1 Introduction

How may physics help us to understand the phenomena of life? This question has raised substantial scientific, historical, and philosophical attention throughout the history of science, particularly in the 20th and 21st centuries. Physicists have often been attracted to the mysteries of living matter, migrating to biology for several reasons. These include the application of methods and practices rooted in physics to tackle biological problems (see, e.g., Fleming 1968; Olby 1974; Kay 2000), the expectation that living matter could be reduced to physics (see, e.g., Fuerst 1982), or, after the military use of atomic energy, the appeal of a science attached to life over a science attached to death (see, e.g., De Chadarevian 2002). Among the many physicists who have attempted to understand the phenomena of life from the perspective of physics are important figures such as Niels Bohr, Pascual Jordan, Max Delbrück, (about these historical figures, see Joaquim et al. 2015), Erwin Schrödinger, Léo Szilárd, Nicolas Rashevsky, George Gamow, Seymour Benzer, Walter Gilbert, James Till and Francis Crick. The list of physicists who have played a memorable role in biological research is long and other important names are probably missing here. The history of the relationship between physicists and biology is rich, fertile and impossible to deal with fully in this paper. Our focus lies rather in the current migration of physicists to biology and their role in biological research, particularly in the search for systemic biological explanations.

We carried out an empirical study focused on the field of systems biology. The field finds its roots in a quantitative approach to biological phenomena, especially at the cellular level. Systems biology has benefitted from contributions from several disciplines, such as physics, mathematics and computer science. More recently, the use of the term integrative systems biology (ISB) has become widespread. ISB is an emerging field that applies and develops mathematical and computational methods to model large-scale biological systems. It integrates mathematics, computation, a variety of quantitative concepts and methods, and biological experimentation.

We used oral history as one of the methodological tools to gather the empirical material, conducting interviews with physicists working in systems biology, but we also based our results on observations made in their laboratories, informal conversation with research group members, occasional group meetings, lectures, and so on. The approach is explained in further detail below. The goal of this paper is to offer a contribution to build a richer picture of integrative aspects of systems biology and to understand this interdisciplinary field in a more sociologically and philosophically informed manner. Moreover, we are concerned with communication issues within systems biology and in particular with the sources of miscommunication, whether they lie in an epistemological or linguistic sphere.

In this paper, we focus on several specific issues: the reasons why physicists move from physics to biology; the extent to which they adapt their practices to biology and the extent to which they become part of the biology community; and how physics itself contributes conceptually and methodologically to biology, according to the interviewed physicists. We conclude that there are common reasons to move, that the transition must be understood in terms of degrees, and that there are typical features of systems biology that are rooted in physics, such as the search for general principles. We also consider questions about interdisciplinarity, namely, the ways in which both fields can and should be integrated. Additionally, we address challenges regarding the coexistence of many epistemological cultures in the scientific community and discuss results from the interviews that illustrate cultural issues concerning biologists and physicists and their distinct ways of thinking. Many cultural issues come along with consequences, particularly to the exchange of ideas in the community. Several episodes of misunderstanding were reported in the interviews. For instance, we have seen that the judgment of what is considered as a model is a matter of interdisciplinary debate. The interviews illustrate several misunderstandings that are more epistemological than merely linguistic and, consequently, indicate that some accommodations are necessary. Before getting into the details of our empirical study and its results, let us provide a brief historical context.

1.1 Physicists going into systems biology

The present work focuses exclusively on recent science. In the post-genomic era, the great scientific challenge of converting an unprecedented amount of data into knowledge has been dependent on multidisciplinary skills. In particular, the so-called omics (genomics, proteomics, etc.), systems biology and synthetic biology draw on theoretical and methodological approaches that strongly involve interdisciplinary research. Physical scientists — among other experts — have been required in biology for support, particularly quantitative support.

Throughout the final part of the twentieth century, our understanding of biological systems has changed substantially. The development of molecular analysis techniques and tools has given rise to a huge amount of data at the molecular level of living systems. It has increasingly become clear that a restricted focus on sequencing was not enough to provide a full understanding of the systems and, therefore, systemic approaches came to the fore (see, e.g., Ideker et al. 2001; Kitano 2002, Hood 2003, Carusi 2008). Sidney Brenner (2010, p. 207) summed up the situation metaphorically: “Sequencing the human genome was once likened to sending a man to the moon. The comparison turns out to be literally correct because sending a man to the moon is easy; its [sic] getting him back that is difficult and expensive”.

The research on protein folding developed by the physicist Eytan Domany — who was an interviewee in our empirical study — and his group exemplifies the shift in focus from sequences to dynamics. Their theoretical approaches aim to predict the structure of a protein from its sequence by using methods from statistical physics and computational tools. Sequencing turns out to be a step towards tackling the crucial problem: the ways a protein chain folds into complex shapes, according to physiological conditions and evolutionary factors, in order to perform a specific function. (see Domany 2000).

Overall, the scientific challenge for those concerned with living systems has been to deal with more dynamic and systemic problems, and also with big data. For this task, biological research has increasingly relied on computational methods and high-throughput technologies and consequently on the skills of those trained to deal with complexity in other disciplines (see e.g., Auffray et al. 2003; Ideker et al. 2001). In this context, new institutes, programs, departments, conferences, chairs and journals dedicated to systems biology have proliferated (Agrawal 1999; Powell et al. 2007), clearly showing the institutionalization of the new discipline in the sense of a dynamic structure “for assembling, channeling and replicating the social and technical practices essential to the functioning of the political economy and the system of power relations that actualize it” (Lenoir 1997, p. 47).

It is important to clarify, however, what we mean by systems biology. The term is a flexible one, since it encompasses different kinds of analytical approaches (see Keller 2005a, 2007). Previous attempts to apply systems theory to biology in the past, notably by von Bertalanffy in the 1930s and Weiss in the 1950s, established a systemic approach that strongly influenced many scientific endeavors. However, these approaches did not generate an institutionalized scientific field at the time. We focus here on systems biology as a present day scientific field, which has reached the levels of institutionalization mentioned above. Therefore, we take systems biology as the study of how molecules and cellular components interact and come together to give rise to sub-cellular machineries that are capable of the operations required for physiological functions, dynamics and processes. It encompasses both top-down and bottom-up approaches, i.e., starting both from a description of whole systems and going down to the components, and from cellular components to the higher-level system (see e.g., Bruggeman and Westerhoff 2007).

Systems biology is an integrative field that investigates biological systems by combining experimental practices with theoretical work for predictions and model building using a mathematical language. Quantitative techniques are applied to analyze large database sets collected from wet-lab experiments (e.g., clustering, data visualization techniques, network construction, and gene-set enrichment analyses) and subsequently to model phenomena, systems, and processes of interest. To perform such combinations of tasks, the field of systems biology gathers biologists, physicists, mathematicians, computer scientists, and engineers with the goal of extracting knowledge from biological data (e.g., Allen 2001; Kell and Oliver 2004; Fan et al. 2005; O’Malley and Dupré 2010; Frické 2015).

Physicists are, once more in history, playing a central role in biological research. The set of activities related to the quantitative scenario of systems biology is totally suitable for physicists, even though many of the issues – such as how to incorporate experimentation into quantitative work – are not solely issues for physicists but for any quantitative people in the field. Our focus on physicists was inspired by Keller’s argument (Keller 2005a) that physics is the very discipline claimed to be an important candidate to offer a theoretical framework to systems biology, as well as by the relevance of current contributions from physicists to systems biology and by the historical significance of the long standing relationship between physics and biology. There seems to be a scarcity of written sources about this issue, at least in terms of peer-reviewed papers, something not uncommon in works on contemporary science. The topic is more often discussed through sources such as opinion papers, editorials, features, and synopses (e.g., Knight 2002; Ouellette 2003; Wolgemuth 2011).

Physicists are joining biological departments with the conviction that their mindset may provoke an impact on mainstream biology. This state of affairs raises a number of questions, such as: What are the circumstances under which physicists approach biological problems in systems biology? From the perspective of the physicists, what kind of interdisciplinary challenges must be tackled?

1.2 When disciplinary worlds collide: Issues concerning physicists working as systems biologists

Due to its distinctive interdisciplinary character and demands for broader collaboration between scientists from different backgrounds, systems biology has become an important subject of sociological and philosophical investigation. The introduction of new methods into biology is generating novel methodological strategies for managing complexity that researchers are only beginning to investigate from philosophical and sociological perspectives.

Researchers have been examining fundamental issues in systems biology and asking for more sociologically- and philosophically-informed collaborations (O’Malley and Dupré 2005). For instance, Calvert and Fujimura (2011) and Kastenhofer (2013) investigated epistemic cultures and tensions in systems biology. Rowbottom (2011) and Fagan (2016) explored interdisciplinary collaboration inside the scientific community, including barriers that confront attempts to work across disciplinary boundaries. Fagan examined a case of failed interdisciplinary collaboration between experimental stem-cell biologists and theoretically-inclined modelers. Rowbottom explored issues arising from the interaction between condensed matter physicists and molecular biologists, particularly related to modeling practices. MacLeod and Nersessian (2016) performed an empirical ethnographic study of systems biology labs. They mapped out some of the methodological scenery of problem-solving within the interdisciplinary field of systems biology. Recently, MacLeod (2018) illustrated some of the cognitive barriers between molecular biologists and systems biologists, which have contributed to the failure of interdisciplinary interactions.

