In the Sociology of Scientific Knowledge (SSK), most sociologists think that science is more or less socially constructed. It seems to me, however, that there are significant difference in the way they use the word "social". The purpose of this paper is to illuminate the ambiguity by introducing a distinction between two senses of 'social construction': social construction as a social conditioning process and social construction as a consensus formation process. There are two reasons why I introduce this specific distinction. First, this distinction is useful to classify typical SSK research. The classification has a further implication for effective sociological debates. Second, the distinction has a heuristic value in certain kind of sociological research. The distinction will provide us a conceptual scheme to think about the paradoxical nature of science, namely what scientists do is deeply determined by society and still science has its own life independent of society.
2. What is social construction? -- Sismondo's review
Besides the distinction I am going to introduce, the term 'social construction' has been used many different ways by sociologists of science. Sismondo's review article "Some social constructions" provides a good overview (Sismondo 1993). The current usage of 'social construction' was introduced by Berger and Luckmann in The Social Construction of Reality (1966), but they were not particularly interested in science. The introduction of the word into sociology of science was during the 1970s, and the earliest example of the use of the word Sismondo could find was by Mendelsohn, Weingart and Whitley (1977). In his contribution to this book, for example, Mendelsohn summarizes his enterprise as follows: "what I have tried to illustrate in this section is the manner in which the forces active in the social order played a critical role (though not the only role) in the establishment of the scientific way of knowing -- the epistemology of modern science." (Mendelsohn et al 1977, 19. cited in Sismondo 1993, 526).
After the introduction to the sociology of science, 'social construction' has been used in a variety of meanings. Sismondo distinguishes four distinct meanings of 'construction': (1) construction of institutions including knowledge and facts; (2) construction of scientific theories; (3) construction of artifacts in laboratories; (4) construction of the object of representation (Sismondo 1993, 516). As is easily seen, this classification is based upon what is constructed. According to Sismondo, Knorr-Cetina (1981) uses the word in the senses (1)~ (3), and Latour and Woolgar ( 1986) use the word in all of these senses (Sismondo 1993, 527-536).
Sismondo also points out a difference between 'mild' constructivism and more radical variations (Sismondo 1993, 526). Mild constructivism, as is shown in the above quotation from Mendelsohn, admits that the social factor is just one of the factors affecting science. On the other hand, many other sociologists use the word 'construction' in a stronger sense. Pinch and Bijker provide a good example: "Within such a program [Bloor's "strong programme"] all knowledge and all knowledge claims are to be treated as being socially constructed; that is, explanations for the genesis, acceptance, and rejection of knowledge claims are sought in the domain of the social world rather than in the natural world" (Pinch and Bijker 1987, 18). In this sense, "scientific knowledge can be, and indeed has been, shown to be thoroughly socially constituted" (Pinch and Bijker 1987, 18-19).
Sismondo's study reveals the interesting and different ways the term 'social construction' is used. Nevertheless, there seems to be another difference among social constructivists which is overlooked in the study. This is the distinction in the way the word 'social' is used, which I discuss in the following sections.
3. Two kinds of social construction
To illustrate the distinction I have in mind, the following two quotations may be a good starting point:
"One way of grounding our reasoning behavior in society is to study the way in which it is harnessed to particular social interests. The 'interest model' has been shown to work convincingly and in detail in a large number of cases" (Bloor 1981, 210)
"Both approaches are social in that they consider the objects of knowledge as the outcomes of processes which invariably involve more than one individual, and which normally involve individuals at variance with one another in relevant respects" (Knorr-Cetina 1981, 117)
Both Bloor and Knorr-Cetina think that science is essentially social, but the reason why they think so is remarkably different. On the one hand, Bloor thinks that science is social because our "reasoning behavior" is associated with extra-scientific social interests. On the other hand, in Knorr-Cetina's view, science is social simply because it involves "more than one individual". As my review in the following sections shows, these views represents two typical ways sociologists argue for social construction of scientific knowledge.
I call the two views 'social conditioning' and 'consensus formation'. The following is a rough description of them:
social conditioning -- Society, as a causal factor in scientific practice , constructs scientific knowledge . E.g. Society shapes each scientist's world view, motivation, material environment etc.
consensus formation -- The scientific community, as a small society with its own structure, norms and decision processes, creates scientific facts as a result of consensus.
Before I move on to detailed analyses of literature using these notions, a minimal clarification on the distinction is in order. In social conditioning, society appears as a causal factor to regulate and to guide scientists' pursuit of their own research project. I do not intend to deny the free will of scientists by the word "conditioning", but apparently in SSK literature society is supposed to be quite strong. I also use the word "conditioning" in quite a broad sense, as my examples show below.
On the other hand, in consensus formation, the scientific community itself is conceived as a small society. In this small society, scientists have to negotiate to evaluate each other's work. Of course extra-scientific interests can come into such negotiations, but this is not a necessary factor when we use the word 'social' in this sense. Moreover, the scientific community is usually more than a causal factor in scientific practice. It is a precondition of the very existence of science.
The distinction is not totally my invention. For example, as I discuss later, Cole's criticism of social constructivists concentrates on two major points, namely "constructivist problems in accounting for consensus" and "constructivist problems in demonstrating causality" (Cole 1992, 53-81). These two problems seem to roughly correspond to my distinction between social conditioning and consensus formation. Sociologists themselves sometimes refer to similar distinctions to clarify differences among different schools (e.g. Knorr-Cetina 1981, 116; Latour and Woolgar 1986, 152).
