Hans-Jorg Rheinberger * **
Experimental Complexity in Biology: Some Epistemological and Historical Remarks
Philosophy of Science
Vol. 64, Supplement
Dec. 1997, S245-S254.
My paper draws on examples from molecular biology, the details of which I have developed elsewhere (Rheinberger 1992, 1993, 1995, 1997). Here, I can give only a brief outline of my argument. Reduction of complexity is a prerequisite for experimental research. To make sense of the universe of living beings, the modern biologist is bound to divide his world into fragments in which parameters can be defined, quantities measured, qualities identified. Such is the nature of any “experimental system.” Ontic complexity has to be reduced in order to make experimental research possible. The complexity of the research object, however, is epistemically retained in the rich context of an experimental landscape, where the eruption of “volcanic systems” can change the scenery dramatically as the result of particular, unprecedented findings.
1. Epistemology. The following remarks are not aimed at a philosophy of biological complexity. Rather, the intent of my deliberations is epistemological. Epistemology is concerned with the spatial and temporal structure, the noumenal modus vivendi, and the material character of the activities involved in engendering scientific objects. It does not pretend to describe or to explain the properties of these objects themselves. Epistemology deals with the properties of scientific practice, the practice of biology in the event.
My claim is this: having a close look at what life scientists do when they occupy themselves with the objects of their experimental procedures may lead to a better understanding of scientific practice in gen-
* Max Planck Institute for the History of Science, Wilhelmstr. 44, D-10117 Berlin, Germany.
** Richard Burian and Lindley Darden are acknowledged for their patience, help, and criticism.
eral. Above all, it seems to me, this concerns the role and the status of sharply bounded concepts as well as the predictive force of explicitly formulated theories.
In closely following the historical path of biological practices, we may arrive at a view of how the sciences operate that is quite different from views that have taken classical physics as their point of departure. With this statement, I do not intend a counter-colonial conquista, nor do I wish to revive the age-old and weary debate traditionally labeled a critique of “reductionism,” and I offer no holistic remedies. Much has been said about constitutive, explanatory, and theoretical reduction in recent years (see Sarkar 1991 for a succinct summary). I do not want to add to this philosophical debate. Let me adopt, instead, a different point of view, less theory-inclined and more practice-centered, of the sciences as an irreducibly multilayered ensemble of epistemic practices.
2. Experimental Systems. The basic concept I envisage as a point of orientation in the hypercomplex network of the modern empirical sciences, and of the life sciences in particular, is that of the “experimental system.” The notion is firmly entrenched in the everyday practice and vernacular of twentieth-century life scientists, especially of biochemists and molecular biologists. Scientists use the term to characterize the scope, as well as the limits and the constraints, of their research activities. Ask a laboratory scientist what he is doing, and he will speak to you about his “system.” Experimental systems constitute integral, locally manageable, functional units of scientific research. It is through them that particular scientific objects - epistemic things in my terminology - gain prominence in a wider field of epistemic cultures and practices. These practices include instruments and inscription devices as well as model organisms to which biological research objects are inextricably tied. Theorems become attached to and detached from them in a rather contingent manner. As William Bechtel and Robert Richardson (1993) demonstrated in Discovering Complexity, strategies of “decomposition” and of “localization” are crucial to the analysis of such systems.
Consequently, on an epistemological level we need, to quote Gaston Bachelard, a “philosophy of epistemological detail,” the counterpart to what he calls the “integral philosophy of the philosophers” (Bachelard 1966, 12, 14). Many philosophers have seen in this situation a deplorable and detrimental, yet intrinsic limitation to empirical knowledge. I argue that this is a mistake. We should recognize that, properly analyzed, the details and particulars of practice, far from imposing limits on our knowledge, are prerequisites for, and provide the very means of, achieving scientific knowledge. In particular, it is in the fabric
of properly “tuned” experimental systems that scientific events materialize. It is in the nature of an event that it cannot be anticipated. Novelties are always the result of spatiotemporal singularities. Experimental systems are precisely the arrangements that allow scientists to create epistemic spatiotemporal singularities. They allow researchers to arrive at unprecedented, surprising results. In this sense, such systems are “more real,” if you will, than ordinary reality. The reality of epistemic things is their resistance, their resilience, their capacity, as “jokers” of practice, to force us to abandon preconceptions and anticipations. To cite Michael Polanyi: “This capacity of a thing to reveal itself in unexpected ways in the future, I attribute to the fact that the thing observed is an aspect of reality, possessing a significance that is not exhausted by our conception of any single aspect of it. To trust that a thing we know is real is, in this sense, to feel that it has the independence and power for manifesting itself in yet unthought of ways in the future” (quoted in Grene 1984, 219). From the perspective of scientific research, this is the best definition of “reality” I have encountered so far.