An important body of literature concerns modeling practices. For instance, MacLeod and Nersessian also investigated how integrative systems biologists deal with the complexity of the systems they are managing, and the constraints on model-building, through the affordances of model-based reasoning (MacLeod and Nersessian 2013a), as well as how integrative systems biologists build simulation models in the absence of a theoretical base (MacLeod and Nersessian 2013b). From a philosophical point of view, Carusi (2008, 2011, 2016, 2018) analyzed modeling, simulations and visualizations in the context of computational biology. By taking social relations, technologies and symbolic systems into account, she proposes that the very notion of modeling needs to be reformulated (Carusi 2016). She also paid attention to social, institutional and epistemological aspects in the context of the ongoing tensions between mathematically- and experimentally-inclined modelers in systems biology (Carusi 2014).

Of particular relevance to the present paper is a study that used interviews as a methodological resource. Green (2017) interviewed a selection of systems biologists and philosophers to explore their approaches, aspirations and interests, investigating their diverse views on significant aspects of systems biology, as well as on its future directions. She also analyzes, elsewhere, current methodological strategies in systems biology (Green and Jones 2016, Green et al. 2018, see also section 3.5.).

We mentioned above some recent literature, however, it is also indispensable to mention some classical laboratory studies that adopted an ethnoanthropological point of view and paved the way to studies like ours. They tackled epistemological questions by assuming lab dynamics as belonging to an autonomous culture, addressing how certain entities become objects of research, how scientific knowledge is constructed, what is adequate knowledge, and other notorious conundrums. Karin Knorr-Cetina, for instance, provided an anthropological description of the knowledge cultures of science, introducing the term “epistemic culture”, that is, a culture that creates and warrants knowledge. By the term she refers to “an amalgam of arrangements and mechanisms — bonded through affinity, necessity and historical coincidence — which in a given field, make up how we know what we know” (Knorr Cetina 1999, p.1). Her main tenet is that different laboratories do not share the same way of producing and validating scientific knowledge, as they are endowed with distinct epistemic cultures. She is concerned about the machine deployed in knowledge production and looked upon scientists as embedded into construction machineries that are organized dynamically without being governed by single actors. The notion of epistemic cultures represents a useful guiding assumption of the present research: epistemic cultures constitute specific ways of producing knowledge, which determine different ways of inquiring and interpreting science.

While Knorr-Cetina aims at a sociological/anthropological description of cultures of scientific knowledge, Evelyn Fox Keller (2002) focuses on a description of the epistemological assumptions of these cultures. She looked at key twentieth-century figures under the lens of what she calls epistemological cultures, that is, “the norms and mores of a particular group of scientists that underlie the particular meanings they give to words like theory, knowledge, explanation, and understanding, and even to the concept of practice itself”. (Keller 2002, Cf. also 2005a, 2007). Through a sophisticated analysis of the cultures of physics and biology, particularly in relation to the problems of developmental biology, Keller pleaded for proper attention to the meanings of the words used by scientists and the way they use them, as well as to the linguistic and narrative dimensions of explanations. The notions of epistemological culture and the appreciation of linguistic clarity in science also played a guiding role in our analysis.

Another guiding theoretical reference to our work is Peter Galison’s study on the collaboration of instrumentalists, experimentalists, and theoreticians in high-energy physics. (Galison 1997, cf. also Galison 1996). He developed the metaphor of “trading zones” to explain how physicists and engineers from different cultures worked together to develop particle detectors and radars. To explain their successful communication, Galison examined the movement of ideas, objects and practices in the context of the establishment of pidgin and creole languages, and claimed that two different groups are able to find a common ground to communicate by means of such languages. From this perspective, we can, then, ask whether contemporary systems biology may be a case of a successful trading zone.

2 Physicists as interlocutors

We conducted and recorded semi-structured interviews with thirteen leading physicists working in systems biology in four different countries: Brazil, Germany, Israel, and the United States. In Germany, we conducted five interviews with three research group leaders, Nikolaus Rajewsky at Max Delbrück Center for Molecular Medicine, Hanspeter Herzel at Humboldt University, and Peter Arndt at Max Planck Institute for Molecular Genetics, and two postdoctoral researchers, Roman Brinzanik and Navodit Misra, also at Max Planck Institute for Molecular Genetics. In Israel we interviewed physicists who coordinated systems biology research groups at the Weizmann Institute of Science: Uri Alon, Joel Stavans, and Eytan Domani. In Brazil, we interviewed Suani Pinho,Footnote 1 who is unfortunately the only woman in our sample and is one of the leaders of a research group in Statistical Physics and Complex Systems at the Institute of Physics, Federal University of Bahia. In the United States, we interviewed the following leading physicists: Erel Levine, from Harvard University, Eric Siggia, from the Rockefeller Foundation, and Ned Wingreen and Thomas Gregor, from Princeton University. Research group leaders and postdoctoral researchers have a panoramic view to provide information about their own migration to biology, as well as about their students and co-workers.

Another relevant source of oral information was a conversation with Evelyn Fox Keller, in which we also explored her transition from physics to biology in the 1960s.Footnote 2 All the physicists gave informed consent for the interviews and for the use of the information derived from them. In order to respect the interviewees’ privacy, we do not reveal their identities when mentioning their reports, with few exceptions, for instance, when the report is very related to the interviewee’s particular research topic, laboratory/institute or personal experience.

By using oral histories we adopted a method widely applied in the history of contemporary science. It was first used to study the history of quantum mechanics through the Archive for the History of Quantum Physics led by Thomas Kuhn (Kuhn et al. 1967). The collections of oral histories made available by the American Institute of Physics illustrate its current wide use.Footnote 3 However, oral history, as any other historical method, is not exempt of weaknesses, particularly if it is considered a theoretical resource or historical genre, and not only as a methodological tool. As remarked by Green and Troup (1990), the main problems are related to its inherent dimension of subjectivity, to the mechanisms of production of memories through interviews and to how these memories may evolve over time. Indeed, according to Florence Descamps (2010), oral history is currently a polysemous term, meaning a type of contemporary history, and a historiographical movement, as well as a widely used method in contemporary history. We have used oral history in the latter sense.

Interviews are, however, widely used in distinct scholar fields ranging from the social sciences to the philosophy of science. For instance, Green (2017) interviewed systems biologists and philosophers of science in order to get a glimpse of current relations between these two domains. In the same vein, Schlosshauer (2011) interviewed physicists and philosophers of physics in order to gain a perspective on the issues concerning the controversy over the foundations of quantum mechanics. We have used interviews in our research not for historical purposes but as a resource widely shared by different scholar professionals. While a few questions in our interviews dealt with the interviewees’ past — for instance, the motivation to move to biological research —, most of the questions were intended to capture their current views about their practice in systems biology. Thus, we do not need to discuss the problems arising in the production of memories about the past, which is a common issue in oral histories.

Each interaction happened in person, with the first author meeting the interviewees in their respective countries. In our view, the oral history approach can be greatly enriched by personal contact between the interviewer and interviewee, and is impaired by an exclusively virtual interaction. Online interaction would reduce the complex interactional perception to a mere collecting of reports through a computer. Another advantage of the physical presence of the interviewer in the research environments is that our results also benefitted from observations of the scientists’ working places, including everyday lab routine, offices, group meetings, supervision sessions, lectures, and informal meetings in coffee rooms. The interviewer attended group activities as often as possible, and relevant informal conversations with other members of the groups, students and secretaries also took place and, when possible, were recorded.

The physicists set both time limits and the places where the interviews were conducted. The average time granted for an interview was around one hour. The longest took around 85 min and the shortest, 42 min.

The physicists themselves were of great help in pointing out the main characters within the scientific community working on systems biology. We selected the scientists to be interviewed considering our geographical and financial restrictions. We argue that the geographic dimension and the qualitative approach — which was chosen in order to provide a maximum focus on the details — justify the choice of thirteen physicists. Further studies both in other countries and through quantitative approaches are, however, worth pursuing.

The interview protocols were constructed for each physicist, considering the particularity of his or her research interests, publications, careers, working places, etc. Thus, each interview was preceded by extensive preparation, particularly, researching the interviewees’ career and relationship with the field of systems biology, and working with their papers. They were encouraged to speak about, for instance, their careers, motivations to do research in biology, their background in physics, their research work and group, the culture of systems biology, and issues of interdisciplinary interactions. While the interviewer attempted to provide guidelines for the interviews, she also kept the interaction flexible enough for the physicists to talk freely.

The nearly 15 h of interviews together with other empirical material (i.e. notes, recorded informal conversations, non-recorded informal conversations, pictures, etc.) were qualitatively analyzed. Digitally recorded copies of the interviews conducted in the USA were deposited in the Niels Bohr Library & Archives. The remaining interviews are available to future scholars by contacting the first author. If the person interviewed agrees, a copy of the interview may be shared for the purpose of further studies.

