meant to explain how humans with normal human cognitive capacities manage to do modern science. One way, it is suggested, is by constructing distributed cognitive systems that can be operated by humans possessing only the limited cognitive capac- ities they in fact possess. Moreover, Latour himself now seems to agree with this assess- ment. In a 1986 review of Hutchins’s Cognition in the Wild (1995), he explicitly lifts his earlier moratorium claiming that “cognitive explanations . . . have been . . . made thoroughly compatible with the social explanations of science, technology and for- malism devised by my colleagues and myself ” (Latour, 1986a: 62). How this latter statement is to be reconciled with his theory of actants is not clear. Here I would agree with Andy Pickering (1995: 9–20), who is otherwise quite sym- pathetic to Latour’s enterprise, that we should retain the ordinary asymmetrical con- ception of human agents, rejecting both Knorr Cetina’s super-agents and Latour’s actants. Thus, even in a distributed cognitive system, we need not assign such attrib- utes as intention or knowledge to a cognitive system as a whole but only to the human components of the system. In addition to placating common sense, this resolution has the additional virtue that it respects the commitment of historians of science to a narrative form that features scientists as human actors. Laboratories as Evolving Distributed Cognitive Systems Applying the notion of distributed cognition, Nancy Nersessian and associates (2003) have recently been investigating reasoning and representational practices employed in problem-solving in biomedical engineering laboratories. They argue that these lab- oratories are best construed as evolving distributed cognitive systems. The laboratory, they claim, is not simply a physical space but a problem space, the components of which change over time. Cognition is distributed among people and artifacts, and the relationships among the technological artifacts and the researchers in the system evolve. To investigate this evolving cognitive system, they employ both ethnography and historical analysis, using in-depth observation of the lab as well as research into the histories of the experimental devices used in it. They argue that one cannot divorce research from learning in the context of the laboratory, where learning involves build- ing relationships with artifacts. So here we have a prime example of the merger of social, cognitive, and historical analyses built around the notion of distributed cog- nition—and in a technological context. MODELS AND VISUAL REPRESENTATIONS Although mental models have been discussed in the cognitive sciences for a genera- tion, there is still no canonical view of what constitutes a mental model or how mental models function in reasoning. The majority view among cognitive scientists assimi- lates mental models to standard computational models with propositional represen- tations manipulated according to linguistic rules. Here the special feature of mental models is that they involve organized sets of propositions. Work in the cognitive study of science generally follows the minority view that the mental models used in rea- 266 Ronald N. Giere soning about physical systems are iconic. An exemplar of an iconic mental model is a person’s mental image of a familiar room, where “mental image” is understood as highly schematic and not as a detailed “picture in the mind.” Many experiments indi- cate that people can determine features of such a room, such as the number and place- ment of windows, by mentally examining their mental images of that room. While not denying that mental models play a role in the activity of doing science, I would emphasize the role of external models, including three-dimensional physical models (de Chadarevian & Hopwood, 2004), visual models such as sketches, diagrams, graphs, photographs, and computer graphics, but also including abstract models such as a simple harmonic oscillator, an ideal gas, or economic exchanges with perfect infor- mation. External models have the added advantage that they can be considered as components in distributed cognitive systems (Giere, 2006: chapter 5). Combining research in cognitive psychology showing that ordinary concepts exhibit a graded rather than sharply dichotomous structure, together with a model- based understanding of scientific theories developed in the philosophy of science, I (1994, 1999) suggested that scientific theories can be seen as exhibiting a cognitive as well as a logical structure. Thus, the many models generated within any general the- oretical framework may be displayed as exhibiting a “horizontal” graded structure, multiple hierarchies of “vertical” structures, with many detailed models radiating outward from individual generic models. Using examples from the 1960s revolution in geology, I argued that scientists some- times base their judgments of the fit of models to the world directly on visual repre- sentations, particularly those produced by instrumentation (Giere, 1996, 1999). There need be no inference in the form of propositional reasoning. Similarly, David Gooding (1990) found widespread use of visual representations in science. In his detailed study of Faraday’s discovery of electromagnetic induction, he argued that the many diagrams in Faraday’s notebooks are part of the process by which Faraday constructed inter- pretations of his experimental results. Most recently, Gooding (2005) surveyed work on visual