Finally, I uses the terms "social conditioning" and "consensus formation" to refer to both social processes as subject matters of research, and research programs concentrating on these processes. I think that in most cases the sense in which I use the terms is obvious by the context.
4. Social construction as social conditioning
4-1. Bloor's "strong programme"
One of the clearest statements of this kind of approach can be found in Bloor's Knowledge and Social Imagery (Bloor  1991). At the beginning of this book, Bloor proposes "the strong programme" of sociology of scientific knowledge (5-8). The strong programme consists of four tenets. First, sociology of scientific knowledge should be "causal", that is, "concerned with the conditions which bring about belief or states of knowledge. Naturally there will be other types of causes apart from social ones which will cooperate in bringing about belief" (7). Second, sociologists should be impartial with respect to truth and falsity, rationality and irrationality etc., in the sense that both sides require explanation. Third, the explanation should be symmetrical, namely "the same types of cause would explain, say, true and false beliefs" (ibid.). Finally, the explanation should be reflexive, namely it has to apply to sociology itself.
The first and third tenets are especially important to understand his position. Scientific knowledge is an object of sociological research because it is caused by social factors. As the above quotation shows, Bloor admits the existence of other kinds of causal factors (he emphasizes this point in the afterword to the second edition of the book; Bloor 1991, 165-166). In this sense, he is a 'mild constructivist' in Sismondo's classification. But his overall position is not mild at all, because of the requirement of symmetry. By this requirement, Bloor claims that social factors are not biases which distort scientific reasoning. Rather, social factors are an essential part of (good and bad) scientific reasoning. Anyway, this is a typical example of constructivism in the sense of social conditioning.
Bloor does not tell us exactly what kind of social factors he has in mind. His paradigm case is social interests, but he seems to talk about other kinds of social causes also (e.g. Bloor  1991, 129). Many sociologists have studied the social conditioning process in science, and we can discern several different kinds of social factors from their works. I will examine some of the representative works in this field in the following subsections.
4-2. Forman's study on Weimar culture and quantum mechanics
Forman's famous study on the Weimar culture and quantum mechanics (1971) claims that the intellectual atmosphere of the Weimar culture influenced the anti-deterministic interpretation (so-called Copenhagen interpretation) of quantum mechanics. In his 1971 paper, Forman's argument is divided into three parts. In the first section, Forman establishes that the intellectual milieu in Germany after World War I (the so-called Weimar culture) was anti-scientific (8-37). Spengler's Decline of the West was a representative of this sentiment. Especially causality and determinism in science are attacked. Forman also points out that German physicists and mathematicians were well aware of this cultural environment (8-15). In the second section, Forman argues that these scientists started to adapt to the cultural environment at an ideological level (38-63). For example, Wilhelm Wien converted from positivism to Lebensphilosophie, and Forman also reports other physicists' capitulation to Spenglerism. In the third section, Forman describes physicists' conversion to an acausal interpretation of physical laws (63-108). Forman reveals the remarkable fact that physicists' conversion at this level started before they started to get evidence for the acausality of quantum mechanics. This fact suggests that their conversion a physical level was caused by their conversion at an ideological level.
Forman clearly states that his work is a causal analysis (3). He declares: "it seems to me that the historian cannot rest content with vague and equivocal expressions like 'prepared the intellectual climate for,' or 'prepared, so to speak, the philosophical background for,' but must insist upon a causal analysis, showing the circumstances under which, and the interactions through which, scientific men are swept up by intellectual currents" (ibid.). His later work (1979) is an attempt to strengthen the power of the causal analysis. In this work he compares British culture and German culture at that time, and concludes that British empiricist culture prevented a majority of British physicists from an acausal interpretation of physical laws. He consciously treats the two cultures as independent variables, and physicists interpretation of physical laws as a dependent variable.
Forman also clarifies that this is not a psychological analysis, but a sociological analysis, "treating present mental posture as socially determined response to the immediate intellectual environment and current experiences" (Forman 1971, 3). Here is an answer to the question about what kind of social factors influence scientific knowledge. Society affects physicists' choice by determining their mental posture.
4-3. Collins' "experimenter's regress"
Another representative work in this category is Collins' study on the gravitational radiation debate (Collins 1985, 79-111). Einstein's general theory of relativity predicts that moving bodies will produce gravity waves. These waves are so weak that ordinary detectors cannot detect them. In the late 1960s, Joseph Weber constructed a detector which, according to him, could detect gravitational radiation, if there really is such a thing. In 1969, he reported that his detector had really detected some waves. Other scientists tried to replicate the result, but none of them succeeded. By 1975, scientists reached an agreement that there had been something wrong with Weber's experiment.
Collins uses this case to support his "experimenter's regress" argument (Collins 1985, 83-84). In this kind of situation, no one really knows what the correct outcome of the experiment is, because "what the correct outcome is depends upon whether there are gravity waves hitting the Earth in detectable fluxes" (84). To determine this, we need to construct a better gravitational wave detector, but to know which one is a better detector, we need to know what a correct outcome would be. As it is, this is a never ending process, and this is what Collins calls 'the experimenter's regress'.
However, as a matter of fact, scientists reach agreements and determine which experiments are valid. Then, how do they reach such agreements? Collins tries to answer this question by interviewing scientists who have participated in the gravity wave debate. He finds that 'non-scientific' reasons have played an important role in the debate. His list of such reasons includes faith in experimental capability and honesty, personality and intelligence of experimenters, reputation of running a huge lab, whether the scientist worked in industry or academia, nationality, and so on (87). At the final step of the debate, many scientists have mentioned the failure of calibration of Weber's detector as a reason for disbelieving his result, but according to Collins this is merely an expression of conservatism. For, there is no assurance that a detector which can detect other kinds of waves (i.e. which can be calibrated) can also detect gravity waves (again, this is a part of the experimenter's regress). Therefore, if scientists took a different attitude, "the anomalous outcome of Weber's experiments could have led toward a variety of heterodox interpretations with widespread consequences for physics. They could have led to a schism in the scientific community or even a discontinuity in the progress of the science" (105).