The analysis of the raw data obtained through the interviews was to isolate recurrent topics with historical and epistemological significance. The results must be seen as trends concerning the contemporary movement from physics to biology. In the following sections, a discussion of the main general findings is presented.

3 Results and discussion

The results presented here must be seen as trends concerning cultural challenges in contemporary interdisciplinary biology. In the next three sections, we discuss the reasons why physicists move from physics to biology; the extent to which they adapt their practices to biology and the extent to which they become part of the biology community; and how physics itself contributes conceptually and methodologically to biology, according to the interviewed physicists. Subsequently, we will focus on the challenges regarding the coexistence of many epistemological cultures in the interdisciplinary environment of systems biology. We will also present episodes of miscommunication, with either successful or unsuccessful endings. Finally, we will discuss local strategies to overcome cultural challenges, such as the process of learning a new scientific language and the role of the mentor as a mediator.

3.1 Why to move? Motivations behind the movement from physics to biology

The interviewees were motivated to move from physics to biology by a perception that there is something fresh to be pursued in the biological sciences. They gave several personal reasons for being attracted to biological research, such as a particular book, a particular person, or the awareness that they could apply their knowledge and methods to innovatively solve problems in the new field. Nonetheless, one reason has been often enthusiastically highlighted: the perception that biology has a “smell of breakthrough in the air”, as the interviewee Eytan Domany elegantly put it. This comes together with a sense that the field of physics has reached satisfactory maturity for the time being.

In physics, they claim, there is a higher likelihood of getting involved with well-known problems, in the sense that they have been established and thoroughly analyzed by the great pioneers of the past. A physicist reported that previously, when working in the field of condensed matter physics, much of his work was to find what scientists like Philip Warren Anderson and Lars Onsager have not done: “You have to look at the little corner which has been relevant back then when they have done the theory”. Nevertheless, the fascination for the field of physics remained strong among the interviewees and was often combined with some initial aversion to biology, as we can see, for example, in the following comment: “Why would any physicist want to waste his or her time with something that is so soft and irreproducible?” (Physicist 12). In turn, biological research and, in particular, systems biology and the even younger field of synthetic biology offer room for an abundance of questions that are still wild in a unique way. Physicists tend to engage with these questions, to the extent that they are related to the theory and modeling of complex systems.

We want to illustrate this process of tackling new and stimulating questions by commenting upon the prominent research topic of RNA-based regulation, which is currently being investigated by many of our interviewed physicists (Erel Levine, Eric Siggia, Eytan Domany, Hanspeter Herzel, Joel Stavans, Thomas Gregor, and Uri Alon). The noncoding RNA molecules are transcripts that function directly as structural, catalytic or regulatory RNAs, in contrast to mRNA encoding proteins. Biological research over the past years has revealed that noncoding RNA molecules are involved in significant regulatory control mechanisms of gene expression (see, e.g., Eddy 2001; Daneholt 2006). Findings regarding gene regulation have transformed our views about noncoding RNA and exciting new research questions of great relevance have started to emerge and have attracted many researchers, including physicists.

The approaches towards these questions are inevitably rooted in the disciplines the scientists come from. When studying regulatory machineries in different organisms, such as bacteria, worms, drosophila and alike, biologists are often concerned with the evolutionary link between the species, since their mindset is commonly embedded in evolutionary thinking. Physicists, in turn, tend to look for unifying principles that dictate regulation, since their thinking style relies on the assumption that organisms share the same physics. Keller (2002, 2005a) conjectured that physicists and biologists ask different kinds of questions and look for different kinds of answers. In our empirical study, her assumption is supported.

The general point we want to make is that the contribution of physicists to biology goes far beyond the technical application of knowledge, methods, and problem-solving approaches. It reaches the process of question making in a field perceived as being less established and more in flux than physics. This state of affairs is extremely attractive for physicists. Physics is undoubtedly a field of big questions, but sometimes, as in any field, physicists may feel unmotivated. Physicist 11 reported: “I got a little bored with physics and drifted to biophysics and biology”. In the new field, physicists seem seduced again by the possibility of being the first ones to know a novel bit of the furniture of nature.

It is evident that the awareness of possible upcoming breakthroughs and the consequent motivational feeling are embedded into perceptions concerning the history of physics and the history of biology. In particular, as mentioned above, there is the distinctive impression that the field of physics has reached a state of maturity, whereas the field of biology seems to be the place where the action is (Wise 2004, 2007). A popular historical account among the interviewees was that, although biology has been enormously successful in the past by asking questions that do not necessarily require mathematics, at some point it turned out that without quantitative approaches the distance between what molecular biologists were doing “and the understanding of the phenomena that they were really interested in was growing” (Physicist 6). Therefore, the physicists place themselves in this moment in history and claim that the field of systems biology experiences a great creative phase that partly explains the transition. However, it is worth emphasizing that many other factors are involved in their moving from one field to the other, particularly more pragmatic ones, such as the proportion of funding and positions, which seems to be currently larger in biology than in physics (Keller 2005a).

It is not unprecedented in history that the perception of distinctive creative phases turns into a reason for changing fields. For instance, Max Delbrück related his transition to biology to the perception that quantum mechanics has become “the final word” on the “behavior of atoms”, while biology was a field that “was not yet at the point”, where they were “presented with clear paradoxes” (Delbrück 1949). Even though Delbrück was referring to his transition in an earlier phase of physicists moving to biology, there seems to be a strong similarity between his perception and those reported in our interviews. Many interviewees consider themselves as lucky to be “in the right place at the right time” (Physicist 6). Some physicists explicitly expressed a comparative view that “biology is today what physics was in the first half of twentieth century with the advent of all the big revolutions” (Physicist 7), since “the things are moving in an extremely fast pace” (Physicist 8). To sum up, the feeling that in comparative terms biological research is considered by many physicists to be at earlier stages than physical research (see Keller 2002) is an important and recurrent factor underlying physicists’ move from one field to the other.

3.2 To what extent? Different degrees of commitment to biology

To move, or not to move: that is not the only question a physicist faces. The transition to biology must be understood in terms of degrees. Therefore, a crucial question for a physicist-turned-biologist is: to what extent do I get involved with the new field?

One can find many historical examples of great physicists who worked on biological topics to different degrees. On the one hand, we may point out discrete interactions, as in the case of Niels Bohr, who speculated on the relationship between quantum theory and biological phenomena from a theoretical point of view, or Pascual Jordan, who developed Bohr’s suggestions and proposed to unify quantum physics and biology (see e.g., Joaquim et al. 2015). Or, later, George Gamow, who worked on the specific problem of protein synthesis, after becoming enthusiastic about the double helix model (see e.g., Segrè 2011). On the other hand, we can mention stronger commitments, such as those of Seymour Benzer, Max Delbrück, and Francis Crick, who have built a whole career in genetics and molecular biology (see e.g., Kay 2000).

Nowadays, along with the rise of more quantitative approaches to biology and, consequently, the increasing ways through which a physicist can apply technical and intellectual knowledge to the new field, the range of degrees of involvement has expanded. Physicists can apply their mathematical skills, and may also go into the wet laboratory and engage with experiments involving organisms in various ways. The different research strategies connected with different ways of getting involved in biological research leads to different speculations and abstractions, and result in distinct degrees of closeness to the biological systems (MacLeod and Nersessian 2013a; Carusi 2014). Thus, there are many degrees of proximity between the physicists and the biological realm, and, accordingly, different ways and degrees of moving, as an interviewee (Physicist 2) pointed out, according to his own experience:

I will define a move myself: you can stay in physics, you use your tools, your models to analyze some data and some biological problems. This I wouldn’t call a move. It’s more an application of a certain concept to other types of data. (…) Until ‘96 I was a physicist applying techniques to biological problems. In ‘96, I got the chair here in theoretical biology and this was a real move (…) not only in terms of institute (…) but also in spirit. Until ‘96 I had a lot of methods and some applications. After 2000, I had a lot of biological problems that I asked myself critically ‘can I contribute to these topics?’ (…) Then I learned biology over the years, started collaborations, got to know experimental data. Then I had a new topic. (…) It was not a move; it was a graduated transition (…) from thinking like a physicist to now. In the last 10 years, I feel a bit like a biologist.

Prima facie, the fact that there are many degrees of involvement may seem an obvious fact, but it has important implications, particularly for the laboratory structure and organization, as well as for institutional policies. In the next paragraphs, we will discuss these implications from the interviewees’ perspective as well as from our own field observations. The interviewees were encouraged to talk about their research environments and communities, and, due to the fact that the interviewer visited their institutions, she could visit a number of labs, offices, departments, and institutes. Here we describe some observations alongside oral information provided by the physicists. For those acquainted with the work in a systems biology laboratory, the present account will look very familiar.