I have shown elsewhere that productive experimental systems are located at the cutting edge of the decompostition and localization procedure just mentioned (Rheinberger 1992, 1997). They operate most prolifically at the fuzzy boundary between the trivial and the complex. Experimental systems are machines for reducing complexity, but to escape triviality, they must remain connected to the complexity of an “epistemic horizon.” It is the network of surrounding experimental systems that makes each of its elements take on its epistemic value. If ontic complexity has to be reduced in order to make experimental research possible, this very complexity is epistemically retained in the rich context of an experimental landscape, in which new connections and disconnections can happen at any time, and where the boundaries of a scientific object continually fluctuate.
3. The Gene as a Fluctuating Object of Molecular Biology. In this section, I very briefly touch on the changing epistemic and experimental fate of the “gene” as one of the prime epistemic objects of molecular biology. The few remarks that follow are not meant as a systematic assessment of the intricate pathways through which molecular biology has appropriated this object. Nor can I retrace the meandering history of the gene as an object of experimentation in molecular biology. My concern is simply to point out, in a rather loose and associative fashion, some questions that I think will have to be addressed if we wish to understand experimental complexity. We will see that instead of solving the riddle of the gene and rescuing it forever from the deep unknown, molecular biology has managed to redefine its boundaries repeatedly.
Today, biologists continue to reshape this strange epistemic object and to alter it almost beyond recognition.
Molecular biology is a hybrid science combining physics, chemistry, genetics, and the search for biological function at the molecular level. Not surprisingly, it presents itself as conceptually hybrid as well, which is not to say that it has no consistency. The discourse of molecular biology pervades contemporary biology as a whole. I would like to learn more about such hybrid consistencies: how they come about, how they work, and how they evolve.
Let me give a few examples of perspectives on genes both enabled and constrained by experimental systems on these different levels. For a biophysicist working with a crystalline DNA fiber, a gene might be sufficiently characterized by a particular conformation of a DNA double helix. If asked, he or she might define a gene in terms of the atomic coordinates of a nucleic acid. For a biochemist working with isolated DNA in the test tube, genes might be sufficiently defined as stretches of nucleic acids exhibiting certain stereochemical features and sequence recognition patterns. The biochemist can reasonably try to give a macromolecular, DNA-based definition of the gene. For a molecular geneticist, genes might be defined as instructive elements of chromosomes that eventually give rise to defined functional or structural products: transfer RNAs, ribosomal RNAs, enzymes, and proteins serving other purposes. Molecular geneticists certainly will insist on considering issues in terms of replication, transcription, and translation and will require examination of the products of hereditary units when speaking of genes. For evolutionary molecular biologists, genes might be the products of mutating, reshuffling, duplicating, transposing, and rear-ranging bits of DNA within a complex chromosomal environment that has evolved through differential reproduction and selection. Therefore, they will rely on concepts such as transmission, lineage, and history. For developmental biologists, genes might be sufficiently described, on the one hand, as hierarchically ordered switches that, when turned on or off, induce differentiation, and on the other hand, as patches of instructions that are realized in synchrony through the action of these switches. Thus, developmental biologists are likely to refer to the regulatory aspect of genetic circuitry when defining a gene or a larger transcriptional unit such as an operon. We could go on and add more items to the list.
We can ask whether it is necessary or even desirable to have a unified concept of the gene in order to hold all these disciplinary specializations together and to develop them in a coordinated fashion. Obviously, this has not been needed in the half century since molecular biology came into existence. I do not think that it would have helped the development
of the field in appreciable ways; further, I contend that an attempt to do so today would produce nothing more than an exercise in rhetoric. The coherence of molecular biology - which does not exclude Kuhnian incommensurabilities - is not tied into an axiomatic structure or an algorithm; it is embedded in a complex set of experimental systems, each with its genuine epistemic practices, that have evolved over time and that have constrained earlier interpretations as well as allowed new ambiguities to arise. Genes as we now know them are boundary objects par excellence that are crafted, more than by any theory, by the practices and instruments that helped to create the new biology.