In this study, major extra-scientific factors are personal evaluations of experimenters. But are they 'social' factors? Collins himself does not argue this point, but we can answer affirmatively. The evaluation of personality, honesty, and status in the community depends on our related cultural norms. Therefore, these cultural norms affect scientists' judgement through their evaluation of each other. If this is a correct interpretation, Collins' study reveals a social conditioning in the same sense that Shapin and Schaffer's study on the Boyle-Hobbes debate does (Shapin and Schaffer 1985). According to Shapin and Schaffer, many of Boyle's air pump experiments were conducted at the Royal Society, and he exploited the reputation of members of the Royal Society to increase the credibility of his experimental results (55-60). In this case the extra-scientific factor is the reputation of gentlemen in the Royal Society.
4-4. Other studies on social conditioning
There are many other studies which fall in this category. Even though not usually classified as social constructivists, feminist epistemology literature (e.g. Keller and Longino 1996) provides a rich examples of gender biases in science. For example, Martin's work argues that biologists had assumed that eggs play a passive role in fertilization, and sperms an active role (Martin 1996). Later they found that eggs play a much more active role than they had thought. Martin shows that both the old view and the new view are associated with cultural stereotypical imageries. In this case, the extra-scientific factor is various gender stereotypes.
On the other hand, there are studies on material influences of society on research design. For example, Callon's study on scallops (1986) can be understood as a case study of this category. He tells us that the research design of biologists studying scallops in St. Brieuc Bay was determined by negotiations between biologists, fishermen, and scallops (!). Biologists, fishermen, and scallops had their own desires and problems to solve, and the research was designed to meet those desires and to solve those problems. This is not an influence on scientists world view, but an influence on material conditions of of research design itself. I said that this study "can be understood" as a case of social conditioning, because Callon himself would rather be classified in the next section. We will return to his work later.
5. Social construction as consensus formation
There are many different ways to explore the process of consensus formation in science. Each of them reveals different aspects of the process. In this section, I discuss typical methods, namely interview, anthropological research, ethnomethodology, and rhetorical analysis. I also briefly discuss two distinct views about consensus formation, namely consensus among human beings and consensus including non-humans.
5-1. Interview -- experimenter's regress again
The study by Collins discussed above also provides a representative work on the consensus formation process. His interviews reveal the process through which scientists have negotiated to determine which experiments are valid. When we understand Collins' work in this way, the gravity wave debate is social not because scientists use extra-scientific considerations, but simply because scientists negotiated. Which is Collins' original intention? His reply to Franklin's criticism seems to support the latter interpretation (Collins 1994; Franklin 1994). Franklin's argument is based mainly upon published works of the scientists. He finds many scientific reasons to reject Weber's results. We do not have to bring extra-scientific factors into consideration to break the experimenter's regress. Franklin concludes: "Collins's view that there were no formal criteria, applied to deciding between Weber and his critics, may be correct. But, the fact that the procedure was not rule-governed, or algorithmic, does not imply that the decision was unreasonable" (Franklin 1994, 471). To this criticism, Collins replies that "I have never suggested that scientists' actions were unreasonable" (Collins 1994, 501). Collins admits that these scientific reasons played an important role in the debate. But these reasons are not enough to reach a conclusion. Nevertheless, most of these extra-scientific reasons are also reasonable in some sense. Collins concludes: "it is just that 'reasonableness' is a social category; it is not drawn from physics" (Collins 1994, 503).
The last sentence suggests that Collins thinks that scientific rationality (a version of reasonableness) is a social category. Thus even when scientists settle arguments from scientific reasons, what they are doing is still social. If this interpretation is correct, Collins is talking about the consensus formation process in my sense.
5-2. Anthropological research -- laboratory life
Anthropological research, namely in situ observation of participants' activity, is utilized by some sociologists, and Latour and Woolgar's Laboratory Life: The Construction of Scientific Knowledge is one of the earliest attempts at such a study (Latour and Woolgar  1986). Latour spent two years in the Salk Institute as an outside observer inside the laboratory, 'outside' in the sense he did not have enough background knowledge about biochemistry to participate in the scientific activity itself, and 'inside' in the sense that he was sitting in the laboratory and taking notes on the conversations among scientists.
Latour and Woolgar are very conscious of the way they use the word 'social'. First, they refuse the use of 'social' as opposed to 'technical' factors in science (roughly the usage I discussed in the previous section). Rather, they "regard the use of such concepts as a phenomenon to be explained" (27). Thus, they propose a working definition of their own sense of 'social': "it could be said that we are concerned with the social construction of scientific knowledge in so far as this draws attention to the process by which scientists make sense of their observations" (32; emphases in original). In what sense is this process 'social'? Of course they are providing a definition, but if the defined term 'social' has nothing to do with the common use of the term, then the validity of the definition itself can be questioned. The answer can be found in their criticism of attempts to differentiate scientific activity from non-scientific activity (151-183). Latour and Woolgar analyze the construction of facts in conversations, retrospective construction of thought processes, and construction of artifacts and reality. In all cases they find nothing mysterious about scientific activity, and conclude that we do not need to accept that scientific activity has any special character (183). I think that this account can be understood as a specific case of my consensus formation process. What Latour and Woolgar describe is the process of consensus formation about facts, artifacts etc. in the laboratory as a small community.