Each research group we visited is strongly interdisciplinary and gathers biologists, physicists, computer scientists, among others. The division of labor varies from lab to lab. There are people doing pure experiment, people doing pure theory, and people doing both. The organization of the workspace also varies remarkably in terms of the disciplinary setting and the source of the biological data. For example, at the Weizmann Institute of Science, in Israel, the interviewer visited two independent research groups, both with wet and dry labs side by side. Nonetheless, one of them is located in the Department of Physics and the other in the Department of Biology. She also interviewed a third physicist and research group leader from the Weizmann Institute of Science, located in the Department of Physics. At Harvard, in the United States, the office building and the laboratory building are walking distance from each other. At Humboldt University, in Germany, the biological data come from collaborations outside the campus. At the Federal University of Bahia, in Brazil, the biologists usually go to the Institute of Physics for the official meetings and the biological material upon which mathematics is applied comes from another university in the same state, from sources located in another Brazilian state, and from international databases.

Three main research approaches were identified, which are not isolated but often combined: experiment-oriented, theory-oriented, and tool-oriented. The groups that perform wet experiments, such as the group of Rajewsky, at the Max Delbrück Center for Molecular Medicine, in Germany, deal with biological matter by applying techniques from molecular biology and biochemistry in order to generate data. The theory-oriented groups with no wet lab must either work in close liaison with experimental labs or find database sets collected from wet labs. For instance, Brinzanik and Misra — from the Max Planck Institute for Molecular Genetics, in Germany—and Suani Pinho — from the Federal University of Bahia, in Brazil — do not go to the wet lab to perform high-throughput analysis of cancer material. All the groups had a theory-oriented section. Finally, some groups were also concerned with the development of the instruments and artifacts they apply, such as Gregor’s group at Princeton University, in the U.S., which builds custom microscopes for live imaging.

In a similar fashion, MacLeod and Nersessian (2013a, 2014) analyzed the affordances and limitations of unimodal and bimodal strategies for integrating modeling and experimentation in systems biology. The unimodal strategies rely on collaboration between experimenters and modelers in distinct laboratories and focus more on mathematical or algorithmic methods to reduce complexity. The researchers only do modeling in collaboration with experimental researchers outside the lab. In a bimodal strategy, in turn, modelers perform their own experiments, i.e., researchers conduct their experiments in the course of building models. According to the authors, the bimodal strategy is a novel and difficult strategy pursued by a small number of researchers, as it requires both high-level modeling and experimental skills. The unimodal strategy is, then, the predominant mode of working. They also claim that both strategies have advantages and limitations. Their division of strategies is supported by our empirical observations, as we witnessed both unimodal and bimodal strategies at work in the researchers’ work places with a predominance of the former. This division also seems to represent well distinct views and opinions in the field of systems biology about how to approach complex biological systems, as far as our data allowed us to ascertain.

When it comes to considering the migration of physicists, we can conclude that, in addition to the question we posed previously — to what extent does a physicist involve herself/himself with the new field? - there are still two crucial derived questions for the physicist-turned-biologist, which relate to the distinction between unimodal and bimodal strategies: Do I go to a wet lab and handle biological matter? Should I work at the biology building?

The first question is to be tackled by taking into account the research circumstances and the scientists’ personal career interests. According to our observations, a research group leader plays a crucial role in this decision. Physicist 6, who is also a group mentor, defended the need to stimulate interactions between scientists, so that the biologists try to plan the experiment together with the physicists, and the physicists, in turn, invite the biologists to plan the analysis together. In his lab, the physicists are encouraged to understand the wet experiment in further detail. Some other groups were less concerned with these interactions. For instance, there was no biologist in one of the groups we visited, since the group leader does not see any need for close interaction between physicists and biologists. He (Physicist 11) explained in his interview that the people he hires are exclusively physicists, mathematicians, and computer scientists, due to their stronger quantitative approaches, and the biological material comes from other labs.

Regarding the second question mentioned above, — Should I work at the biology building? — we would like to comment on what we perceived as a historical innovation in institutional terms. Today, cutting-edge research environments often present exclusively theoretical labs located in biology departments, and experimental biology labs located in physics departments. Therefore, very similar workplaces, in which similar scientific questions are asked and similar aims are pursued, may be in distinct buildings under the banners of physics or biology. It was even reported: “I’m sitting in physics but I could sit in biology, anyway.” (Physicist 8). Still, just because a lab is located in a department of physics this does not mean that it is necessarily theory-oriented. Joel Stavan’s lab is a graceful example: in his interview, he emphasized the strong experimental point of view of his group, which is based in the Department of Physics of Complex Systems, at the Weizmann Institute of Science, in Israel, and includes a wet lab. He even commented on the process of building up the wet lab in the beginning, when he came from physics and had to learn the biological experimental world from scratch. Another possible format is a lab that belongs to the environments of both biology and physics, such as Levine’s lab, at Harvard University, which is part of both the Department of Physics and the Center for Systems Biology.

By visiting most interviewees’ work places, we observed that the diversity of institutional formats provides a new intertwined picture of the interfaces between physics and biology. Thus, the current scenario may amount to a unique episode in the long historical relationship between the disciplines, in which physicists, biologists and their respective research places are combined in complex and miscellaneous ways. It seems that it is the first time in history that the interface between physics and biology reaches such a complex and concrete configuration. From a more philosophy-oriented approach, MacLeod and Nersessian (2013a, 2014) report an important connection between philosophical divisions in systems biology and the structure of research in the field. Philosophy of science is in need of a detailed picture of interdisciplinary research practices in systems biology (Cf. also MacLeod and Nersessian 2016, MacLeod 2018), and this effort can benefit from oral history approaches like the one we present in this paper. Clearly, such a scenario has many implications for research policies, funding structures, and university teaching. It is worth pursuing further historical, philosophical and sociological investigation on this institutional interface and its new configurations.

3.3 To which effects? Intellectual contributions and philosophical issues

Physicists tend to search for general principles and for simplifying the systems under study: that is the gist of their thinking and the way they provide satisfactory explanations, and, consequently, the way they look at biological systems, the questions they ask, and how they search for answers. In our study, this is particularly visible when it comes to comparing modeling styles. The following quote from Physicist 1 illustrates a popular claim among the interviewees:

I think there are some differences when you discuss a certain problem with a biologist and a physicist... let’s say there is an idea of how to dissect the problem or how to solve the problem... and the biologists will all the time and in many cases they will tend to bring in ‘yeah but this you haven’t really considered in your model’ (...) they have been used to easily add the missing layers of complexity in the model right away (...) the physicists tend to try to simplify the problems with the hope of some unifying principles and try also to get a clearer understanding of the scales involved..... ‘maybe there is the complex level A and B but maybe the A is only important for complex data regarding certain scales which is different from complexity B. Maybe you shouldn’t be so concerned about the complexity type B because you are only interested for now maybe in the world living on the scale A.

As this interviewee states, physicists ask different questions from biologists: they usually favor global accounts over detailed descriptions and search for systemic explanations by looking for general principles. That is, they are more inclined than biologists to rely on what Batterman and Rice (2014) call “minimal models”, models that are explanatory due to a story about why a class of systems will display the same large-scale behavior because the details distinguishing them are irrelevant. It was reported that the expression of such a style is evidently stronger or weaker according to the physical sub-cultures the physicists are trained in. On the whole, for a long time the physics community has been using global accounts, as conceptual tools that are very powerful in explaining many aspects of reality, such as the principle of conservation of energy or the principle of least action.

Another aspect suggested by this interviewee’s statements is that a crucial problem is: how to tell what exactly must be considered in a model? Overall, it was often reported that biologists tend to consider more factors, given their presumed descriptive tradition, the fact that they are mainly concerned with the biological reality behind a phenomenon, and their supposed emphasis on its complexity, and physicists tend to consider fewer factors, since they see themselves as oriented towards simple (even minimal) models that uncover general principles and are mainly concerned with equations or other mathematical aspects behind a phenomenon. These differences concerning models and the search for simplicity and/or generality will be pursed later in this paper (section 3.5).

Finding general principles is a major goal for contemporary systems biology. It may be a winning choice, but it is crucial to devote critical attention to evaluate in which situations it is useful to simplify, in the search for maximizing generality, and in which situations it is a better choice to pay attention to the complexity, historicity, variation of living systems. Moreover, it is important to understand the kinds of analytical tools suitable for inquiring styles maximizing generality or not.

Another typical outcome of the physicists’ mindset is the search for new physics in biology. Historically, this is a recurrent topic: Niels Bohr, Max Delbrück, and Erwin Schrödinger, for instance, investigated the idea that new physics could emerge from the study of life (see Bohr 1933; Schrödinger 1944; Delbrück 1949), though they differed in their assessment of what kind of new physics it was likely to be (see Stent 1998). Today, along with the advances in our understanding of living systems, there are reshaped hopes of finding new physics in the unique properties of living matter. The question is the same: is there something that has been hidden and cannot be easily revealed in the inanimate world? In the face of what scientists currently know about biological complexity, the search for a new physics has turned out to be more a hope or a maturing goal than an acknowledged research aim. For instance, physicist 13 considered the hope of finding some new physics in the living world as a long-term goal, instead of a research question to be framed in a project, and physicist 10 called the quest for fundamental principles in biological systems his holy grail.

