches. This was of course a lengthy process that required open minds, patience and perseverance. Along the way there have been failures due to overly impatient steps taken on the basis of naive views. Nothing of this sort could have happened without some people thinking ‘out of the box’ and dreaming the impossible to reach the boundaries of the possible. Support for risk-taking is key for such things to happen.
As mentioned before, dealing with the huge amounts of data that all “omics” sciences require was only possible not only thanks to new computing power provided by the massively expanded capacity of computers but also thanks to the development of efficient algorithms to deal with these data in a manageable and efficient way. This actually meant a quite radical transformation of Biology and turning part of it into a “Big Science” when its tradition favoured small teams working on quite different and narrow projects.
The development of another new discipline is also worth looking into: Cognitive Science. The basis of this is the study of the processes by which the mind builds cognition from information that various senses gather so that several functions are articulated properly, from perception to memory, from memory to action, from emotion to language, etc. The fields dealing with such knowledge are very diverse: Physiology, Neuroscience, Psychology, Linguistics or Mechanics. Getting actors in these fields to interact and propose complementary representations could not be achieved without developing new concepts and suggesting how computational models can integrate data collected, hence the role also played in these disciplines by Computer Science. New investigative methodologies are constantly being added. A good example of this is proton magnetic resonance spectroscopy coming from sophisticated physics: it now allows one to follow the electro-chemical activity of the brain with a high level of detail and in real time. The process of analysing these signals is very involved, so that referring to it as an “imaging” process is hugely oversimplified.
It must be noted that the level of information that has become accessible through these approaches is beyond anything that was expected some 30 years ago. Indeed, it allows one to relate very directly any physical activity to what is happening in the brain. To achieve such a high level of integration of competences is very challenging as it requires people with totally different backgrounds to coordinate efforts and to accept that new methodologies invade their usual practice. This does not go without resistance. The development of a new, more comprehensive approach by these actors remains key for future progress. This is what is happening with a Cognitive Science approach to Behavioural Economics, one of the new frontiers in the area.
The reality of interdisciplinary research seen from the ERC
Let me come to something quite different, namely the share of Interdisciplinary Research in the present activity. With to date almost 10,000 ambitious research projects funded, the European Research Council (ERC) provides a good observatory to get an idea of the share and impact of interdisciplinary projects.
Let us start with a picture of the nature of projects funded by the ERC. For the purpose of project evaluations, presently 25 scientific panels cover all domains of knowledge. Outside a few cases, such as Mathematics and maybe Computer Science, very few of these panels are strictly disciplinary. This means that already a number of projects evaluated by one single panel and eventually funded by the ERC are multidisciplinary. Still, it is interesting to look at a map of grants that are “cross-panel” as they represent most of the time more radical levels of interdisciplinarity. Here is a diagram giving the intensity of cross-panel grants. It shows, for example, that with the present structure of ERC panels a number of Principal Investigators in the Life Sciences (LS) need to be evaluated by experts from several different panels.
Another piece of interesting information comes from the statistics of the ex-post evaluation of ERC projects. What is this? In an effort to get a global view of the impact of ERC-funded projects (mainly scientific but also economic), the ERC Scientific Council has for the past four years annually conducted, and will continue to conduct, an evaluation of around 200 randomly chosen projects two years after their completion. The possible grades independent evaluators can give are A for breakthroughs, B for major scientific advances, C for incremental scientific advances and D for projects with little achievements. The differences in the level of performances of interdisciplinary projects has been striking, as shown in the diagram below.
Another decision showing the attention given by the ERC Scientific Council to interdisciplinarity is the relaunch of Synergy Grants calls –grants that allow proposals be submitted not by 1 Principal Investigator but by 2, 3 or 4 scientists together with the objective of tackling a truly ambitious scientific challenge. It was primarily motivated by the will to create a space where interdisciplinary projects could be naturally submitted and better evaluated. This had to do with the possibility of having several researchers come together with a common ambition in a synergetic style. In the end, 77% of the successful applications were considered interdisciplinary. More precisely, out of the 26 Synergy projects funded in 2018, 3 used keywords typically coming from 2 usual panels; 3 used keywords coming from 3 panels, 13 from 4 and 1 from 5. Great care has been taken to ensure that the evaluation respects this diversity by introducing a 3rd step evaluation and forming the panels for the Synergy calls in charge of making decisions only after the list of applications is known. In this way, it is possible to match to an optimal degree the expertise needed to do the evaluation.
Challenges in terms of evaluation
This leads me naturally to stress that Interdisciplinary Research activities face different types of challenges that need to be properly analysed. Let us begin with what concerns evaluation at the level of projects. I already mentioned it when I briefly spoke of ERC Synergy Grants.
This is again an issue that the ERC considered very seriously as we noticed a persistent slight discrepancy in success rates between cross-panel applications and the others. This has been happening in the first step of the evaluation of ERC applications, where only opinions of generalists are gathered. In the second step, where the opinions of remote referees with a very specific competence of the project are solicited on top of the ones assembled for the first step, this discrepancy was no longer visible.