representation in science and provided a new theoretical framework, abbre- viated as the PSP schema, for studying the use of such representations. In its standard form, the schema begins with a two-dimensional image depicting a Pattern. The Pattern is “dimensionally enhanced” to create a representation of a three-dimensional Structure, then further enhanced to produce a representation of a four-dimensional Process. In general, there can also be “dimensional reductions” from Process down to Structure and down again to a Pattern. Gooding illustrates use of the scheme with examples from paleobiology, hepatology, geophysics, and electromagnetism (see also Gooding, 2004). JUDGMENT AND REASONING There is a large literature devoted to the experimental study of reasoning by individ- uals, typically undergraduate subjects but sometimes scientists or other technically trained people (Tweney et al., 1981; Gorman, 1992). Here I consider first two lines of Cognitive Studies of Science and Technology 267 research that indicate that reasoning by individuals is strongly influenced by context and only weakly constrained by normative principles. I then describe a recent large comparative study of reasoning strategies employed by individuals in research groups in molecular biology and immunology in the United States, Canada, and Italy. Biases in Individual Reasoning The Selection Task One of the most discussed problems in studies of individual rea- soning is the so-called selection task devised by Peter Wason in the 1960s. In a recent version (Evans, 2002), the subject is presented with four cards turned one side up and told that one side shows either the letter A or some other letter while the other side shows either the number 3 or some other number. The four cards presented have the following sides facing up: A, D, 3, 7. The subject is instructed to select those cards, and only those cards, necessary to determine the truth or falsity of the general propo- sition (“law”) covering just these four cards: If any of these cards has an A on one side, then it has a 3 on the other side. The correct answer is to select the card with the A on front and the card with the 7 on front. If the card with an A on front does not have a 3 on the back, the law is false. Likewise, if the card with a 7 on front has an A on the back, the law is false. The cards with a D or a 3 showing provide no decisive information, since whatever is on the back is compatible with the law in question. On average, over many experiments, only about ten percent of subjects give the right answer. Most subjects correctly choose to turn over the card with an A on front, but then either stop there or choose also to turn over the uninformative card with the 3 on front. Many have drawn the conclusion that natural reasoning does not follow the idea long advocated by Karl Popper (1959) that science proceeds by attempted falsification of general propositions. If one were trying to falsify the stated law, one would insist on turning over the card with the 7 facing up to determine whether or not it has an A on the back. Others have drawn the more general conclusion that, in ordinary cir- cumstances, people exhibit a “confirmation bias,” that is, they look for evidence that agrees with a proposed hypothesis rather than evidence that might falsify it. This leads them to focus on the cards with either an A or a 3 showing, since these symbols figure in the proposed law. A striking result of this line of research is that the results are dramatically different if, rather than being presented in abstract form, the proposed “law” has significant content. For example, suppose the “law” in question concerns the legal age for drink- ing alcoholic beverages, such as: If a person is drinking beer, that person must be over 18 years of age. Now the cards represent drinkers at a bar (or pub) and have their age on one side and their drink, either a soft drink or beer, on the other. Suppose the four cards presented with one side up are: beer, soda, 20, and 16. In this case, on the average, about 75 percent of subjects say correctly that one must turn over both the cards saying beer and age 16. This is correct because only these cards represent possi- ble violators of the law. 268 Ronald N. Giere This contrast is important because it indicates that socially shared conventions (or, in other examples, causal knowledge) are more important for reasoning than logical form. Indeed, Evans (2002: 194) goes so far as to claim that “The fundamental com- putational bias in machine cognition is the inability to contextualize information.” Probability and Representativeness A battery of experiments (Kahneman et al., 1982) demonstrate that even people with some training in probability and statistical infer- ence make probability judgments inconsistent with the normative theory of proba- bility. In a particularly striking experiment, replicated many times, subjects are presented with a general description of a person and then asked to rank probability judgments about that person. Thus, for example, a hypothetical young woman is described as bright, outspoken, and very concerned with issues of discrimination and social justice. Subjects are then asked to rank the probability of various statements about this person, for example, that she is a bank teller or that she is a feminist and a bank teller. Surprisingly, subjects on the average rank the probability of the con- junction, feminist and bank teller, significantly higher than the simple attribution