This conception of 'social' leads Latour and Woolgar to the conclusion that everything is social. In the first edition, the subtitle of the book was "the social construction of scientific facts" (emphasis added). The word "social" was dropped in the second edition. They explain the reason for doing this as follows: "In all such uses [uses by Merton and Edinburgh school], 'social' was primarily a term of antagonism, one part of a binary opposition. But how useful is it once we accept that all interactions are social? What does the term 'social' convey when it refers equally to a pen's inscription on graph paper, to the construction of a text and to the gradual elaboration of an amino acid chain? Not a lot. By demonstrating its pervasive applicability, the social study of science has rendered 'social' devoid of any meaning" (Latour and Woolgar 1986, 261; emphasis in original).
5-3. Ethnomethodology -- discovery of a pulsar
The ethnomethodological approach is another way to study the consensus formation process. Ethnomethodologists also analyze conversations, but their conversational analysis is much more detailed than in the study of Latour and Woolgar. In a conversational analysis, ethnomethodologists analyze structural features of conversations, like turn taking (Lynch 1992, 239). Ethnomethodologists also emphasize "indexicality" of language, namely context dependence of our utterances (233-235). Even seemingly abstract rules are not really context-independent, because they just elaborate order which is already produced within activities described by the rules (242).
The study of an optically discovered pulsar is a representative work of the ethnomethodological study of science (Garfinkel, Lynch, and Livingston 1981). The material they study is an approximately one-hour-long tape recording of the conversations among scientists (Cocke and his colleagues) at the moment of discovery of the pulsar. Through transcribing and analyzing the tape recording, Garfinkel et al. reveal the process in which an observation of pulse on the screen was developed into a discovery of an object, the Independent Galilean Pulsar (IGP).
Lynch (1992) points out several particularly ethnomethodological aspects of this study. First of all, the IGP is treated by Garfinkel et al. as a "cultural object" instead of a physical or natural object (Lynch 1992, 248-249; Garfinkel et al. 1981, 141). Why can a pulsar be a cultural object? Because it "is 'extracted' from a succession of observational runs with the optical and electronic equipment" (Lynch 1992, 248-249). The IGP is not a cause of the discovery, but is constituted in the "local historicity" of the scientists' work (249).
It is still not clear to me why they use the word "cultural" instead of "social", but the difference does not seem to be a serious one in this context, and I think that it is safe to classify this work in the same category as Latour and Woolgar's one, namely as a case of the consensus formation process. Here the consensus is the establishment of the discovery of an object, and the process is analyzed by a structural analysis of the conversation among scientists.
5-4. Rhetorical analysis
Rhetorical analysis of scientific writings, such as Latour (1987) and Gross (1990), sheds light on another aspect of consensus formation. Let us take Latour's analysis of what he calls "black boxes" (Latour 1987, 1-100). Latour proposes to look at science from two sides, "ready made science" and "science in the making" (4). For example, the double helix structure of DNA is a ready made fact for many of us, and the complex process of Watson and Crick's discovery of the structure is a black box for us. But, according to Latour, to understand science we have to open up the black box, to see science in the making.
Latour analyzes the strategies used in scientific controversies by rhetorical analysis of scientific articles. He finds that the strategy is fairly simple: to anticipate oppositions, and to bring in as many supports as possible to meet them. To cite other papers is a common way to bring in authorities (33-38). At the same time, to write a paper which will be cited in the future is also important, because the reputation of a paper depends on the number of citations in other papers (38-44). Of course supports from experimental results in a laboratory is essential, but these results themselves can be analyzed rhetorically, because experimental results without interpretation and commentary do not mean anything (70-74). To do experiments and to interpret results, scientists need to use much taken-for-granted knowledge, that is, other black boxes.
This analysis also suggests the way to attack a scientific article. First, the dissenter should check the ties between the article and supportive facts. If she wants to dissent from the experimental results, she can first check the tie between the experiments and their interpretations and commentaries, but ultimately she may need to build another laboratory stronger than the one used for the original article (74-94). Of course the dissenter can borrow other black boxes to support her claim. Anyway, the controversy is settled when one side of the debate gives up attempts to get stronger support. By the settlement of the debate, the claim of the article becomes a new black box.
To understand Latour's rhetorical analysis, it is important to note that the word 'rhetorical' is not supposed to be in opposition to 'rational', but is supposed to mean all persuasions including rational ones (Gross is more explicit on this point; see Gross 1990, 9-12). In this sense, rhetorical analysis does not primarily concern extra-scientific factors in scientific articles (even though rhetorical analysis can deal with these factors also, if there are any). This point is reflected in the way Latour explains why this kind of rhetorical analysis shows that science is 'social': "the more technical and specialised a literature is, the more 'social' it becomes, since the number of associations necessary to drive readers out and force them into accepting a claim as a fact increase" (Latour 1987, 62; emphasis in original).
I think that this is a good study of the consensus formation process. A scientist needs to organize resources in the scientific community through citation and other ways of borrowing black boxes. Consensus appears as a result of this activity. Rhetorical analysis helps us to understand rules of such consensus formation.
5-5. Society including non-humans
All of the above discussed cases deal with consensus among human beings. This sounds like a trivial statement -- how can a non human participate in the consensus? However, there are several influential sociologists who challenge this common sense.
