Replication and the Experimental Ethnography of Science

31

- -31

Replication and the Experimental Ethnography of Science

By

Ryan D. Tweney

Department of Psychology

Bowling Green State University

Bowling Green, OH 43403

USA

419/372-2301

419/372-6013 (fax)

[In press, Journal of Cognition and Culture]

Replication and the Experimental Ethnography of Science

Ryan D. Tweney[*]

ABSTRACT

The present paper attempts to define an experimental ethnography as an approach to the understanding of scientific thinking. Such an ethnography relies upon the replication of contemporary and historical scientific practices as a means of capturing the cultural and cognitive meanings of the practices in question. The approach is contrasted to the typical kind of laboratory experiment in psychology, and it is argued that replications of scientific practices can reveal dimensions of the microstructure of science and of its context that otherwise may remain invisible. An extended example is presented, based upon replications of the experimental procedures used by Michael Faraday (1791-1867) in 1856 during his examination of the optical properties of gold.

Introduction

In the popular imagination, the image of science is replete with material objects; bubbling flasks, microscopes, sparking coils and the like, all presided over by a white-coated figure with an expression of quiet confidence, passionate certainty, or – sometimes – sheer madness. Indeed, with some of the facial expressions of scientists being a possible exception, the lab coat does have its place in science, flasks and microscopes are still in use, and the sparking coils, while anachronistic now, once did signal the forefront of electrical research. Clearly, science occurs in a realm of materiality, a realm that has recently produced a huge outpouring of scholarship by scholars. Clearly also, this materiality provides an important dimension of inquiry for a cognitive anthropology of science.

The present paper is an attempt to define how an experimental approach to a cognitive ethnography can contribute to the understanding of scientific practices. By experimental ethnography, I mean the attempt to understand scientific practices, both contemporary and historical, using procedures that replicate the practices under study, in an effort to more fully characterize their nature. Such methods are ethnographic insofar as they are meant to capture the richness and cultural meaning of the practices in question, and they are experimental in that the agency of the inquirer is used to directly create, manipulate, and participate in the epistemic practices under examination.

There are two broad categories of such replication that I will consider in this paper. The first stems primarily from cognitive psychologists and involves the attempt to isolate the essential processes of scientific practice. The second grows out of recent work on “cognitive history” and derives from earlier work on historical issues.

The goal in most laboratory research in psychology has of course been an abstractive one, that is, one in which general laws of cognition and behavior are to be abstracted from the particularities of actual behavior, usually the aggregated performance of statistically large numbers of individual “subjects.” The historical emergence of the “subject” as a faceless “object” of psychological investigation has been documented by Danziger (1990), who argued that the clients of American psychology in the first half of the 20th century, primarily bureaucratic elites in education, the corporate world, and the military, were not interested in the inner workings or experiences of individuals and instead wanted the ability to predict and control the behavior of large numbers of people. As a result, the goals of psychological science shifted in the direction of seeking “nomothetic” laws instead of “idiographic” understandings. This aggregative approach had its critics (e.g., Allport 1940), but nonetheless, by mid-century, the dominance of an aggregative approach within psychology was overwhelming. Coupled with the use of statistical analysis (in part as a legitimating procedure for psychology’s “scientific” status, Tweney 2003), the reductive character of experimental psychological method opened a large gap between psychology and the methods of ethnography. By the end of the century, the contrastive “normative” rules of research given by Fetterman (1998) for anthropologists, versus those outlined by Rosenthal & Rosnow (1991) for psychologists, reveal almost no common ground. In the end, most anthropologists would consider the laboratory setting of psychological tasks to be far too artificial, and hence useless as adjuncts to the methods of ethnography, while most psychologists consider the methods of ethnography to be too limited as approaches to nomothetic laws and as little better than anecdotal in character.