4. Conjunctures, Hybrids, Bifurcations, Experimental Cultures. Let me come back to the concerns of epistemology and sort out, if only in preliminary fashion, a few features of such webs of epistemic activities. If experimentation has “a life of its own” with respect to theories (Hacking 1983, 150), experimental systems, as we have glimpsed through the short discussion of the molecular “gene” above, do not live alone for that reason. The number of such systems is enormous, and variation among them is a prerequisite for experimental complexity to come into play. With Peter Coveney and Roger Highfield (1995, 7), we can state that complexity like this rests on “macroscopic collections of such units that are endowed with the potential to evolve in time,” whose “interactions lead to coherent collective phenomena.”
Such interactions can be specified. The collective action of experimental systems may lead to “conjunctures,” meaning the emergence of an extraordinary constellation. The notion should not be confounded with that of an “anomaly” or with that of a “paradigm shift” in the sense of Thomas Kuhn (1962). It designates neither an irritating irregularity within an established and accepted conceptual framework nor the replacement of an encompassing theory by a new one; rather, it points to unforeseen directions opened up within the experimental process. Conjunctures derive from unprecedented events and may lead to major rearrangements and recombinations of given representational spaces in an experimental system. Unprecedented events are about things not sought after - they come as a surprise, but nevertheless do not just happen. They are made to happen through the inner workings of the experimental machinery “for making the future” (Jacob 1988, 9), and yet, they may lead experimenters to completely change the direction of their research activities. Conjunctures can take different forms, and it remains for historical case studies to elaborate these forms in depth.
To illustrate this point, let me give a brief example whose details can be found elsewhere (Rheinberger 1997). In 1953, a first test tube system
of protein synthesis was established. Its dependence on ATP as a supply of biochemical energy led, in 1954, to the characterization of a novel epistemic entity: activated amino acids consisting of a combination of an amino acid, the building block of proteins, and a molecule of ATP lacking two of its phosphates. With that, protein synthesis became detached from the earlier context of oncology and inserted into general biochemistry. Two years later, in pursuit of this biochemical perspective, yet another molecule emerged as an additional intermediate of the protein synthesis reaction chain. It was a small ribonucleic acid to which the activated amino acids became attached before condensing to protein molecules. Today, this new epistemic entity is known as transfer RNA. A major conjuncture followed from this finding. First identified as an intermediate in a biochemical reaction chain, transfer RNA was soon to become an intermediary in genetic information transfer. It bridged the gap between the genes as carriers of genetic specificity, and between the proteins as carriers of biological specificity in terms of cellular function. As a result, the experimental system of in vitro protein synthesis became part of molecular biology.
There is another kind of event deriving from the fuzzy contours of experimental systems: events that produce linkages between independent systems, thus leading to hybrid formations. Interfaces can be created between two or more experimental arrangements. Such coincidences connect particular experimental systems to integrated setups. From a hybridization of different, originally unconnected experimental systems, research arrangements with totally unexpected qualities can result. Things thus are brought together whose articulation, amalgamation, or even blending was not assumed to lie in the “nature” of these things. The history of molecular biology is replete with hybridization events. The fusion, e.g., of Francois Jacob’s bacterial conjugation and phage replication system with Jacques Monod’s system of induced enzyme synthesis led to the emergence of another novel RNA entity, messenger RNA, and to a pathbreaking model of genetic regulation.
A third type of event is complementary to hybridization. It can result in the bifurcation of a particular experimental system and thus lead to offspring systems. Such offspring arrangements tend to form ensembles, or clusters, that yield an experimental space for enlarged scientific communities. Generally speaking, bifurcations of an experimental system occur when it has reached a level of complexity that allows researchers to pursue slightly divergent, but sufficiently different epistemic tracks to enable them to arrive at significantly different results. Typically, clusters of such bifurcated systems remain linked for a while by sharing one or more of their material constituents and so may take advantage of each other’s achievements or of each other’s services. But this must not nec-
essarily remain so. They can become completely disconnected from the maternal system, or integrated into other ensembles.
