Review of Barry Barnes, David Bloor and John Henry, Scientific Knowledge: A Sociological Analysis, University of Chicago Press, Chicago, 1996.

Scientific Knowledge: A Sociological Analysis is an introductory book into the sociology of scientific knowledge. Barnes et al do not pretend to provide a balanced overview of the field. It is a biased introduction that largely ignores alternative approaches to SSK (the sociology of scientific knowledge), such as those provided by Bruno Latour, Steve Woolgar and Andrew Pickering, to name a few. Barnes et al make no apologies for this bias. They believe their version of SSK to be “particularly well-suited” for an introduction to SSK (p. 200). Part of this ‘well-suitedness’, for them, is their allegiance to materialism. They believe too many versions of SSK have been idealist and have sought to deny the existence of an extra-mental reality. Here they exclusively cite Harry Collins, though Woolgar would have been a more poignant example. At any rate, materialism they claim puts them “in touch with other fields that study human beliefs and dispositions — psychology and biology for example” (p. 210).

Overall, the approach they take to introducing their version of SSK is foundationalist. They start in chapter one with an examination of empirical, perceptual psychology. The ground they cover here is familiar to philosophers. Their focus is the debate between Jerry Fodor and Paul Churchland concerning whether perceptual processes are theory-laden to the extent that what one perceives is a product of what theory one believes. Here they side on empirical grounds with Fodor’s modularity theory according to which the phenomenal character of perceptions is, to a degree, impervious to cognition. This is clearly the more difficult path for defenders of SSK, since the social constructedness of knowledge would be assured if theoretical beliefs informed our perceptions in the way Churchland suggests. On this psychological basis, they next examine the socio-cognitive processes involved in interpreting perceptual experience (chapter two). Their discussion of this topic is unclear and unsystematic since they focus entirely on Robert Millikan’s discovery of the quantization of electric charge (the famous oil-drop experiment). Apparently, Millikan’s work is paradigmatic of how perceptions are interpreted. It is true that Millikan’s work is considered a ‘classic’ in the history of science, as Barnes et al point out (p. 18). And let us grant for the sake of argument that the methods Millikan used to support his beliefs about the quantization of energy were both epistemically suspect (Millikan presumably discarded legitimate data) and socially motivated — Barnes et al comment, “it is the experiment qua institution that is at issue here” (p. 40). It follows that as evidence for the social constructedness of knowledge, the Millikan case is quite compelling. Nonetheless, if one wants to defend generalizations concerning the social mechanisms of perceptual interpretation, then one should proceed generally, as they did in chapter one. By examining exclusively the interpretive strategies use by Millikan and citing their social character, one does not demonstrate, in a general way, the social character of interpreted perceptions.

From a discussion of how perceptions are interpreted, they next consider how language evolves from a foundation of interpreted perceptions (chapter three). The content of this chapter is central to their book, and central to their SSK. For it introduces their favoured view concerning the nature of linguistic classification — the ‘finitist’ view. Finitism can be simply stated as follows: the application of any classificatory term is logically imprecise; thus, to render its application more precise, one has to supply extra-logical considerations, which for Barnes et al involve psychological or sociological considerations. In other words, finitism entails the legitimacy of SSK: the precise application of any classificatory term must involve social negotiation. Obviously, the key to their finitism is their defense of the logical imprecision of classificatory terms, which proceeds as follows. They start with the assumption that terms are connected to the physical world through (i) acts of ostension, and (ii) feelings of similarity between objects which compel speakers to extend the use of ostended terms to further non- identical objects (see pp. 49-53). They then suggest that any two objects, however dissimilar, could be linked by a feeling of similarity and so grouped under one term — that is, logically they could be so linked, although perhaps not psychologically nor sociologically. Thus, classificatory terms are inherently logically imprecise and need supplementation by psycho-social forces to be given a consistent usage. It is here that Barnes et al’s defense of SSK becomes extreme. The logical possibility of alternative classificatory schemes is too weak a foundation on which to ground the necessity for sociological contrivance. For there is another question to answer before making this leap — i.e., is the grouping of two objects under the same term not only logically possible but epistemically possible as well? This question Barnes et al ignore to the detriment of their program. For it is logically possible to group red and blue under the same colour term, despite the psychological and sociological discomfort in doing this. But don’t we agree that, psychological and sociological factors to one side, it just doesn’t seem reasonable to group red and blue under the same colour term?

Let me pause to consider this important question. What I’m suggesting is that one’s grounds for distinguishing red from blue is a separate issue from the question of whether one is psychologically or sociologically compelled to separate these colours. One can have (non-logical) grounds for separating these colours, despite a psycho-social compulsion not to do so, and vice versa. Barnes et al wish to ignore this fact. Perhaps this is fair enough, since their topic isn’t epistemology but SSK. Nevertheless, a distinction needs to be drawn between having reasons for drawing a classification and being psychologically or sociologically compelled to do so, for whether a belief constitutes knowledge is often taken to depend on whether one has reasons for one’s belief. And, after all, Barnes et al are doing the sociology of scientific knowledge!