In a paper entitled "Mixing humans and nonhumans together: the sociology of a door-closer" (1988), Latour (using a pseudonym 'Jim Johnson') argues that sociology should deal with non-human artifacts like door closers on a par with humans. His major reason is that these artifacts are often delegates for humans (door-closers are delegates for humans who stand by and close doors).
The aforementioned study by Callon is an example of this kind of sociology (Callon 1986). The biologists tried to find out a way to protect scallops in St. Brieuc Bay, and they had to negotiate not only with fishermen but also with scallops. They had to prepare the situation so that the scallops could properly react. Throughout the paper Callon tries to show that the relationship between biologists and fishermen is symmetrical with the one between biologists and scallops.
According to these views, science is social even when there is no interaction between scientists. If they use experimental instruments or they interact with the object of research, this is already social. This view is too radical for other sociologists, and Collins and Yearley discuss that Latour and Callon have gone too far (Collins and Yearley 1992a). The debate caused by the paper is known as an "epistemological chicken" debate, but the debate is beyond the scope of this paper (see Callon and Latour 1992, Collins and Yearley 1992b, Pickering 1995, for more about the debate).
6. Problems with each approach
In the last three sections (sections 3-5), I have explained what I mean by social conditioning and consensus formation, and used these notions to classify and analyze typical works in the sociology of scientific knowledge. Now, the question is whether this distinction is important or not -- whether it is more important than other classifications, e.g. Sismondo's. In the next two sections, I try to justify my taxonomy from two points of view. In this section, I argue that the taxonomy corresponds with particular problems sociologists face. In the next section, I argue that by distinguishing these two processes clearly, we can find interesting sociological questions on the relationship between the two processes.
6-1. Problems with the social conditioning approach
If a sociologist want to argue for the existence of specific social conditioning in science, she faces a problem proper to this kind of approach. As I mentioned above, Cole deals with this problem in a chapter in his book called "constructivist problems in demonstrating causality" (Cole 1992, 53-81). Cole argues that the causal claims by social constructivists are inadequate in two points (Cole 1992, 61-62). First, they often fail to identify social variables and cognitive contents independently. According to Cole, "they frequently refer to decisions made on cognitive or idiosyncratic psychological bases as being social, because the decisions are made in a social environment" (61). But if we should admit this as a causal relationship between society and the cognitive content of science, then everything is social by definition. This makes social constructivism uninteresting. The second point is that social constructivists fail to demonstrate any linkages between social processes and "knowledge outcomes" (ibid.). Cole means by knowledge outcomes the specific content of a scientific idea, like E= mc2. Social constructivists have succeeded in showing that scientists' everyday activity is influenced by social variables, but at the same time social constructivists are "black-boxing" knowledge outcomes.
I think Cole's criticism makes a good point. For example, Collins' study (as is understood in section 4-3) has cultural norms for evaluating people as a social variable, and the result of the gravity wave debate as dependent cognitive content. But if this is really a causal relationship, then under other cultural norms the outcome of the debate would be different. Collins surely asserts this (Collins 1985, 105), but this assertion is not supported by any argument at all.
Forman's study did a good job in establishing a causal relationship between Weimar culture and the interpretation of quantum mechanics, but still there are many criticisms, e.g. Hendry (1980), Brush (1980), and Kraft and Kroes (1984). All of them point out some gaps in Forman's argument, but Brush's thesis provides the most interesting case. He finds that both science and culture have swung between realism and romanticism since 1800 periodically (he finds roughly 35-year cycle). The cycle in culture and that in science synchronize with each other, and this seems to suggest some correlation between them. But Brush proposes an alternative explanation that practitioners in science and those in culture have similar response characteristics, like "a tendency to oppose any such trend after it has gone to an extreme, and to react by moving in the opposite direction" (Brush 1980, 394). If we assume that they have these similar response characteristics, and that both science and culture started the period of romanticism around 1800 (at that time science and culture were not clearly divided), then we can explain the synchronism of Weimar culture and the acausal interpretation in physics without appealing to any causal relationship between them. This ingenious alternative explanation shows that establishing a causal relationship is really difficult, especially in a historical case study.
6-2. Problems with the consensus formation approach
Cole also points out "constructivist problems in accounting for consensus" (Cole 1992, 33-60). He claims that "constructivist sociologists of science have no convincing way to explain why some ideas win out in the competition for communal acceptance" (59). Social constructivists describe how a scientific claim has been accepted, but they do not explain why one claim is accepted instead of another, why it has been accepted so rapidly, and so on. The problem is more serious with multiple discoveries. If scientific discoveries are constructed in their social settings, why is it scientists in different settings can find the same thing independently from each other? Cole also thinks that if constructivists use "cognitive interests", i.e. a scientist's interest to accept new works which support her own past contributions, these interests can explain why one theory is accepted instead of another, but this explanation "tends to be a tautological and ex post facto argument", and even worse, it is hardly a sociological explanation (53).
Cole does not seem to be aware of the fact that those sociologists who discuss consensus formation use the word 'social' in a peculiar way. They do not distinguish cognitive factors from social factors, so even a purely cognitive explanation can be a social one. Cole's reply would be that such a use of the word 'social' makes sociological explanation vacuous. Or, to capture his intention, maybe 'superfluous' is a better word; those sociologists can even accept a traditional philosophical explanation of consensus formation, and add "and this is social". But in such an explanation, the role played by sociologists is totally superfluous.