In an early signal that this division was perhaps too sharply drawn, Herbert Simon (1967/1996) used the metaphor of an ant crossing a sandy beach. The ant is a relatively simple behavioral system, but the geometry of its path across a beach is apparently very complex. The complexity, however, resides in the environment, not the ant. Simon argued that a similar consideration applied to human problem solving; the apparent complexity is in the symbolic environment of thought not the thought itself which follows, according to Simon, a relatively simple set of heuristic principles. For Simon, broadly general principles of problem solving could be used to untangle the complex environmental manifestations of the problem solving of a single individual. His approach was strikingly successful in such domains as chess playing, and its reach has even been extended to include claims about scientific discovery (e.g., Klahr 2000; Langley, Simon, Bradshaw, & Zytkow 1987; Kulkarni & Simon 1990). In contrast to most psychological research, Simon and his colleagues tended to downplay the aggregative methods of experimental psychology in favor of computer simulation, often based upon single-subject protocols of great complexity.

Simon’s tradition of research has led others to a broader range of methods that begin to resemble the ethnographic approaches that I consider in this paper. Thus, Kevin Dunbar (1999) used laboratory analogs of real-world scientific experiments in genetics as a means of refining the questions used in his in vivo studies of four molecular biology laboratories. These latter were based upon extensive video recorded observations of lab meetings and interviews held over an extended period. Some such studies within cognitive science have converged in scope and richness with efforts in which computer models of scientific thinking have been based upon instances of “real science” (often elaborated in a historical context), for example the studies of Darden (1992) on the processes of theory change in modern genetics, or the simulations by Kulkarni & Simon (1990) of the discovery of the ornithine cycle by Hans Krebs, a simulation of experimental practice based upon a historical analysis of Krebs’s diaries (Holmes 1980; 2004).

In my own case, the work of my colleagues and I initially used laboratory-based “model systems” and an aggregative approach as methods for the understanding of scientific thinking (Mynatt, Doherty, & Tweney 1978; Tweney, Doherty, & Mynatt 1981). Having found, in these studies, evidence for several heuristics involving the assessment of evidence, we sought to confirm our findings in real-world contexts, an attempt that eventually led to my studies of Faraday’s practices (described later in this paper). Our first efforts were conceived under the assumption that the simpler environment of laboratory studies of science would prove superior to a direct immersion in the messy world of “real science.” Thus, my first efforts to understand Faraday (Tweney 1985; Tweney & Hoffner 1987) were essentially hypothetico-deductive in character, in the sense that I was trying to test hypotheses derived from our laboratory work. Only later did I realize that the messiness of Faraday (to take the example closest to my gaze at the time) was only apparent. It appeared so only to one whose hypotheses-laden vision was constricted to selected aspects of Faraday’s work. In fact, broader approaches were necessary.

In the present paper, I adopt a perspective that emerged from this experience, to wit, that everything in the record left by Faraday – his publications, his diary, and now even his specimens – is potentially relevant. The task of the scholar is to understand how to interpret this record in the light of the best available theory, whether that theory derives from cognitive science, from sociology, or from anthropology, and the methods must be at least in part ethnographic in character.

The second broad category of research that I consider is based upon a historical-cognitive approach within the history of science and cognitive science. Nancy Nersessian (1995) introduced the term “cognitive history” to describe this body of work, characterizing it as an attempt to understand the processes by which “vague speculations get articulated into scientific understandings, are communicated to other scientists, and come to replace existing representations of a domain” (p. 194). Here, the goal has been to use insights derived from cognitive science to understand historical cases of scientific thinking and practice, as well as to import findings from the history of science into cognitive science itself. For example, Nersessian (1999) used Maxwell’s development of his field theory as a source problem for claims within cognitive science about the generative role of analogy in scientific thinking. For some in this tradition, the replication of specific historically relevant cases has played an important role. For example, Gooding (1989; 1990) was able to understand some aspects of Faraday’s 1821 experiments on the rotation of a current-carrying wire in a magnetic field by replicating the procedures used by Faraday. In the process, he discovered that the apparently straightforward discovery recorded in Faraday’s notebooks of circular motion (the result of transverse forces rather than radial forces) actually must have involved a long series of interactions between “eye,” “hand,” and “mind.” Only through replication was Gooding able to reconstruct a path by which a seemingly chaotic phenomenon was gradually ordered and shaped by the experimenter’s activity (Gooding & Addis 1999). In Gooding’s words, “Empirical results are never completely independent of the practices that produce them. Facts are practice-laden as well as theory-laden.” (1989, p. 64). Such examples of the implicitly ethnographic use of replication will constitute many of my examples in the remainder of this paper.