The inquiring styles traditionally based on physics that we described above must be taken carefully, particularly due to the conspicuous differences between biological and physical complexities. Complexity in biology is fundamentally different from the homologous concept in physics: it has its own specificities concerning, for instance, constraining factors, hierarchy, nonlinearity, and non-generality, which come into existence by the evolutionary processes (see e.g., Keller 2005b). Although complex living systems obey the laws of physics, these basic laws do not fully explain their behavior: the multiple interactions generate irreducible emergent properties (El-Hani and Emmeche 2000; Cohen and Harel 2007). Physics is indeed all over the place, but so is evolution, as mentioned in one interview: “if you cannot isolate a subsystem, then you have to address the whole cell as a whole body, and then physics is very limited. (…) Biology is a kind of history. The history of evolution dictates what kind of solution is found” (Physicist 2).

Finally, we would like to comment on some other assumptions grounded on physics reported by our interviewees. Their discourse varies from complex to simple narratives regarding philosophical debates about their work. They often engaged in self-reflection during the course of the interview, raising philosophical issues such as: what counts as theory and as experiment in physics and biology; the distinction between theorizing and modeling; conceptual differences between the disciplines (concerning, e.g., laws, complexity, emergence); the status of constraining factors in physical and biological systems; what the experimental status of computer simulations is - Are computer simulations in biology considered experiments as in physics? Or running a computer program is just part of the theorizing process performed once the experimental approach is over? These are concerns of both scientists and philosophers of science (e.g., Winsberg 1999; Galison 1996; Keller 2000, 2003; Carusi 2011, 2014).

The interviewees gave both similar and dissenting answers to the many philosophical doubts and showed both appreciation and dislike for comparisons between the fields, as well as for alternative views of the philosophical topics they raised. An interviewee, for instance, raised a few speculations: “I think phenomenology is an obscene word in biology. You are not allowed to say this word so instead you say modeling, or… but from the theoretical perspective, it is not even clear. So we have our personal belief, and we do believe that there are fundamental principles to biology... but can we prove that they are there?” (Physicist 10).

Our findings support the claim that systems biology is still a field in search of a philosophical foundation (Cf. e.g. Boogerd et al. 2007, Keller 2005a), that the field has no theoretical base (Cf. e.g. MacLeod and Nersessian 2013a), that systems biologists are developing their own philosophy of science (Cf. e.g. Calvert and Fujimura 2011), and that scientists and philosophers need to develop sociologically-informed philosophies of systems biology in order to offer guidance to scientific practice (O’Malley and Dupré 2005; MacLeod and Nersessian 2016). Given the increasing popularity of systems biology, or more broadly speaking, interdisciplinary research on biological systems, we need to face philosophical issues head on, increasing the thorough concern and conscious debate about epistemological aspects of this new field. They are crucial to forging productive research strategies. Through studies like this, we hope to contribute to articulate a foundation for this forging.

3.4 Cultural issues between biologists and physicists

As physicists and biologists come from two clearly distinct epistemological cultures, they have different traditions, goals, and ways of dealing with the unknown. Physicist 6 reported in interview that they face problems like those of an intercultural communication:

This is a major topic that I struggle with over the years and I discovered ways to work but it’s something that really needs a lot of work: biologists and physicists, and computer scientists they all come from different cultures. It’s almost like people from very different countries, like two continents.

Physicist 10 also used a similar analogy to address cultural differences:

Every newcomer to the lab or to the field has the same language problem that any immigrant moving to a new country. Even if you did learn the language in the school in your own country, when you come to this new country you very quickly discover that you don’t know what people is talking about, that they are using slangs that you don’t understand, that they are using words in a way that does not make any sense to you. All you need is to overcome this language barrier. The worst thing you can do is to try to hide from it.

The learning of a new language was reported as a crucial step for cultural adaptation. The need for cultural involvement was emphasized, instead of involvement restricted to the mathematical context. As physicist 12 puts:

As physicists we have to come across to learn the language of the biologists to come with an equation and say ‘listen this is what we need solved’ you are gonna wait a long time. You have to be able to go and say ‘we understand the terminology (…) what you mean when you talked about activators, repression, transcription factors’. We don’t have to be experts in all the details the way biologists are but we have to be the ones who turn that kind of mental picture of what is going on into a real quantitative mathematical description. You cannot just say ‘you package it for me, and then I will do the math that I’m so good at’, you have to go through how do they do experiments? What should I be careful about in interpreting the data? (…) You have to be aware of all these kinds of considerations and try to formulate what is going on and not be misled by all the complexity of these biological systems.

Cultural learning is certainly not only unidirectional, but bidirectional: physicists must adapt to biology and biologists to physics. However, the demands for these professionals are distinct, as physicists are the ones changing fields and dealing with a new type of system. Physicist 7 declared: “I have found in my experience with biologists that I had interacted with, that it’s we, the physicists, that have to make the much more significant effort to adapt to their language and way of thinking than they to us”. At his lab, the interviewer was kindly invited to talk to other scientists and the hunch that the main effort to adapt to the language comes from the physicists was confirmed by the research group members. But, as an interviewed biologist stated, the adjustment indeed goes both ways:

[…] I should understand their (the physicists’) terms but I cannot learn all those years of education. I just have to understand the main, or crucial terms. But actually [Physicist 7] knows how to deal with people like me, so he would talk to me in a different way or explain me something in a basic way. When I explain the biological system — you can see my diagrams here — I give them not too many details. We communicate in a way that everyone knows how to explain to the other with his/her own terms. That’s the idea. So it’s not only that I adjust, he also adjusts to me.

Through learning and teaching strategies, the scientists develop ways to successfully understand each other. The interviewees reported many episodes in which scientists found a common ground to communicate, overcoming language barriers. Physicist 13 told us a case of successful exchange in his lab:

[…] There are biologists and physicists; these are the two main groups that I have in the lab. They need to learn each other’s language, they need to learn each other’s way of thinking, and it takes a while for them to be able to communicate. One of the most successful work that has been going on in my lab over the last three years was the connection between a developmental biologist — a postdoc who used to work with zebra fish and now works with drosophila — and a physicist — grad student who came in as a string theorist. So think about that, take a string theorist and a developmental biologist, put them in a box and shake it really hard and something nice emerges. That’s the fun part of this job. It took a while at the beginning to make them communicate in the right way ‘cause one was essentially getting the data and the other was analyzing the data”.

However, communication was not always reported as flowing and the literature corroborates this finding. For instance, Fagan (2016) reported that different explanatory preferences can hamper the acceptance of the results of mathematical modelers by molecular biologists. MacLeod (2018) examined domain specificity in scientific practice and how it constrains interdisciplinary research, contributing to difficulties in work gathering scientists from several disciplines. Accordingly, a consequence of domain specificity is that “the technical skills, understanding and experience required to operate within a domain can be opaque or intractable to non-specialists without adequate training” (p. 707). This may lead to communication problems. Here, we claim that the overall situation seems to be in disarray. The interviewees commented on the stress between different worldviews expressed by referees and it was mentioned that sometimes there is even aggression at interdisciplinary meetings (Physicist 6).

3.5 Multiple thinking styles

Keller has emphasized that physicists and biologists have traditionally different ways of thinking (2002, 2005a). In interview, she also outlined that while biologists want to know how systems work, the primary question for physicists is how they could work. Overall, in our interviews, her conjecture was supported. Physicist 10, for instance, reported: “It was very interesting to me how biologists were asking very different questions from the questions I would ask”. This physicist also connected the current migration to biology to this difference in question-making: “there are enough people out there that do fantastic biology (…) the only way to justify (the migration) is that I perhaps ask different questions and come with different approaches”. Following up on it, the interviewee described an episode in which he was working in a collaborative project and raised the question “how many?” to his co-workers, more precisely, “how many proteasomes?”. His colleagues considered it an odd question. As a physicist, asking about quantities is both natural and a priority: “coming from physics, the first thing you want to know is the numbers of what we are talking about”. Along with a sensitivity for numbers, the interviewees claimed that physicists have a particular concern with the bigger picture of a system, that is, general principles, constraining factors, equilibrium and linear laws.

According to Green and Jones (2016), there is a conflict between those who emphasize the explanatory power of mechanistic detail and those who consider this same detail irrelevant. In our study, we report the same: there is disagreement concerning what to consider as relevant in a system (see section 3.6.1). Green and Jones introduced the notion of constraint-based reasoning, emphasizing that mathematical abstraction is supposed to impose formal constraints on a search space for possible hypotheses. Constrained-based reasoning is related to the search for simplicity, as we will describe below.