of being a bank teller. This in spite of the law of probability according to which the con- junction of two contingent statements must be lower than that of either conjunct since the individual probabilities must be multiplied. The accepted explanation for this and related effects is that, rather than following the laws of probability, people base probability judgments on a general perception of how representative a particular example is of a general category. Thus, additional detail may increase perceived representativeness even though it necessarily decreases prob- ability. On a contrary note, Gigerenzer (2000) argues that representativeness is gener- ally a useful strategy. It is only in relatively contrived or unusual circumstances where it breaks down. Solomon (2001 and chapter 10 in this volume) discusses the possi- bility that biases in reasoning by individuals are compatible with an instrumentally rational understanding of collective scientific judgment. Comparative Laboratory Studies of Reasoning For over a decade, Kevin Dunbar (2002) and various collaborators have been examin- ing scientific reasoning as it takes place, in vivo, in weekly lab meetings in major mol- ecular biology and immunology labs in the United States, Canada, and Italy. In addition to tape recording meetings and coding conversations for types of reasoning used by scientists, Dunbar and colleagues have conducted interviews and examined lab notes, grant proposals, and the like. Among the major classes of cognitive activ- ity they distinguish are causal reasoning, analogy, and distributed reasoning. Causal Reasoning Dunbar and colleagues found that more than 80 percent of the state- ments made at lab meetings concern mechanisms that might lead from a particular cause to a particular effect. But causal reasoning, they claim, is not a unitary cogni- tive process. Rather, it involves iterations of a variety of processes, including the use of inductive generalization, deductive reasoning, categorization, and analogy. The Cognitive Studies of Science and Technology 269 initiation of a sequence of causal reasoning is often a response to a report of unex- pected results, which constitute 30 to 70 percent of the findings presented at any par- ticular meeting. The first response is to categorize the result as due to some particular type of methodological error, the presumption being that, if the experiment were done correctly, one would get the expected result. Only if the unexpected result continues to show up in improved experiments do the scientists resort to proposing analogies leading to revised models of the phenomena under investigation. Analogy Dunbar et al. found that analogies are a common feature of reasoning in lab- oratory meetings. In one series of observations of sixteen meetings in four laborato- ries, they identified 99 analogies. But not all analogies are of the same type. When the task is to explain an unexpected result, both the source and target of the analogies are typically drawn from the same or a very similar area of research so that the difference between the analogized and the actual situation is relatively superficial. Nevertheless, these relatively mundane analogies are described as “workhorses of the scientific mind” (Dunbar, 2002: 159). When the task switches to devising new models, the differences between the analo- gized and actual situation are more substantial, referring to structural or relational fea- tures of the source and target. Although they found that only about 25 percent of all analogies used were of this more structural variety, over 80 percent of these were used in model construction. Interestingly, analogies of either type rarely find their way into published papers. They mainly serve as a kind of cognitive scaffolding that is discarded once their job is done. Distributed Reasoning A third type of thinking discussed by Dunbar and associates is collective and is most common in what they call the Representational Change Cycle. This typically occurs when an unexpected result won’t go away with minor modifi- cations in the experiment and new or revised models of the system under investiga- tion are required. In these situations they find that many different people contribute parts of the eventual solution through complex interactions subject to both cognitive and social constraints. Here causal reasoning and analogies play a major cognitive role. Culture and Scientific Cognition Richard Nisbett (2003) has recently argued that there are deep differences in the ways Westerners and Asians interact cognitively not only with other people but also with the world. Dunbar argues that one can also see cul- tural differences in the way scientists reason in the laboratory. He compared the rea- soning in lab meetings in American and Italian immunology labs that were of similar size, worked on similar materials, and used similar methods. Members of the labs pub- lished in the same international journals and attended the same international meet- ings. Many of the Italians were trained in American labs. Nevertheless, Dunbar found significant differences in their cognitive styles. Scientists working in American labs used analogies more often than those working in the Italian laboratories. Induction or inductive generalization was also used in the 270 Ronald N. Giere American labs more often than in the Italian labs, where the predominant mode of reasoning was deductive. In American labs, deductive reasoning was used only to make predictions about