To provide an example, let us return to transfer RNA. With this molecule, the point was reached where in vitro systems of protein synthesis began to proliferate and to develop in diverging directions. Different research agendas arose. Characterizing the nucleotide sequence of transfer RNA promised insights into the secrets of the genetic code. Probing the interaction of transfer RNA with ribosomes attracted many research groups interested in the molecular mechanism of decoding. And the replacement of rat liver in vitro systems with systems based on E. coli extracts opened the possibility to link the genetics of this organism with detailed studies of its physiology.
Concepts such as conjuncture, hybridization, and bifurcation permit us to envisage ensembles of experimental systems and to articulate their intricate interactions. They permit us to conceive of a structured experimental network of objects and practices that, just as in the case of individual experimental systems, is tinkered and pulled together from different elements. The cohesion of such a reticulum of experimental practices and systems is not conceptual in the first place. It is due to, and reaches exactly as far as, the circulation and the exchange of epistemic entities, model compounds, technical subroutines, and tacit skills throughout the network. Conjunctures, hybridizations, and bifurcations basically describe types of shifts, linkages, and descents through which the dynamics of reorientation, fusion, and proliferation of particular experimental systems is made possible. The consideration of these processes permits us to extend the epistemic analysis from the microdynamics of localized and situated experimental settings to the larger dynamics of cultures of experimentation.
Epistemic things (such as the gene and transfer RNA), experimental systems, ensembles of such systems, and experimental cultures are the conceptual elements with which I try to prepare the ground for a history and epistemology of experimentation that dissolves the traditional distinction between context of justification and context of discovery. Such an epistemology of scientific practice frees the experiment from its subsidiary role in rationalistic accounts of theory development and theory change. My framework, however, does not focus on the social history of scientific institutions and disciplines. It is an attempt to understand the epistemic dynamics of the empirical sciences in terms of the structure of the practices from which these sciences spring and in which they dwell. Experimental cultures are as “patchwork” as the experimental systems they are composed of. But they are held together by a specific kind of glue: material, not only formal, interaction; and practice-centered, not merely theoretical, compatibility.
5. The Patchwork View of Research. With this, we come very near to one of the ideas - and remain very far from others - that Stuart Kauffman develops in his recent book (1995), and which he calls the “patch procedure”:
The basic idea of the patch procedure is simple: take a hard, conflict-laden task in which many parts interact, and divide it into a quilt of nonoverlapping patches. Try to optimize within each patch. As this occurs, the couplings between parts in two patches across patch boundaries will mean that finding a “good” solution in one patch will change the problem to be solved by the part in the adjacent patches. Since changes in each patch will alter the problems confronted by the neighboring patches, and the adaptive moves by those patches in turn will alter the problem faced by yet other patches, the system is just like our model coevolving ecosystems... We are about to see that if the entire conflict-laden task is broken into the properly chosen patches, the coevolving system lies at a phase transition between order and chaos and rapidly finds very good solutions. Patches, in short, may be a fundamental process we have evolved in our social systems, and perhaps elsewhere, to solve very hard problems. (Kauffman 1995, 252-253)
In its application to science, let me call this the “patchwork view of research.” To understand its workings, we have to look for a transition zone that lies somewhere between the rationality of individual actors and the constraints of disciplinary communities. The patches, i.e., the experimental systems, are the subcritical elements of a network that, as a whole, takes on the features of a supracritical process we call science in the making. For the time being, of course, this does not amount to much more than a seductive metaphor. The few remarks I have presented in the central section of my paper on the gene as a boundary object in molecular biology are scarcely more than shreds that may hint at the direction to be taken. We have only started to look at the non-Cartesian, emergent properties of the inextricable web of epistemic practices that unfolds in utterly unforeseeable ways but nevertheless shows a pattern.
How shall we describe this endeavor? There are resources. In his Inaugural Address to the College de France in 1970, Michel Foucault traced a series of notions that will prove helpful for assessing the historical dynamics of scientific practice: “The fundamental notions now imposed upon us are no longer those of consciousness and continuity (with their correlative problems of liberty and causality), nor are they those of sign and structure. They are notions, rather, of events and of series, with the group of notions linked to these [including regularity,
the aleatoric, discontinuity, dependence, and transformation] it is around such an ensemble that this analysis of discourse I am thinking of is articulated, certainly not upon those traditional themes which the philosophers of the past took for ‘living history’” (Foucault 1972b, 230).  Foucault (1972a, esp. Part 4) has characterized his version of such an endeavor as “archeology” and, with respect to the history of science, has spoken of an “archeology of knowledge.” The archeologist digs out the material sediments, the dispositions and depositions in which all theoretical knowledge is embodied and embedded.