A fine example of searching for general principles is the research on network motifs developed by the interviewee Uri AlonFootnote 4 and his group (see also Green et al. 2018). In large networks, including biological networks (e.g., gene regulatory, protein, metabolic networks), there is a plethora of possible interaction patterns. Surprisingly, a few types of recurring and statistically significant interaction patterns called motifs have been identified as local properties of many biological networks. They found out that the network motifs appear to function as simple building blocks of transcription networks from bacteria to mammals (cf. Alon 2006; Milo et al. 2002). Accordingly, Alon’s research provides a fundamental understanding of a huge class of systems. This research indicates that one level of simplicity can be generalized to a large set of biological networks (see also Bruggeman and Westerhoff 2007). In interview, Alon pointed at examples of typical inquiries for generalities: “Are there general principles for how this biological matter is made? Why do you see all these particular molecules interacting the way they do? (…) How can precision work despite the randomness?” (Interview with Alon, Weizmann Institute of Science, July 2012).

Both physicists 2 and 7 drew attention to the notion of equilibrium as a distinguishing feature. Physicists have been educated to use the notion of equilibrium in most of their systems of study. In Biology there are mainly non-linear dynamics and an overall tendency to non-equilibrium. Therefore, it is impossible to understand many aspects of biology by regarding living systems in terms of equilibrium schemes, the way physicists do. There are fundamental features of biological systems that physicists are not familiar with. Physicist 7 gives the example of preservation of biological information through generations: “It’s impossible to understand the accuracy of replication using equilibrium ideas, you need to involve non-equilibrium schemes. (…) At least for me it has been very shocking: that some of the frameworks that were very well established in our minds to treat certain systems cannot be applied to biology”.

The search for simplicity was emphasized by most of the interviewees as a typical rationale rooted in physics. The following quotes exemplify the expectation of some underlying simplicity from the physicists’ perspective:

Biology is very, very complex. There is no way one can doubt about that. But there are certain ways in which one can see simple principles that can explain some aspect of why it’s built. That’s what physicists are trained to do and works fantastically when we try to understand the simple stuff like metal, plastic, not living matter. The surprise is that it also works – at least the way I look at science and the results we get – it works remarkably really well if you know the model and you think about biology (…) The simpler, the better for me (Physicist 6)

(…) you realize that there are very simple biophysics in complex biological systems, underlying that there were something that was simple. I think it’s very appealing to the physicists... that’s what the physicist’s training was: finding underlying simplicity (…) If we are lucky and maybe have good partners, and someone has good taste we can dig in to that system and find some underlying simplicity. And then with more hard work and some more thoughts we can build backup from that simplicity to a real understanding of biology (Physicist 12).

Another thinking style mentioned was the traditional distaste of mathematics by the biologists (see also Keller 2002; Carusi 2014). Physicist 10 mentioned that there is “the fear of math (…) when the biologists feel that, by being a biologist, he or she has the license not to know mathematics”. He shared an episode in which a student came to him to ask for help with the theorizing part of her work. She had the experiment and even the differential equation, but she could not solve it by herself. She approached him with the excuse she was allowed to not to know it as a biologist. Instead of solving the equation, the professor used the opportunity to explain two different approaches: she indeed could simply plug her experiment into mathematics, but it would be much better if she understood what the solution actually meant: “solving a differential equation that you already have is not physics, and this we definitely work hard to change”.

Physicists often commented on the descriptive tradition of biology. It was pointed out that biologists are trained to describe phenomena and complexity in detail, which may potentially compromise the search for what is essential. Physicist 5, for instance, criticized an apparent confusion between factual description and understanding: “when I listen to a biologist and they say ‘this gene does it’, ‘we found the function’, ‘this gene regulates that gene’, or ‘this transcription factor bonds to that’, these are facts. It does not teach me anything”.

Finally, an important key difference in ways of thinking is related to the biologist’s evolutionary point of view. Physicist 2 reported: “Physicists are looking for data-related simple principles, simple models, toy models. Mathematicians also use the same equations but they are not close to experimental data. They want to make it per se (…) they want to prove something and so on. So even if they study the same equations, physicists want to explain some effects and mathematicians want to prove something and the biologists are asking how is the functioning and how it evolved”. Physicist 8 puts emphasis on the process of evolution and the need to consider it carefully. He explains that physicists are trained to identify causes and to understand time as moving forward: “There is one thing that happened now, and another thing that happened a minute from now, and what happened now could cause what happened a minute from now, and not the other way around (…) We feel that we understand something if we understand how A causes B”. Evolution, in turn, requires a backward looking.

The physicist argued, through an example, that a lack of understanding of that jump of perspective leads in the wrong direction. The example concerned one of his group’s project, on leukaemia in children affected by Down syndrome. The aim was to understand why children with Down syndrome have much higher probability of suffering from leukaemia. During the study, they made two important findings: that a big fraction of the cells under study has an amplification of a particular receptor, and that a sizeable fraction of the cells that have this particular amplification also have a mutation in a gene coding for a kinase that interacts with that receptor. These two elements together give the cell a boost of cell division, which can lead to cancer, and so, this could be a way of triggering leukaemia. The main evidence was that all those cells that have the mutation also have the amplification. Accordingly, the natural question for them was: what is the relationship between the receptor amplification and the mutation in the kinase gene? “As a physicist”, he reported, “it is obvious that the amplification somehow causes the mutation because in order to have the mutation you must have the amplification (…) So there must be a causative connection between the amplification and the mutation”. However, in the course of the study, it became clear that the gene, which is amplified, is in one chromosome and the mutated kinase gene is in a completely different chromosome. Therefore, the causal relation was not obvious and the question for him turned out to be “what could the causal path be going from the amplification to the mutation?”, and the physicist then “spent sleepless nights trying to figure this out”. The denouement of the story happened through an intervention of a collaborator who was a medical doctor and biologist. This collaborator pointed out that there was no causal relation between the events, as a physicist would expect, and that what might explain the concomitance of the two events could be a selective process. The two events together, which happened by chance a long time ago, conferred the cell an advantage, that is, fast proliferation rates. Thus, the events were selected together, having no causal relationship with one another. In situations like this, the physicist must learn how to look backwards in time and to be careful with inferences of simple causality, such as: A and B are correlated, so A causes B. Perhaps A and B are correlated because of C, and C may be a process of selection taking place in the past: “this is something that a physicist who works in biology should be aware of” (Physicist 8).

The distinct thinking styles we presented illustrate that the agreeing on the research questions and the ways of looking for answers are issues that require mutual adaptation and careful work. It is fruitless to underestimate the fact that “there are different cultures, and it’s very clear that this is a big issue in such interdisciplinary collaborations. In order to bridge the gap in terminologies and approaches, even in research questions, we have to agree in the questions that you want to study together” (Physicist 4).

3.6 Miscommunication across disciplines

Miscommunication across disciplines has been documented in the literature. For instance, Fagan (2016) reported a “failure to communicate” between two different communities, namely, experimental stem-cell researchers and theoretical systems biologists, which was rooted in divergent views of explanation. Chandrasekharan and Nersessian (2013) reported a lack of understanding between experimentalists and modelers. The former often do not understand modeling and the latter often do not know how to frame a request appropriate to the affordances and constraints of experimentation.

In our empirical study, several interviewees reported episodes of miscommunication or misunderstanding within the scientific community, including cases of both that were easily managed and more consequential ones. Physicist 10 told us about a research situation in which it took many months for him to be understood by a biologist colleague. He said the understanding was a matter of time: “I could hear the click and then he got it (…) and now we even have a student that we are co-advising. It just took some time”. In turn, physicist 12 reported a more consequential case, since the misunderstanding affected the development of a paper. The members of the collaborative group meant different things by the word ‘active’ and had to clarify the issue. The interviewee stated that these kinds of misunderstandings often happen:

We just had this experience very recently with one of our collaborations, we were writing a paper together and we meant different things by the word active. One may think: ‘it’s a simple word’. As physicists we meant the probability to be in a particular state that we were calling active and the biologists were talking about biochemical activity which depends on the presence of substrates, as well as the state, and so on. As a result, obviously it was not a long-term problem but these things do come up. I think this is an opportunity to physicists to force their biology colleagues to be very precise in their languages; this is something that physicists can bring.

Similarly, many misunderstandings related to scientific concepts were mentioned. For instance, physicist 3 commented that the term ‘isochore’– which in general refers to patterns of large-scale variation of base composition in the genome – is often understood in different ways. Physicist 10 gave a fine example related to quorum sensing. The lack of understanding of a specific fact — namely, the fact that bacterial cells stop growing and accumulate proteins when they move from the exponential phase to the stationary phase — has resulted in misleading research questions. Physicists who do not know that fact, which “any microbiologist would tell” (Physicist 10), often think that the concentration of protein has something to do with the circuit of functionality of the system, instead of concluding from that fact that there is a phase transition, and this may lead them to raise misleading hypotheses. Therefore, the lack of specific knowledge may lead to mistaken scientific questions.

3.6.1 Modeling strategies in integrative systems biology

The most recurrent issue raised by our interviewees regards distinct conceptions of model and modeling strategies. Modeling is a crucial task for systems biologists, as the growth of biological understanding in this field has been and is still strongly dependent on the formulation of models. However, modeling in biology has been a subject of critical debate. Keller (2002), for instance, highlights both traditional attempts of mathematical biology that have been successful from a standpoint of the great triumphs of the field — such as Stéphane Leduc’s efforts in artificial life, D’Arcy Thompson’s classical attempts on mathematical biology, and Alan Turing’s mathematical model of embryogenesis — and current attempts of mathematical and computational modeling that raise diverse ways of making sense of biological data.