the results of potential experiments. There is some evidence that these differences in cognitive strategies among scientists in the laboratory reflect similar differences in the cultures at large. Thus, it seems that no single cognitive process characterizes modern science and research in a given field can be done using different mixes of cognitive processes. Which mix predominates in a given laboratory may depend as much on the sur- rounding culture as on the subject matter under investigation. CONCEPTUAL CHANGE As noted at the beginning of this essay, following the publication of Thomas Kuhn’s Structure of Scientific Revolutions (1962), conceptual change became a major topic of concern among historians, philosophers, and psychologists of science. When the cog- nitive revolution came along a decade later, tools being developed in the cognitive sciences came to be applied to improve our understanding of conceptual change in science. I will discuss just one ongoing program of this sort, Nancy Nersessian’s Model- Based Reasoning. Following the general strategy in cognitive studies of science, Nersessian’s goal is to explain the process of conceptual change in science in terms of general cognitive mechanisms and strategies used in other areas of life. Her overall framework is pro- vided by a tradition emphasizing the role of mental models in reasoning. Within this framework she focuses on three processes: analogy, visual representation, and simu- lation or “thought experimenting,” which together provide sufficient means for effect- ing conceptual change (Nersessian, 2002a). The Mental Modeling Framework Extending standard notions of mental models, Nersessian claims that some models in the sciences are generic. They abstract from many features of real systems for which models are sought. An example would be Newton’s generic model for gravitation near a large body in which the main constraint is that the force on another body varies as the inverse square of its distance from the larger body. This abstraction allows one eventually to think of the motion of a cannon ball and that of the Moon as instances of the same generic model. Analogical Modeling A considerable body of cognitive science literature focuses on metaphor and analogy (Lakoff, 1987; Gentner et al., 2001). The relationship between the source domain and the target domain is regarded as productive when it preserves fundamental structural relationships, including causal relationships. Ners- essian suggests that the source domain contributes to the model building process by providing additional constraints on the construction of generic models of the target domain. The use of analogy in everyday reasoning seems to differ from its use in Cognitive Studies of Science and Technology 271 science, where finding a fruitful source domain may be a major part of the problem when constructing new generic models. It helps to know what a good analogy should be like, but there seems to remain a good bit of historical contingency in finding one. Visual Modeling The importance of diagrams and pictures in the process of doing science has long been a focus of attention in the social study of science (Lynch and Woolgar, 1990). For Nersessian, these are visual models, and she emphasizes the rela- tionship between visual models and mental models. Visual models facilitate the process of developing analogies and constructing new generic models. Nersessian also recognizes the importance of visual models as external representations and appreci- ates the idea that they function as elements in a distributed cognitive system that includes other researchers. Indeed, she notes that visual models, like Latour’s immutable mobiles, provide a major means for transporting models from one person to another and even across disciplines. This latter point seems now accepted wisdom in STS. Simulative Modeling We tend to think of models, especially visual models, as being rel- atively static, but this is a mistake. Many models, like models in mechanics, are intrin- sically dynamic. Others can be made dynamic by being imagined in an experimental setting. Until recently, thought experiments were the best-known example of simula- tive modeling. Now computer simulations are commonplace. However, the cognitive function is the same. Imagining or calculating the temporal behavior of a model of a dynamic system can reveal important constraints built into the model and suggest how the constraints might be modified to model different behavior. Thought experi- ments can also reveal features of analogies. A famous case is Galileo’s analogy based on the thought experiment of dropping a weight from the mast of a moving ship. Realizing that the weight will fall to the base of the mast provides a way of under- standing why an object dropped near the surface of a spinning earth nevertheless falls straight down. Nersessian brings all these elements together in what she calls a “cognitive- historical analysis” of Maxwell’s development of electrodynamics following Faraday’s and Thompson’s work on interactions between electricity and magnetism (Nersessian, 2002b). This analysis shows how visual representations of simulative physical models were used in the derivation of mathematical representations (see also Gooding & Addis, 1999). COGNITIVE STUDIES OF TECHNOLOGY In history, philosophy, and sociology, the study of technology has lagged behind the study of science. The history of technology is now well established, but both the phi- losophy and sociology