Barnes et al have a response to this criticism. Their response is a reiteration of Collins’ notion of the experimenter’s regress. In short, one cannot non-circularly assert the accuracy of one’s classifications, for to assert the accuracy of one’s classifications one must assert beforehand the reliability of one’s classificatory procedure, and one cannot assert the reliability of one’s classificatory procedure without asserting its accuracy. Thus, they believe, one must abandon the epistemic study of experimental (and more generally observational) practice for a more psychological or sociological approach. However, the experimenter’s regress flounders because of its dependence on a dubious epistemic principle: that to be justified in asserting a claim, one must be justified in asserting its justification. For example, they don’t believe that to be justified in asserting a claim, it is enough that one is a reliable producer of true claims (and that one believes this). One must further prove one’s reliability and prove it in the strongest way possible, by demonstrating the accuracy of one’s beliefs. Call the principle at work here the JJ- thesis; it is akin to the KK-thesis, that to know, one must know that one knows. Both the KK-thesis and the JJ-thesis lead to skepticism since both lead to a justificatory regress. Hence, in the absence of strong arguments in their defense — and Barnes et al provide no such arguments — they should be discarded as epistemic constraints. Once we discard the JJ-thesis (along with the KK-thesis), we remove Barnes et al’s objection to the possibility of epistemically grounded classificatory distinctions, where this objection is motivated by the experimenter’s regress. In turn, we re-establish the sense in which the classification of terms can be epistemically grounded. Classificatory precision would then no longer demand the introduction of psychological or social factors.

At this point, Barnes et al have a further response. The basis of their finitist view of classification is the claim that determining the referent of a term is inexorably indefinite — that one really doesn’t know if the use of a term should be extended to a certain object or not. Logically, it could, or it couldn’t; logically, for example, I could call my pen a ‘pencil’, if I wished to stretch my use of terms in this way. Psychologically and sociologically, of course, such an extension won’t succeed, so at this stage the sociology of scientific knowledge takes hold. Thus, Barnes et all arrive at the following tenet of finitism as applied to beliefs: if what terms refer to is open-ended, then what beliefs are about must also be open-ended. Just as there is a sense in which we do not know what our terms mean, in that same sense, we do not know what to believe (p. 71) Again, we have a form of skepticism that is underwriting Barnes et al’s SSK. Given that we lack knowledge about the meanings of terms or what we believe, social and psychological factors are needed to supply definiteness to the meanings of words and our beliefs. However, such skepticism is unwarranted in the usual case. Given that the application of a term is unclear in certain areas, it doesn’t follow that it’s unclear in all areas. For example, I may be unsure about whether to call a particular shade of colour orange or red. But my lack of certainty here doesn’t affect my certainty about calling some colours ‘red’ and some colours ‘orange’. Because I don’t know how to apply a term in some case, that doesn’t mean that “I don’t know what the term means” or that “I don’t know what to believe” in all or even the usual cases. Such an inference is far too strong. Thus, the social and psychological factors introduced above, needed to render meanings and beliefs more precise, are not needed in all cases or even the usual cases.

In chapter four, Barnes et al extend their SSK to theoretical entities. It is here that Barnes et al’s SSK shows its real value as a scientific enterprise. Let me first remark that empiricism is a viable and compelling philosophic position, where by empiricism I mean: the view that claims about the world should always be put to empirical test. True empiricists don’t ‘a priorize’. Thus, true empiricists when examining the nature of scientific knowledge allow empirical results to guide their inquiry. And so when investigating the nature of scientific knowledge, the following sorts of facts become relevant: that people generally believe in the existence of an external reality whose nature may be completely different from how we think or talk about it (p. 88), that what has counted as science has changed dramatically over the centuries with the appellation ‘science’ often being motivated by non-epistemic concerns, e.g., political interests (chapter 6), that the ‘self-evidence’ of geometrical first principles is historically variable (p. 190), that science often evolves in a Kuhnian manner (chapter 4). The bulk of Scientific Knowledge is devoted to examining in this sort of empirical manner the character of scientific knowledge. In essence, Barnes et al are doing empirical history of science (and math) and allowing it to inform their sociology of science. Some of the historical episodes they consider to this end include Henk van den Belt’s recent work on the 19th century legal battle in France concerning the classification of aniline red dyes (pp. 121- 127; van den Belt’s work illustrates the dependence of classificatory schemes on social interests), Robert Kohler’s 1972 study on 19th century enzyme theory (pp. 129-139; this episode exhibits the limitations in social explanations of institutional explananda), and the controversy surrounding Robert Chamber’s 1845, best-selling, proto-evolutionist book Vestiges (pp. 154-168; it is a controversy that demonstrates the political motivations underlying the way scientists draw disciplinary boundaries). Without a doubt, an empirical methodology such as this is an excellent way to examine science given the pride of place science holds for empirical data. It renders their work, like any scientific work, testable, revisable and defeasible. In fact, it has been put to recent test by Andrew Pickering in his book The Mangle of Practice (Chicago, 1995). Pickering provides convincing historical, empirical evidence against the sort of interests approach Barnes et al advocate in chapter 5. I believe Barnes et al would recognize whole- heartedly the value of this sort of empirically-based criticism, though, of course, they might disagree with Pickering’s conclusions. For this reason, because of its pro-empiricist standpoint, the version of SSK Barnes et al present in Scientific Knowledge is highly commendable.

author’s address: hudsonr@alcor.concordia.ca