This is a real danger for the works I discussed above. For example, Tilly (1981) argues that Latour and Woolgar's Laboratory Life corroborates Popper's view that scientists try to refute their own hypotheses. If Tilly is right, what Latour and Woolgar is doing is to support Popperian explanation of science and to add "and this is social". Tilly's method is easily applicable to Garfinkel et al. (1981) and Latour (1987). For example, the tape recording used by Garfinkel et al. reveals that Cocke and Disney tried to eliminate counterpossibilities that the observed pulse was an artifact before they called up another colleague. Popperians can easily describe this process in Popperian terminology, such as conjecture and refutation. If this explanation is enough, to add "and this is social" is superfluous. Latour's analysis of black boxes also allows traditional philosophical descriptions. Confirmation theorists would be able easily to show that the role of borrowed black boxes is to confirm the hypothesis in the paper.
Sociologists sometimes claim that such philosophical factors are only a part of the picture and cannot explain everything (Collins 1994, 503; Latour and Woolgar 1986, 280-281). Then they are arguing for the position that extra-scientific factors (or at least factors which cannot be dealt with by traditional philosophy of science) play an essential role in consensus formation. But this move is not consistent with the way they use the word 'social'. Rather they are arguing for social construction as social conditioning. This means that they are now responsible for the problem discussed in the previous section.
Actually I am not sure whether the superfluity is really problematic. Rather, it seems to me that this is a good place where sociology of scientific knowledge can cooperate with traditional philosophical approach to reveal the rules of the game. It may be true that Popper himself and his theory refuses such a cooperation, but recent development of philosophy of science has made the cooperation much easier. For example, I think Cherniak's (1986) notion of the "special reasons requirement" is a plausible candidate for a rule actually used in scientific debates. The special reasons requirement says that to raise a doubt about a knowledge claim, you need to show a plausible specific reason to doubt. If you want to doubt an experimental result, a mere possibility of a mistake is not enough, but you should specify a particular part of the experiment which could plausibly have gone wrong. I think that the special reasons requirement provides a deeper insight when combined with, say, Latour's black box analysis. By the cooperation, the sociologist gets a tool to analyze the deeper structure of the black-boxing process, and the philosopher gets a material basis for the philosophical thesis.
6-3. For more effective debates
What is the point of pointing out the different problem situation between two types of research? Many social constructivists seem to assume that when scientists negotiate, extra-scientific factors inevitably come into play (this is why the distinction is not usually made). This is most notable in the reply to Franklin by Collins, cited above. He says that scientific reasons were not enough to settle the gravity wave debate, but he does so without any argument. If he had been conscious about the distinction, then he would not have answered in such a way. He could just assert that scientific rationality is social as is. This reply causes the problem pointed out by Cole above, namely the word 'social' can become superfluous in such an explanation. But I believe that sociologists will be better off by consciously struggling with this latter problem, rather than keeping on asserting that extra-scientific considerations are essential in consensus formation. By having in mind two different senses of 'social', we can expect more effective debate among sociologists, and hopefully between philosophers and sociologists, too.
7. A heuristic value of the distinction
7-1. A new research program for sociology of scientific knowledge
So far, I have argued that there are two distinct senses of 'social' in the constructivist literature, and research projects associated with each of the two senses have their own problems. Now I want to turn to another advantage of the distinction. I think that by distinguishing the two senses clearly, we can think about the correlation between two kinds of social processes, and this leads to a new fruitful field of sociological research.
What I want to capture by the distinction is the paradoxical fact that what scientists do is deeply determined by society (by social conditioning) and still science has its own life independent of society (by consensus formation). To see the point, let us consider a question often raised to constructivists: "If scientific facts are socially constructed, why do airplanes fly anyway?" This naive question needs an explication. When we look at scientific activity, what many of us are most impressed is its ability to create novel phenomena and to control (replicate, predict, etc.) them. This ability becomes even more impressive when we compare scientific activity with other system of knowledge. For example, aerodynamics is an indispensable part of making airplanes, and if we replaced aerodynamics with other system of knowledge, like crystal ball gazing, the airplanes would not fly. Even in its own field, crystal ball gazing does not create novel phenomena in the way science does. Then the question boils down to this: what is so peculiar about scientific activity?
I think that sociologists tend to overemphasize the fact that science is just another social activity. To admit that there is something peculiar about science does not exclude sociological research on science. To the opposite, I think here is a genuine sociological question. Imagine that there is a country which has a significantly lower crime rate than other countries. What sociologists would do is not to argue that there is nothing special about this country, but to attempt to find some sociological explanations of the peculiarity. Similarly, here is a community with significantly higher rate of creating and controlling novel phenomena than other communities. I think that it is sociologists' task to find sociological explanation of this peculiarity.
My distinction between social conditioning and consensus formation offers a research program to answer such a question. We should find some peculiar structure, norms, and decision processes in science. This is an old Mertonian theme, but the situation is now more difficult than in the time of Merton, because now we know that scientists are neither disinterested, communalist, nor universalist. Rather they are deeply situated in their social context, and are influenced by the ideology of the age and their own interests. If these social factors are causally relevant to knowledge outcomes (in Cole's term), i.e. if Forman, Bloor and others are right, maybe what we need to look for is some peculiar causal relationship between them. If they are not causally relevant to knowledge outcomes, i.e. if Cole is right, maybe we need to suspect that there is some systematic way to rule out or to neutralize such factors. In either case, the focus of the study will be the paradoxical relationship between the scientific community and larger society.
I would like to call this research program "the even stronger program". Bloor's "strong programme" advocated the necessity of a sociological explanation of scientific knowledge without appealing to scientific rationality. In this sense, scientific rationality itself was out of the picture of Bloor's program. Now, I advocates the necessity of a sociological explanation of scientific rationality itself, especially an explanation of the asymmetry between science and non-science. In this sense, my program is "even stronger" than Bloor's one.