of technology have only recently moved into the mainstream, and in both cases there have been attempts to apply to the study of technology 272 Ronald N. Giere approaches first established in the study of science. This is apparent in the work pre- sented in The Nature of Technological Knowledge: Are Models of Scientific Change Relevant? (R. Laudan, 1984) and The Social Construction of Technological Systems: New Directions in the Sociology and History of Technology (Bijker et al., 1987). The closest thing to a comparable volume in the cognitive study of technology, Scientific and Technological Thinking (Gorman et al., 2005b), has appeared only very recently, and even here, only five of fourteen chapters focus exclusively on technology rather than science. An obvious supplement would be the earlier collaboration between Gorman and the his- torian of technology, Bernard Carlson, and others, on the invention of the telephone (Gorman & Carlson, 1990; Gorman et al., 1993). Gary Bradshaw’s “What’s So Hard About Rocket Science? Secrets the Rocket Boys Knew” (2005) can be read as a sequel to his paper in the Minnesota Studies volume on the Wright brothers’ successful design of an airplane (Bradshaw, 1992). Bradshaw, who was initially a member of the Simon group working on scientific discovery, begins with Simon’s notion of a “search-space.” Invention is then understood as a search through a “design space” of possible designs. Success in invention turns out to be a matter of devising heuristics for efficient search of the design space. In the case of the teenaged “rocket boys” working on a prize-winning science project following Sputnik, launch-testing every combination of attempted solutions to a dozen different design features would have required roughly two million tests. Yet the boys achieved success after only twenty-five launches. Bradshaw explains both how they did it and how and why their strategy differed from that of the Wright brothers, thus revealing that there is no universal solution to the design problem as he conceives it. Contextual factors matter. Michael Gorman’s (2005a) programmatic contribution, “Levels of Expertise and Trading Zones: Combining Cognitive and Social Approaches to Technology Studies,” sketches a framework for a multidisciplinary study of science and technology. He begins with Collins and Evans’s (2002) proposal that STS focus on the study of expe- rience and expertise (SEE), which, he suggests, connects with cognitive studies of problem solving by novices and experts. Collins and Evans distinguished three levels of shared experience when practitioners from several disciplines, or experts and lay people, are involved in a technological project: (1) they have no shared experience, (2) there is interaction among participants, and (3) participants contribute to devel- opments in each other’s disciplines. Gorman invokes the idea of “trading zones” to characterize these relationships, distinguishing three types of relationships within a trading zone: (1) control by one elite, (2) rough parity among participants, and (3) the sharing of mental models. Finally, he characterizes the nature of communication among participants as being (1) orders given by an elite, or (2) the development of a creole language, or (3) the development of shared meanings. He clearly thinks it desir- able to achieve state 3, with participants sharing meanings and mental models and contributing to each other’s disciplines. Whether intended reflexively or not, this would be a good state for multidisciplinary studies in STS itself, particularly ones involving both cognitive and social approaches. Cognitive Studies of Science and Technology 273 CONCLUSION Looking to the future, my hope is that when the time comes for the next edition of a Handbook of Science and Technology Studies, cognitive and social approaches will be sufficiently integrated that a separate article on cognitive studies of science and tech- nology will not be required. Notes I would like to thank Olga Amsterdamska, Nancy Nersessian, and three anonymous reviewers of an earlier draft of this article for many helpful suggestions. 1. For other recent introductions, see Carruthers et al. (2002); Gorman et al. (2005); Nersessian (2005); and Solomon (chapter 10 in this volume). 2. For a philosophical introduction to this understanding of cognition, see Churchland (1989, 1996). References Bijker, W., T. Pinch, & T. Hughes (eds) (1987) The Social Construction of Technological Systems: New Direc- tions in the Sociology and History of Technology (Cambridge, MA: MIT Press). Bradshaw, Gary (2005) “What’s So Hard about Rocket Science? Secrets the Rocket Boys Knew,” in Michael E. Gorman, Ryan Tweney, David Gooding, & Alexandra Kincannon (eds), Scientific and Technological Thinking (Mahwah, NJ: Lawrence Erlbaum): 259–76. Bradshaw, Gary (1992) “The Airplane and the Logic of Invention,” in R. N. Giere (ed), Cognitive Models of Science, Minnesota Studies in the Philosophy of Science, vol. XV (Minneapolis: University of Minnesota Press): 239–50. Carey, Susan (1985) Conceptual Change in Childhood (Cambridge, MA: MIT Press). Carruthers, Peter, Stephen Stitch, & Michael Siegal (eds) (2002) The Cognitive Basis of Science (Cambridge: Cambridge University Press). Chi, Michelene T. H. (1992) “Conceptual Change Within and Across Ontological Categories: Examples from Learning and Discovery in Science,” in R. N. Giere (ed), Cognitive Models of Science, Min- nesota Studies in the Philosophy of Science, vol. XV (Minneapolis: University of Minnesota Press): 129–86. Churchland, Paul M. (1989) A Neurocomputational Perspective: The Nature of Mind and the Structure of Science (Cambridge, MA: MIT Press). Churchland, Paul M. (1996) The Engine of Reason, The Seat of the Soul: A Philosophical Journey into the Brain (Cambridge, MA: MIT Press). Clark, Andy (1997) Being There: Putting Brain, Body, and World Together Again (Cambridge, MA: MIT Press). Collins, H. M. & R. Evans (2002) “The Third Wave of Science Studies,” Social Studies of Science 32: 235–96. Darden, Lindley (1991) Theory Change in Science: Strategies from Mendelian Genetics (New York: Oxford University Press). de Chadarevian, Soraya & Nick Hopwood (2004) Models: The Third Dimension of Science (Stanford, CA: Stanford University Press). 