Nersessian (1995) argued that the relationship between the disciplines of cognitive science and the history of science was reciprocal; the findings and methods of cognitive science had a great deal to offer history of science, but so also did history of science have much value for the cognitive sciences themselves. Thus, one goal of the present paper is to expand upon the examples of the second kind, to show that an experimental ethnographic approach can draw upon the resources of historical case studies of science in ways that enlarge the scope of current work on cognition itself.

Both kinds of research in cognitive history, the cognitive scientific research (like Simon’s), and the cognitively minded historical work (like Gooding’s and Nersessian’s) suggest a need to provoke discussion about the role of the active replication of scientific experiments and scientific observations as a means of opening a window on otherwise invisible processes of science. The intent is to recover the everyday practices of science from the mostly textual representations of those practices found in published papers and books, in diaries, and in lab notebooks. I argue further that, like an archeologist seeking the limits of pyramid-building technology by “trying my hand at it,” replication of procedures is essential to understanding what can and cannot be happening prior to the instantiation of scientific thought in text. Thus, like research notebooks, the intent is to use replication as a means to open “particularly powerful avenues of access to the microhistorical processes of scientific change” (Holmes, Renn, & Rheinberger 2003, p. xii).

I begin with a case history of experimental ethnography, presenting some of my recent work in which replication was used to uncover some of the experimental and theoretical aspects of Michael Faraday’s research on the optical properties of gold. The project has historical implications for the understanding of Faraday’s research, and implications for cognitive approaches to the understanding of science in general. As a contribution to the history of science, the case study documents an unusually complete record left by a great scientist, one that forces certain reinterpretations of a portion of his work. These reinterpretations are aided by the replications. As a contribution to cognitive science, the case study illustrates how scientific discovery is imbedded in the concrete practices of a set of artifacts, requiring attention to the entire dynamic cognitive system involved in scientific discovery -- the scientist, in a laboratory, making and using artifacts. Here, replications contribute to the understanding of the dynamics of cognition and action.

I. Understanding Scientific Artifacts using Experimental Ethnography.

In 1856, Michael Faraday (1791-1867) carried out an extensive program of research to explore the properties of thin films of metallic gold (Faraday 1857; see also James 1985, for an overview of Faraday’s research on optical phenomena). These films had long been of interest to him because they possess the peculiar property of appearing gold in color by reflected light, but green by transmitted light. Faraday therefore hoped that gold could serve as a model for the general interaction of light and matter. Encouraged perhaps by his earlier finding that magnetic fields could affect a beam of polarised light passing through a highly refractive substance, Faraday was hoping for even more general phenomena by which the integration of field theory and chemical theories of matter could be integrated (Faraday 1839-55). In the course of the research, Faraday discovered the first metallic colloids and what is now known as the “Faraday Tyndall Effect” (actually a case of Rayleigh scattering of light). Thus, although his major theoretical goals were not realised, the investigation resulted in two important discoveries.

In attempting to construct cognitive psychological models of scientific thinking, Michael Faraday’s extensive diaries (Martin 1932-36) had been an important resource for my own investigations, as they have been for others (for examples, see Anderson 1994; Fisher 2001; Gooding 1981, 1990; Steinle 1996; 2003; Tweney 1985, 1992). The stretch of diary covering his work on gold, however, had long been puzzling, both to myself and to others. Thus, Faraday’s principal modern biographer, Pearce Williams (1965), suggested that Faraday was beginning to lose his grip in 1856, and that signs of ageing were apparent. Indeed, a reading of his diary for this period made such a claim plausible, in that there is a seeming aimlessness about much of the record. In fact, however, a recent discovery has cast a different light on Faraday’s efforts during 1856.