Systems biological models can differ considerably in their connection to empirical results and in their degree of abstraction from the underlying biological systems. This raises a number of important methodological and epistemological questions and calls for philosophers to investigate research strategies for dealing with biological complexity. For instance, the relationship between modeling and experimenting is a central feature of the epistemology of systems biology (Cf. e.g. Agrawal 1999, Winsberg 1999, Keller 2002, 2003, O’Malley and Dupré 2005, Carusi 2008, 2011, 2014, 2016, Calvert and Fujimura 2011, Rowbottom 2011, O’Malley and Soyer 2012, Fagan 2016, Green 2017, Green and Jones 2016, Green et al. 2018, MacLeod and Nersessian 2013a, 2013b, 2014, 2016, MacLeod 2018).

What models are and what roles they intend to serve divide the epistemological cultures in systems biology, as scientists in the field have distinct ways of grasping and approaching modeling. In the words of one of our interviewees, “we end up with total different meanings for the word model” and even though scientists are aware of that, “it still creates some difficulties” (Physicist 12). Our findings indicate that the difficulties lie in two main unconformities: (i) models mean traditionally different things in physics and biology, and (ii) physicists and biologists have distinct and typical ways of modeling. One of the key factors lies in the traditional search for simplicity by physicists and the greater concern with the biological context by biologists, as mentioned earlier in this paper.

Traditionally, models in biology are qualitative or quantitative and physicists often reject qualitative notions of models. Some interviewees’ quotes exemplify such rejection, for instance: “A lot of biological models are not mathematical (…) they are like ‘this does that’, it’s a very qualitative thing (…) I don’t trust that, because there is no number” (Physicist 5). Physicists 3 also explained the difference:

Words like model mean a completely different thing for biologists and for the physicists. A biologist would call a model something which is more a kind of picture or scheme for a physicist. When you take a paper in biology, written by biologists, and they refer to a model (…) they say well we have this model and they put you a picture and the picture is of course very descriptive, it’s capturing thousands of words in one image. But you could substitute the word picture for model and nothing would change. For a physicist a model is completely different. It’s the ability to distill the important quantities of the problem, be able to link them mathematically in one formalism, let’s say a set of differential equations and analyzing these equations in order to make predictions. So conceptually it is very different.

Despite the disciplinary contrast, physicists reported a trust that they find a common ground to communicate: “In physics a model is always somewhat quantitative, while in biology it could be a qualitative model. Of course I don’t interact with all biologists, but with those with whom I do, I guess we have more or less the same concept of a model and that is a quantitative one” (Physicist 3). Even though such trust is expressed, physicists also show concern about the negative consequences of disciplinary differences, such as disagreements in question making and data interpretation. As Physicist 13 reported:

For physicists a model means: ‘I have mathematical understanding of what I am looking at, I can describe my phenomena in the language of math, which gives a very strong predictive power.’ (…) I think the biologist sees a model more in terms of a pathway, ‘well I know protein A attracts protein B and now I look at protein C and how it links to these two proteins’, so it’s a kind of a ball and stick model. The questions you can ask with this type of model are more yes and no questions. (…) In mathematical models that are phrased in mathematical language you get a number in the end and this number can be closer to one scenario or another.

Along with the traditional conceptual differences about models, there are also disciplinarily rooted distinctions concerning modeling strategies. Physicist 2 expressed it: “Typically, physicists look for toy models, (…) minimal models, and they try to reduce the system as much as possible but not any further. Biologists, engineers and mathematicians try to keep complexity”. A source of disagreement between physicists and biologists is the evaluation of what exactly must be considered as belonging to a model. Overall, it was often reported that biologists tend to consider more factors, given their descriptive tradition, the fact that they are particularly concerned with the biological reality behind a phenomenon, and their focus on its complexity. According to the interviewees, physicists tend to consider fewer factors, since they are simplicity-oriented and concerned with the equations behind the phenomena. Physicist 1, for instance, reported such divergence by describing distinct perspectives towards the formulation of a model and the inclusion of complexity in it (please see the first quotation in section 3.3 again).

The obvious risk for the biologists is to include irrelevant complexity, and for the physicists, to omit something relevant. Different explanatory strategies may prioritize or deemphasize details. As we mentioned above, Green and Jones (2016) described this situation in terms of constrained-based reasoning and mechanistic strategies. They are interested in the reasons why some scientists prefer more abstract representational frameworks even if more detailed models are accessible. This preferential interest explains at least in part the different modeling strategies described by our interviewees.

Batterman and Rice (2014) discuss how minimal models are explanatory because they share large-scale behavior despite details that distinguish them but are ultimately irrelevant. This appeal to minimal models appeared in our interviews, as illustrated by Joel Stavans’Footnote 5 explanations about his study on iron homeostasis. He and his colleagues proposed a mathematical model to capture the main features of an observed behavior related to the role of iron in the damped oscillation of gene expression. The iron homeostasis network in bacteria employs feedback loops to regulate iron usage and uptake (Amir et al. 2010). Along with the model building, it was possible to come up with many variables that, a priori, would be important to describe the process under study. However, they aimed at building the model as minimal and simple as possible:

So we showed, in trying to think of a model, for a physicist what is important is to be able to make a minimal model. A model that will reproduce what you see but without spurious variables. To distill what is important. So we came up with a model of three variables and we could reproduce essentially everything we saw. (…) To arrive to what is essential we tried many things, but we were guided by biologists. (Interview with Stavans, Weizmann Institute of Science, July 2012).

Accordingly, the process consisted of knocking down the components in order to get insights on what the components do and when they are important. They knocked down many different genes and, at some point, they discovered that one particular gene destroys the very oscillation under investigation. It was a gene involved in iron transport through the membrane. It became clear that these genes involved in iron transport would be important to describe the oscillation. The investigation encompasses an effort to identify spurious components. It is “a process that builds up in trying to see what are the important variables. A physicist would really try to minimize that to get a minimal description”. (Interview with Stavans, Weizmann Institute of Science, July 2012).

The big challenge is to find the “sweet spot, where the balance is just right” (Rowbottom 2011, p. 149). The adequate level of simplicity when modeling is a critical issue in interdisciplinary communication and sometimes it may be a challenge to find a common ground in the discussion, as physicist 4 puts it: “As a physicist you try somehow to reduce things to simple models, and biologists, for example, they are very often full of knowledge, all kind of knowledge about the details and it’s very hard to find a common ground in the discussion, because biologists talk about many, many different things (…) And a physicist would like to identify which of those are really important. (…) this is a very difficult thing”. Given that, in the interviews we explored local strategies applied to overcome issues of interdisciplinary communication.

3.7 Local strategies to overcome cultural issues

Physicists do not perceive the cultural issues they reported as stumbling blocks. Overall, even when recognizing disciplinary gaps, they are confident they manage to communicate well enough to work together: “Good communication solve all this. Once you start to talk with someone, you spend time with him or her, and then it’s easy to clarify. It’s a matter of will”. The researchers just have to “identify that they have common interests and that they can help each other. Then the rest is just about trying to be very clear and just communicate” (Physicist 10). However, recent literature about collaborative relationships in modeling shows the opposite, that is, that there are many inefficiencies involved in such relationships (Cf. e.g. Rowbottom 2011, Calvert and Fujimura 2011, Chandrasekharan and Nersessian 2013, Fagan 2016, Green 2017). MacLeod and Nersessian (2016), for instance, emphasized that in problem-solving approaches in systems biology difficult tasks appear due to a lack of theoretical structure and shared epistemic values. However, they also mentioned the highly adaptive manner through which researchers manage to handle these difficult tasks, suggesting that the confidence shown by the researchers we interviewed may not be that misguided. They call adaptive problem-solving a “response to the complex problems and methodological uncertainty faced by researchers working in these interdisciplinary contexts” (p. 403). Communication is indeed essential for the task of searching out good collaborative strategies for transforming complex problems into tractable ones.

In order to communicate successfully, the interviewees listed several strategies, such as getting familiar with the language through books, papers, reading materials, seminars, collaborations, etc. Interacting with other scientists is also mentioned by most of them as an important tool for learning. They highlight the need to talk to people from different fields in order to improve communication:

You can still find when you come here people that speak your native language. They might help you. But the more you rely on them, the less you are gonna fit in, and the less you will understand what is really going on (…) it’s true that if you are talking to someone who is similar to you, that speaks your language, the communication is just faster and easier. But when you talk to someone who has different concepts, different ideas, that might require more energy (…) It’s a lot of fun to talk to someone who agrees with you but it’s much more fruitful to talk to someone who does not (Physicist 10).