274 Ronald N. Giere [...]... M & Robert Evans (2002) The Third Wave of Science Studies: Studies of Expertise and Experience,” Social Studies of Science 32(2): 2 35 96 Doing, Park (2004) “Lab Hands and the Scarlet ‘O’: Epistemic Politics and (Scientific) Labor,” Social Studies of Science 34(3): 299–323 Epstein, Steven (1996) Impure Science: AIDS Activism and the Politics of Knowledge (Berkeley: University of California Press) Fox... framing of the project According to Lynch, any claims about the relation between the endurance of the factual products of the laboratory and the practice at the lab are not actually part of his project At the end of his ethnography, he disclaims specifically that “whether agreements in shop talk achieve an extended relevance by being presupposed in the further talk and conduct of members or whether they... Theoretical Practice and the Theoreticians Regress,” Social Studies of Science 30(1): 5 40 Knorr Cetina, Karin (1981) The Manufacture of Knowledge: An Essay on the Constructivist and Contextual Nature of Science (Oxford: Pergamon Press) Knorr Cetina, Karin (19 95) “Laboratory Studies: The Cultural Approach to the Study of Science, ” in Sheila Jasanoff, Gerald Merkle, James Petersen, & Trevor Pinch, Handbook. .. of Science 35( 4): 58 1–621 Schutz, Alfred (1972) The Phenomenology of the Social World (London: Heinemann) Sims, Ben (20 05) “Safe Science: Material and Social Order in Laboratory Work,” Social Studies of Science 35( 3): 333–66 Traweek, Sharon (1988) Beamtimes and Lifetimes: The World of High Energy Physicists (Cambridge, MA: Harvard University Press) Winch, Peter (1 958 ) The Idea of a Social Science and. .. is, in fact, the compound (in somewhat shorthand) Pyro-GluHis-Pro-NH2 As Latour and Woolgar pursue their analysis of the discovery of the nature of TRF(H), they never lose sight, or let us lose sight, of their antidemarcationist mission, stating and restating it many times, and the field of STS has ever since referred to these statements of their accomplishment as foundational pillars of the discipline... details of laboratory activity The care and love that they had for their subjects was evident and compelling Together, they formed a corpus of new intellectual work with provocative and profound implications for both the project of intellectual inquiry and also the essence of political citizenship Catching the wave of excitement growing around these projects, energetic scholars then built upon the implications... Following the “social life of images,” SIV includes the study of both imaging practice and the performance of scientific imagery with particular attention to its visual power and persuasiveness Scientific images and visualizations are exceptionally persuasive because they partake in the objective authority of science and technology, and they rely on what is regarded as immediate form of visual apprehension and. .. the sciences and technologies, bureaucracies, and classification systems (cf Rajchman, 1991; Davidson, [1996]1999; Hacking, 1999; Rose, 1999) 300 Regula Valérie Burri and Joseph Dumit The practice turn in STS taught us to attend to the work of science and technology in terms of both the processes of production and the resulting products The challenge of SIV is now to incorporate together the work of science. .. embedded Lynch and Woolgar’s collection is a starting point for studying the cultural embeddedness of the practices of the making and handling of visual representations and of the shaping, distributing, applying, and embodying of scientific visual knowledge If seeing is so often believing, SIV must demonstrate how the making and using of images come together with seeing and believing in practices of scientific... to the project of implicating local scientific practice in the products of that practice They both privilege something outside of the life-world of laboratory practice to explain the endurance of a particular technical fact While each may say that the problem with the other is that they unduly privilege (respectively) the natural or the social in their explanation, the important point to understand . Studies of Science and Technology 273 CONCLUSION Looking to the future, my hope is that when the time comes for the next edition of a Handbook of Science and Technology Studies, cognitive and social. philosophy, and sociology, the study of technology has lagged behind the study of science. The history of technology is now well established, but both the phi- losophy and sociology of technology have only. the field of Science and Technology Studies over the next three decades. Refer- ring to the early lab studies as foundational pillars of a new discipline, these scholars analyzed episodes of science