We at the ERC of course looked into the causes of such a difference: the fact that the opinion of cross-panel members is transmitted only in writing did not give to it the same weight (we even tried to measure the loss of weight and it was quite significant) as the one it would get through a physical presence. We have explored the possibility of having panel members making a visit to the other panel with mixed success. We are trying to create better conditions to make it a more systematic tool. It even has a name: “The traveling evaluator”!
Another issue, that is most likely even more important, is the considerable differences in methodology used in different disciplines and recognised as acceptable and pertinent by experts in a given field.
Yet another dimension that is of course important when one looks at truly ground breaking new results is how to ensure the high level of originality requested when one is not completely sure of what the playing field is. Even more to the point when we are talking about originality in an interdisciplinary project is the fact that sometimes the true breakthrough is not in the contributions in separate fields but in crossing knowledge and methodologies from different fields in a radically new way. This often generates a mixed opinion from experts who tend to limit themselves to their usual horizons without accepting the risk of betting on the possibility of a truly new cross-fertilisation between fields.
This stresses the fact that the more averse to risk taking an evaluation panel is, the more likely it is to not give a high assessment for highly interdisciplinary projects.
As you can see, the challenges in terms of evaluation are considerable and require scientists with an open mind and some practice of interdisciplinarity. This brings me to what I consider a must in terms of both organisation of Research and Education. In the end we know very well that what makes the difference are people… and the way they have been trained.
Challenges in terms of training
As we are all aware, the present generation of researchers has been trained mostly in maximising the competence of individuals in a well-defined discipline. Of course such an approach has its value, in particular because it can lead rather quickly to a recognised high level of competence… but most likely it will be a rather narrow one. This brings us to the worry of hyperspecialisation!
We know that, in the last fifty years, a number of new disciplines have emerged and finally gained their “lettres de noblesse”. I mentioned Molecular Biology and Bioinformatics. Another obvious case has been Computer Sciences and more recently, in a process that is still on-going, Cognitive Science and Data Sciences. The recognition of new disciplines tended to be too slow and sometimes viewed only as a branching process. In a number of cases this resulted in endangering the original discipline that was suddenly stripped from resources to create space for the new branch. This is what happened in France for example where for a long while Computer Science departments were created by taking away positions from Mathematics departments. It took a long effort to get all these positions back. It showed the wrong approach that entails identifying the emergence of a new field as a branching process.
The key question ahead of us is: how to train the next generation of researchers to make them more ready for interdisciplinarity?
My opinion is that it is still legitimate to propose a high level of competence in a given field – understanding in depth the underpinnings and the techniques of a field that gains a scientific identity is indispensable to bringing an individual to the frontier of knowledge and to see how to take steps further. For that purpose, departments in universities remain well adapted.
But this is not enough anymore: it has indeed become very important that students be exposed early in their studies to disciplines other than their major because they address other problematics, have developed other techniques and are centred on other concepts, or even more broadly have developed original processes through which new concepts are formed. Patrick AEBISCHER the emblematic former President of the École Polytechnique Fédérale de Lausanne (EPFL) stated it almost as a principle: it can be called the ‘orthogonality of Teaching and Research’. By this he meant that the organisation of teaching requires some stability to the organisation when another organisation for the research is better suited: one that makes it possible, actually natural, for people with different backgrounds and expertise to work together and tackle problems that bring together people for such a purpose. This leads to a much more flexible structure and the possibility of getting involved in different settings and working with colleagues of different disciplines if this is relevant for the objective.
Still an obligation
As you have understood, in its different forms interdisciplinarity is an integral part, and probably in view of the acceleration of the development of science, an increasingly important part of this development.
In an article on the genesis of Bioinformatics published in 2016 in New Genetics and Society, three researchers from Cardiff University, Andrew BARTLETT, Jamie LEWIS, and Matthew L. WILLIAMS wrote: “While much work on the problems of accomplishing interdisciplinarity has focused on epistemic differences and knowledge “deficits,” and while there has been some recognition of the cultural differences at play…, there has been little discussion of the differences in value systems between “generations.” Different disciplinary cultures find value in different things, valuing different kinds of work (in the wet or dry lab, for example), outputs, and priorities, but value systems also change between generations. Despite widespread valorization of interdisciplinarity as a “good” by agents of science policy, the institutions of science have become accustomed to assessing scientific work on disciplinary lines.” This creates of course a lot of inertia in the system since researchers of the next generation often feel that they still have to prove themselves according to the usual standards of the discipline corresponding to their professional identity. This certainly can considerably limit the implication of the next generation in interdisciplinary activities.
Identifying obstacles to the development of interdisciplinarity is a must because of scientific challenges that need to be tackled but also because of the remarkable added value brought by interdisciplinary projects which ERC ex-post evaluations have shown.. It may be even more important to take appropriate measures at the training level to make sure that the next generation will be given the proper exposure to interdisciplinarity. This concerns universities as well as research organisations. At the level of funding agencies there is also the need to continue efforts to monitor and improve the evaluation of interdisciplinary projects.
I thank you for your attention.