The adjustment of communication requires time. The physicists mentioned distinct learning periods. Physicist 13 reported a collaborative project between a physicist and a biologist at his lab that needed more than a year to get on the flow. Physicist 6 stated that a physicist can understand a lot in three months and in one year they can give a talk using concepts they hadn’t understand before. Physicist 2 reported that students learn quickly what is needed to satisfy practical purposes, but that problems to communicate remain from group to group. He mentioned two or three years to solve them. Concerning his own experience, he said: “It took me about 5 years to arrive in biology, to know what my colleagues are interested in, and to get a feeling on how are the topics in the journals. Now I know both languages very well”.

Collaborations between physicists and biologists mediated by a mentor were observed and are considered by the interviewees as an important strategy: “We try to make the biologist invite the physicists to plan the experiment together and the physicist invite the biologists to plan the analysis together. And they present to me together” (Physicist 6). The role of the mentor was often emphasized:

I try to give physicists the minimum they need in biology to be able to operate, obviously I cannot bridge the gap of many years of experience in biology. (…) In the same way as I’m not trying to make from the biologist students physicists, I am not trying to make the physicists biologists. Because we would be very poor biologists, in the same way that biologists would be very poor physicists. But there is a minimum of ability to be able to swim between the two. And for me it is very important, as a teacher. (Physicist 7)

MacLeod and Nersessian (2014, 2016) also commented on the crucial role of lab directors. They are the ones who choose the modeling strategy that best matches their philosophical preferences, which deeply shape and structure systems in biological research. Here, we observed that although most of the interviewees see their roles as mentors in connection with an attention to language, a few do not see mentoring as a high priority. Physicist 11, for instance, reported: “they don’t need to be pushed towards the language. I can help them to do some more efficiently by pointing out the right papers, but they don’t particularly need”.

According to the interviewees, the miscommunications are always fixable. In order to tackle cultural barriers, the physicists defend (1) a deeper understanding of biological explanations on their part (together with adjustments on the part of the biologists, such as more commitment to mathematical aspects of research), and (2) more extensive debates among scientists coming from different disciplinary cultures.

Concerning the claim for deeper understanding, they argue that research cannot be overly compartmentalized. Research must not happen in a way that biologists pack the problem and physicists do the math. The physicists are supposed to learn the nitty-gritty of biology and be careful about interpreting data. It is not enough to come with equations and quantitative approaches. Although the physicists themselves claim this, during the lab visits the interviewer observed many different styles of distribution of labor and degrees of involvement with the biological topics.

The physicists also argue that the misunderstandings are clarified by more extensive debate, in which the scientists express their points of views and what they mean or how they interpret the particular cases. This seems to be a feasible solution for specific cases in interdisciplinary research groups, but it cannot prevent the fact that such misunderstandings come up in the whole scientific community. Therefore, questions that remained unanswered are: In fact, how urgent is it to reach precision of language and research aims in systems biology? And what kind of strategy will satisfy the need for linguistic and epistemological clarity in such a community?

This situation is not unprecedented in science (cf. Kuhn 1962 and Galison 1997 on theoretical physics), and not even in biology. Our finding that physicists, as systems biologists, are confident in their communication skills goes along with Keller’s claim that scientists are rarely troubled by the coexistence of several terms they rely upon, that despite the lack of precision in their conceptual language they have no trouble in knowing what they mean (Interview with Evelyn Fox Keller, May 2013, cf. also Keller 2012b) According to her, the lexicons of genetics, developmental biology, evolution and ecology are filled with overlapping terms that researchers do not precisely bother themselves about (Cf. e.g. Keller 2000, 2012a).

As the scientists are confident that they manage to communicate well enough, the field of systems biology may present similarities to what Galison characterizes as a trading zone. Systems biologists from different backgrounds are able to exchange goods, despite differences in language and culture. Their discourse shows that they find local solutions towards good communication, namely, “the trading partners can hammer out a local coordination, despite vast global differences” (Galison 1997, p.783). Further studies and probably even time for the field to develop are necessary to label systems biology as a successful trading zone. If systems biologists succeed, their field will be a further example of how exchanges across disciplinary boundaries can reach established systems of discourse and which are rich enough to support scientific development.

However, it may well be the case of a less successful status, such as a rupture, since ambiguity and polysemy are sources of division in a scientific community (Cf. Kuhn 1962). There is a chance that this community splits in two, but, also, that the overlap of meanings may persist for a long period of time without exerting any actual pressure on the community to divide. The following comment made by one of our interviewees about modeling diversity might be interpreted as a sign of division: “we already have a kind of physicists’ school, I would say, in systems biology and there of course I don’t face this problem because we agree that models should be explaining, they should be simple” (Physicist 4). Another example would be: “In my group of 10 people or so, we have no serious communication problems” (Physicist 2). What would be the way to establish a successful trading zone, instead of an unhappy ending as a rupture?

We believe that a focus on mini-crisis solving is a venturesome strategy; solving local issues may not access global stumbling blocks. As language shapes the questions scientists ask as well as their ways of answering the questions, language has a significant effect on scientific practices. We argue that linguistic clarity and integration of epistemological aims deserve higher priority, if unreservedly flowing communication and productivity are to be established. However, a claim for a proper handling of conceptual precision may easily become a cry in the wilderness, as it is not obvious what is a sufficient level of precision and the scientists, overall, trust their natural precision skills. Besides, imprecision may perhaps play a certain role in communication flow, which is not necessarily negative (Kitcher 1982). Thus, we have a simpler suggestion. As Calvert and Fujimura (2011), we defend that awareness and appreciation are much more feasible goals. Interdisciplinary communication in systems biology could strongly benefit from a greater awareness and appreciation of the epistemological differences between scientists from distinct fields and also of their consequences.

4 Concluding remarks

In order to present the circumstances under which physicists currently approach systems biology, we focused upon the questions ‘why’, ‘to what extent’ and to ‘which effects’, namely, the reasons why physicists move from physics to biology; the extent to which they adapt their practices to biology and the extent to which they become part of the biology community; and how physics itself contributes conceptually and methodologically to biology, according to the interviewees.

Some of our findings are: The contribution of physicists to biology goes far beyond technical application and reaches the process of question making. The interviewees were motivated to move from physics to biology by a perception that the field of physics has reached a state of maturity, whereas the field of biology seems to be the place where the action is. The transition to biology must be understood in terms of degrees and this has important implications, particularly for the laboratory structure and organization, as well as for institutional policies. Physicists have a rationale for searching for general principles and for simplifying the systems under study. They also typically search for new physics in biology. There are typical features of systems biology that are rooted in physics, such as the search for general principles, The most recurrent issue raised by our interviewees regards distinct conceptions of model and modeling strategies. Systems biological models can differ considerably in their connection to empirical results and in their degree of abstraction from the underlying biological systems.

We explored physicists’ discourses on the routine and challenges of interdisciplinary systems biology. These discourses show that quantitative scientists bring not only analytical tools to systems biology but also traditions and values, and, thus, an important barrier to overcome is essentially cultural. The cultural barrier comes along with consequences, particularly for the exchange of ideas in the community. Many episodes of misunderstandings were reported in the interviews, for instance, the judgment of what is considered as a model seems to be a matter of interdisciplinary debate. The interviews illustrate several misunderstandings that are more epistemological than merely linguistic and, consequently, indicate that some accommodation is necessary.

Regarding competition or confrontation between scientific experts, we hope to have made it clear that our approach implies absolutely no intellectual subservience or subordination between the disciplines. Those who misunderstand this may evoke taking sides. After all, scientists are inevitably rooted in their disciplines and, consequently, tend to identify themselves with specific scientific communities and display sets of common values and, perhaps, stereotypes.

A last general point we would like to make concerns the sense of otherness expressed by the physicists in relation to the biologists. We encountered a pool of stereotypes and labels to characterize scientists and fields in the interviews, such as: physicists are traditionally arrogant, biologists are traditionally loathed to face mathematics, physics is the queen of natural sciences, biology is the new queen of natural sciences. There were even cases in the fieldwork in which the scientific territoriality became ethically challenging for us to deal with. For instance, when an interviewee expressed disdain for biologists and even lumped the interviewer in the same category of personae non gratae. It often occurred that an interviewee expressed scorn for a group of scientists and said things like “I found biologists impossible to communicate with” (Physicist 11) or “there are some people who go into biology because they read some review or something (…) and don’t ask themselves for a sec if this (their work) is anyway relevant or interesting” (Physicist 10). Thus, we have encountered manifestations of disciplinary territoriality in the systems biology community, which may come together with intolerance, lack of interest, and even irritation. Fagan (2016) also reminded us that tolerance for different views on explanation is an important aspect of interdisciplinary work.

Overall, the interviewer strived to use a diplomatic approach, by providing guidelines for the discussion based on the protocol, aiming to follow the advice that “doing oral history requires historians to be at once confrontational and collaborative, objective and personal, and suspicious and trusting” (Hoddeson 2006, p. 195). In conducting the present work we were constantly concerned to keep a view of neutrality and non-subservience. There are no enemies in this arena, except in the minds of those that are in the habit of making enemies. The very general claim we would like to make is for a less indoctrinated and more open-minded position. For that, physicists and biologists should overcome authoritarianism, combine respect with critical confidence, set out with the idea of otherness, and thus, favor collaborationism over competitiveness.