6. Conclusion. At the end of his previously-mentioned book, Kauffman ponders: “I wonder if we really understand very much of what we are creating”; and he continues: “All we can do is be locally wise, even though our own best efforts will ultimately create the conditions that lead to our transformations to utterly unforeseeable ways of being” (Kauffman 1995, 298, 303). Local wisdom, at best, is what characterizes the practice of an endeavor that has never ceased to depict itself as an allegedly global undertaking: Science. Instead of searching for universal theories, the order of the day for epistemology is to learn to understand how local wisdoms, entrenched in research “attractors” such as experimental systems, become connected to knowledge patch-works. Fragmentation, far from being deleterious, appears as one of the basic conditions of unprecedented development. Fragmentation, aiming at simplicity, finally creates complexity. The research object called the “gene” is a good historical example for this process. The many dimensions it has acquired in the course of the last century are not the result of alternative, organismic or holistic approaches called up to counteract reductionistic genetics and molecular biology. On the contrary, local experimental sophistication has exploded a coarse and simplistic gene concept from within. Currently, we are witnessing a similar scenario on the level of what, we used to call biological “disciplines”: the boosting of developmental biology through molecular genetics. Understanding the dynamics of these interactions and transformations is what a “philosophy of the epistemological detail” will have to address.
“There is an incompatibility between precision and complexity. As the complexity of a system increases, our ability to make precise and yet non-trivial assertions about its behavior diminishes” (Zadeh 1987, 23). Exploring this epistemological principle of uncertainty in an effort to manage experimental complexity requires less striving at a “Theory
1. The insertion has been omitted from the English translation.
of Everything” (Pickering 1995) than looking for patterns of “moderate compressibility” (Coveney and Highfield 1995, 39).
Bachelard, G. (1966), La Philosophie du Non, 4th ed. Paris: Presses Universitaires de France.
Bechtel, W. and R. C. Richardson (1993), Discovering Complexity: Decomposition and Localization as Strategies in Scientific Research. Princeton: Princeton University Press.
Coveney, P. and R. Highfield (1995), Frontiers of Complexity: The Search for Order in a Chaotic World. London: Faber and Faber.
Foucault, M. (1972a), The Archaeology of Knowledge. Translated by A. M. Sheridan Smith. New York: Pantheon Books.
(1972b), The Discourse on Language.Translated by Rupert Swyer. Published as an appendix to The Archaeology of Knowledge. New York: Pantheon Books.
Grene, M. (1984), The Knower and the Known. Washington, DC: Center for Advanced Research in Phenomenology & University Press of America.
Hacking, I. (1983), Representing and Intervening: Introductory Topics in the Philosophy of Natural Science. Cambridge: Cambridge University Press.
Jacob, F. (1988), The Statue Within. New York: Basic Books.
Kauffman, 5. (1995), At Home in the Universe. Oxford: Oxford University Press.
Kuhn, T. 5. (1962), The Structure of Scientific Revolutions. Chicago: The University of Chicago Press.
Pickering, A. (1995), The Mangle of Practice. Chicago: The University of Chicago Press.
Rheinberger, H.-J. (1992), “Experiment, Difference, and Writing. I. Tracing Protein Synthesis,” and “II. The Laboratory Production of transfer RNA”, Studies in History and Philosophy of Science 23: 305-331, 389-422.
(1993), “Experiment and Orientation. Early Systems of in Vitro Protein Synthesis,” Journal of the History of Biology 26: 443-471.
(1995), “From Microsomes to Ribosomes: “Strategies’ of ‘Representation”, Journal of the History of Biology 28: 49-89.
(1997), Towards History of Epistemic Things: Synthesizing Proteins in the Test Tube. Stanford: Stanford University Press.
Sarkar, 5. (1991), “Reductionism and Functional Explanation in Molecular Biology”, Uroboros 1: 67-94.
Zadeh, L. (1987), “Coping with the Imprecision of the Real World”, in R. R. Yager, S. Ovchinnikov, R. M. Tong, and H. T. Nguyen (eds.), Fuzzy Sets and Applications: Selected Papers by Lofti A. Zadeh. New York: John Wiley, pp. 9-28.