7-2. Studies on the correlation between two processes
Actually there already are studies on the correlation between social conditioning and consensus formation, though they may not be consciously concentrating on this problem. In this section I would like to critically analyze two such studies briefly, namely Bourdieu's study of science as a social field, and Fuller's social epistemology.
7-2-1. Bourdieu's study of science as social field
Bourdieu (1991) clearly recognizes the paradoxical nature of science, namely science as historically situated social activity can yield universal truth. He regards science as a field of the struggle for symbolic power. In this sense there is nothing special about the scientific field. This may sound relativistic view of science, but he maintains that "it nevertheless remains true that, under certain conditions, that is in certain states of this field of struggle for symbolic power that indeed is the scientific field, these strategies produce their own transcendence, because they are subjected to the crisscrossing censorship that represents the constitutive reason of the field" (20). He tries to explain this paradoxical fact by analyzing the historical development of the autonomy of science from society. Bourdieu points out several independent variables which can affect the degree of autonomy of a scientific field -- such as volume of scientific resources accumulated in the field, intensity of the constraints exercised by external powers (15). The accumulation of scientific resources (like necessary background knowledge to be competent in the field) enables clear cut break between the scientists and the lay people, and also enables intense cross censorship among scientists. Of course, the autonomy alone does not assure that the field becomes scientific. What is required in addition is that "those who participate in the game have an interest in truth instead of having, as in other games, the truth of their interests" (22). According to Bourdieu, the only way to study the emergence of such a game is a "historicized epistemology" (23). In a historicized epistemology, the emergence is described as a result of "the (historical) encounter between the product of two partially independent histories" (ibid.), namely between historical agents under specific social conditions and a historical field.
I think that Bourdieu shares a major part of my concern about the paradoxical nature of science, and the baseline of his strategy is the same as my above suggestion, namely he tries to account for the fact by analyzing the relationship between society and the scientific community. What I do not agree with him is the emphasis on historicity of the emergence of the scientific field. It is true that each scientific field has its own peculiar history, but still I think that there are common features among these histories which were essential for making these field scientific. Anyway, Bourdieu's study is still a program for future studies, and we need to fill in the blanks in our knowledge about the history. In the course of such future studies, maybe Fuller's social epistemology discussed in the next section will be helpful to find common aspects in the histories of different fields.
7-2-2. Fuller's social epistemology
Steve Fuller proposes to resurrect normative epistemology as science policy-making, and to call this attempt "social epistemology" (Fuller 1988). First, let us summarize his basic ideas according to his recent article (Fuller 1992). The starting point of social epistemology is the doubt that "left to its own devices, science will not necessarily produce the sort of knowledge that we are interested in having" (392). The reason we want knowledge is independent of science itself, and we can judge the efficacy of science from this outside point of view. For such judgements, Fuller proposes to use an experimental method. For example, we can conduct experiments taking independent variables such as "group size, communication constraints, and background information" and dependent variables such as "the sort of scientists, scientific audience, and knowledge commodity that is produced" (405). These experiments can decompose the overall effect into its working causal parts, and can also compensate for sociologists' interpretative biases (404-405). Of course the experimental results are artificial in a sense, but this does not bother social epistemologists. If we find effective institutional structures in experimental settings, what social epistemologists may want to do is make "the world behave more like the laboratory than vice versa" (406).
In Fuller's account, extra-scientific factors come into play from two directions. First, each scientist has self interests and other social conditionings in natural settings. But scientists may behave suboptimally in such settings, so we need to intervene to restructure scientific institutions. It seems to me that the motivation of the intervention can be purely epistemic, but Fuller seems to think that this is usually another extra-scientific factor. Anyway, if the intervention is successful, what I have been calling the 'consensus formation process' will emerge as a result of it.
I think that Fuller's program is a promising way to answer the question I raised above. The peculiarity of science comes from the peculiarity of our motivation to shape scientific institutions. Namely, what we want from science is different from what we want from astrology or crystal ball gazing. Fuller's experimental method is helpful to answer the question of in what way science is different from other fields.
One of my objections to Fuller's program is that he does not seem to recognize the paradoxical nature of science as I described above. This is mainly due to his conception of science which makes science totally dependent on society. To me it seems obvious that the exact content of scientific knowledge is largely independent of social variables, and Fuller's account needs to be revised to include this independence dimension. Bourdieu's program may supplement social epistemology in this respect.
In this paper, I have introduced the distinction between two senses of 'social' in social constructivist literature. The distinction is useful not only to classify typical studies in this field, but also to clarify what is at stake in the debate about social constructivism. In the final part of the paper I have introduced a new research program for sociology, namely the inquiry into the paradoxical nature of science. To think about this question, my distinction between two senses of 'social' provides a good conceptual scheme. I have also applied the scheme to two existing studies which can be regarded as attempts to inquire into the paradoxical nature of science (though Fuller does not recognize the situation as such). I think some combination of Bourdieu's study and Fuller's study makes a fruitful research program. Namely, historical study of the development of autonomy of science and experimental study of efficacy of various scientific institutions can be benefited by informing each other.
 In the literature, terms 'sociology of scientific knowledge (SSK)' and 'social constructivism' are sometimes used interchangeably and sometimes used to denote different schools. I use the term 'SSK' to denote the field, and 'social constructivism' to denote the dominant theories in the field (like the relationship between 'physics' and 'the general theory of relativity').
 Ethnomethodologists would add a reservation that rules in the scientific community is not a precondition, but a result of scientific activity. See section 5-3 below.
 I said 'roughly' because Cole includes causality inside scientific community in the problem with demonstrating causality. Cole also makes a distinction between "macro-level" and "micro-level" approaches (Cole 1992, 35), and his "macro-level" roughly seems to correspond to my social conditioning.
 We find similar claims in Latour and Woolgar  1986, 180-182. This shows both group of scholars are committed to Sismondo's fourth category of social construction, i.e. social construction of reality.
 Fuller himself would deny the alleged peculiarity of science. Fuller thinks that scientific activity makes sense "only if you do not look too closely and are of a rather charitable turn of mind" (Fuller 1992, 401). I think Fuller underestimates the difference between science and other systems of knowledge.
Bloor, D. ( 1991) Knowledge and Social Imagery second edition. Chicago: University of Chicago Press.
--. (1981) "The strength of the strong programme" in Philosophy of Social Science 11, 199-213.
Bourdieu, P. (1991) "The peculiar history of scientific reason" in Sociological Forum 6, 3-26.
Brush, S.G. (1980) "The chimerical cat: philosophy of quantum mechanics in historical perspective" in Social Studies of Science 10, 393-447.
Callon, M. (1986) "Some elements of a sociology of translation: domestication of the scallops and the fishermen of St. Brieuc Bay", in J. Law (ed.) Power, Action and Belief, 196-226.
Callon, M. and Latour, B. (1992) "Don't throw the baby out with the Bath school!" in Pickering (ed.) 1992, 343-368.
Cherniak, C. (1986) Minimal Rationality. Cambridge: The MIT Press.
Cole, S. (1992) Making Science: Between Nature and Society. Cambridge: Harvard University Press.
Collins, H. M. (1985) Changing Order: Replication and Induction in Scientific Practice. London: Sage Publications.
--. (1994) "A strong confirmation of the experimenter's regress" in Studies in History and Philosophy of Science 25, 493-503.
Collins, H.M. and Yearley, S. (1992a) "Epistemological chicken" in Pickering (ed.) 1992, 301-326.
--. (1992b) "Journey into Space" in Pickering (ed.) 1992, 369-389.
Forman, P. (1971) "Weimar culture, causality, and quantum theory, 1918-1927: Adaptation by German physicists and mathematicians to a hostile intellectual environment", in Historical Studies in the Physical Sciences 3, 1-115.
--. (1979) "Reception of acausal quantum mechanics in Germany and Britain" in S. H. Mauskopf (ed.), The Reception of Unconventional Science. Washington D.C.: American Association for the Advancement of Science.
Franklin, A. (1994) "How to avoid experimenter's regress" in Studies in History and Philosophy of Science 25, 463-491.
Fuchs, S. (1993) "A sociological theory of scientific change" in Social Forces 71, 933-953.
Fuller, S. (1988) Social Epistemology. Bloomington: Indiana University Press.
--. (1992) "Social epistemology and research agenda of science studies" in Pickering (ed.) 1992, 390-428.
Garfinkel, H., Lynch, M. and Livingston, E. (1981) "The work of a discovering science construed with materials from the optically discovered pulsar" in Philosophy of Social Science 11, 131-158.
Gross, A. (1990) The Rhetoric of Science. Cambridge: Harvard University Press.
Hendry, J. (1980) "Weimar culture and quantum causality" in History of Science 18, 155-180.
Keller and Longino eds.(1996) Feminism and Science. New York: Oxford University Press.
Knorr-Cetina, K. (1981) The Manufacture of Knowledge: An essay on the Constructivist and Contextual Nature of Science. Oxford: Pergamon.
Kraft, P. and Kroes, P. (1984) "Adaptation of scientific knowledge to an intellectual environment: Paul Forman's 'Weimar culture, causality, and quantum theory, 1918-1927': analysis and criticism" in Centaurus 27, 76-99.
Latour, B. (1987) Science in Action. Cambridge: Harvard University Press.
--. [Johnson, J.] (1988) "Mixing humans and nonhumans together: the sociology of a door-closer" in Social Problems 35, 298-310.
Latour, B. and Woolger, S. ( 1986) Laboratory Life: The Construction of Scientific Facts 2nd edition. Princeton: Princeton University Press.
Laudan, L. (1981) "The Pseudo Science of Science?" in Philosophy of Social Science 11, 173-198.
Lynch, M. (1992) "Extending Wittgenstein: the pivotal move from epistemology to the sociology of science" in Pickering (ed.) 1992, 215-265.
Martin, E. (1996) "The egg and the sperm: how science has constructed a romance based on stereotypical male-female roles" in Keller and Longino 1996, 103-117.
Mendelshon, E. Weingart, P. and Whitley, R. eds. (1977) The Social Production of Scientific Knowledge. Dordrecht: Reidel.
Pickering, A. (1995) The Mangle of Practice. Chicago: University of Chicago Press.
Pickering, A. ed. (1992) Science as Practice and Culture. Chicago: University of Chicago Press.
Pinch, T.J. and Bijker W.E. (1987) "The social construction of facts and artifacts: or how the sociology of science and the sociology of technology might benefit each other" in Bijker, W.E., Hughes, T.P. and Pinch, T.J. (eds.) The Social Construction of Technological Systems: New Directions in the Sociology and History of Technology. Cambridge: The MIT Press.
Shapin and Schaffer (1985) Leviathan and Air-Pump: Hobbes, Boyle and the Experimental Life. Princeton: Princeton University Press.
Sismondo, S. (1993) "Some social constructions" in Social Studies of Science 23, 515-553.
Tilly, N. (1981) "The logic of laboratory life" in Sociology 15, 117-126.