The Historiographical Idea of the Automaton-Scientist

"Go Robot: Do Science!"

For the Gallery of Practices to be productive it must be read.  For the Gallery to be created, elements, or portraits, must be selected.  The danger of the Gallery is that the same thing that is read from the Gallery will be the same ideas informing the selection of its elements.  We will take away a conclusion about the relationship between science and the Cold War because we have selected events indicative of our preconceived understanding of what the relationship between “science” and the “Cold War” was like.

However, there is also a danger that even if a reading of the Gallery manages to escape the principles of its construction—say by constructing a gallery from some broad survey—it is still possible to abstract an idea or practice from the Gallery that makes little sense removed from a system of ideas of which it was initially a part.  Instead, the idea becomes incorporated in a system imposed on the Gallery by the historian’s interpretation, much as a Whiggish narrative of history imposes present ideas on the past.

In the last few years, there have been attempts to develop a history of the morality and virtues of the sciences—the history of the “scientific self”.  In The Scientific Life (2008), Steven Shapin drew upon a gallery of historical rhetoric for evidence of changing attitudes concerning whether scientific figures have been morally equated with or separated from ordinary people.  Similarly, in their book Objectivity (2007), Lorraine Daston and Peter Galison have assembled a gallery of scientific “atlases”—collections of images designed to acquaint readers with a body of scientific knowledge—and have connected representational strategies, via a history of ideas about “objectivity”, to a history of the “epistemic virtues” that have inhabited scientific personas .

By making representation a matter of moral imperative rather than a matter of intellectual choice of representational strategy based on a concordant choice of representational goals—such epistemological deliberations do not appear in histories belonging to the Great Escape historiography—Galison and Daston makes sense of perceived patterns in these representational practices by drawing on a longstanding historiographical idea: the automaton-scientist.

The thesis of this post is that the historiographical idea of the automaton-scientist functions as a way of automating history, if you will, by relieving some, or even all, scientists of at least some of the burden of thought, which produces certain historiographically managable kinds of behaviors from historical actors.  Because what is of the most historiographical interest is the development of ideas, reducing the intellectual content of science to ritual allows historians to pass over scientific figures’ work by simply indicating that they “did work”.

Differences in the idea of the automaton-scientist have had implications for the historiographical role of the scientific “genius”. As Simon Schaffer has explained, the idea of the scientific “genius” came to the fore in the early 19th century with the disciplining of scientific inquiry, the articulation of method, and the imagination of an attendant division of labor between the genius-founders of research programs and ordinary scientific laborers.  With his notion of the “paradigm” (1962), Thomas Kuhn automated certain scientific practices as “normal science” that could proceed methodically until too many anomalies accumulated, necessitating the intervention of intellectually and  biographically interesting figures who could initiate a “revolution” and move the science to a new paradigm in which the methodical work of Kuhnian automata could resume.

With the rise of “constructivism” and the initiation of the Great Escape in the 1980s, the intellectually interesting genius-figure was banished.  Also, the intertwining of the social and the intellectual necessitated new universal tools of analysis that did not draw artificial distinctions between the intellectual and social categories of analysis.  Bruno Latour’s semiotic phenomenology accomplished this in a descriptive sense, but did not explain scientific practice at the local level.

Andy Pickering’s Mangle of Practice (1995) did move to the local level by creating a new version of the automaton-scientist, who attempts to advance a theory or practice until “resistance” is encountered—whether the resistance is natural, social, or some inscrutable combination of the two, makes no difference.  Encountering resistance, depending on how the “mangle” works out in any given instance, the Pickering automaton selects a different tactic (it might try to recruit allies, or adjust the theory, or refine the experimental apparatus…) until a path forward opens up.  As with Latour, the specifics of cognitive processes involving an evaluation of the character of situation faced, and concordant choice of tactic, are safely circumvented in favor of the development of a universal language of description for all historical acts.  Automata functioning within this universal language do not automate history so much as free it from philosophical determinism, much as AI liberates computers from a programmed determinism.  Here success seems to come when a Pickering automaton gets lucky in picking a series of tactics; there is no allowance for different levels of talent.

Daston and Galison’s notion of the “epistemic virtue” retains Pickering’s lack of distinction between levels of talent (still no geniuses), but alleviates certain problems with the Pickering automata.  First, the “virtue”, in setting an aesthetic ideal for what science should accomplish reinstates some methodological content in ritual acts—“collective ways of seeing undeniably produce knowledge and therefore qualify as the stuff of epistemology” (369).  Second, Pickering’s model relies on a notion of a self-interested automaton that always tries to advance its own cause.  The Daston-Galison automaton is almost exactly the opposite—racked by a virtuous conscience, it is filled with self doubt that it is not behaving properly scientifically, i.e. objectively.

Daston and Galison spell out a number of different programs automata can run to behave objectively (truth-to-nature, mechanical, trained judgment…), but these seem to have a certain quality of incommensurability.  Opposing notions of objectivity can co-exist in time, but not in personality: the automaton-scientist must be forced to abandon one virtue before using another.  Not being a philosophical being, there is no higher epistemology that allows the Daston-Galison automaton to wear different hats for different tasks, to choose different strategies of representation.  It is therefore rare to move from one version of a virtue to another, which not only gives the epistemic virtues a mesoscopic persistence at the personal level, the virtues are so sticky that they can actually be passed on between individuals and periodized—not in any strict way, of course, but in a general sense that a Victorian will tend to abide by different epistemic virtues than an Enlightenment scientific figure, or someone working in the 20th century.

The resulting picture is startlingly Kuhnian, only without the geniuses.  Scientists working under different paradigms epistemic virtues can never see eye to eye; their encounters with each other have high stakes and are charged with moral overtones.  Epistemic virtues inform the scientific figures’ moral sense of themselves.  When they are forced to abandon the virtues that make them who they are, we can practically see the sweat on their brows as they experience a Kuhnian an existential crisis.  In fact the existential crisis arises out of many of the same circumstances as a Kuhnian crisis.  In the prologue of the book, British physicist Arthur Worthington seems to experience vertigo when suddenly confronted with the anomalies aesthetic irregularities of his milk drop experiments.  In switching from hand-drawn representation of idealized symmetrical splashes to high-speed photography, he does not decide to switch the object of his interest from symmetry to symmetry-breaking, he becomes another kind of scientist with a new aesthetic ideal and a new notion of what it means to be objective, not unlike a Kuhnian revolutionary experiencing a Gestalt shift.

Daston and Galison never stay with one example of scientific practice for very long, preferring to visit only long enough to discuss the representational practice, and thus the sense of scientific self, at play.  The representational practice is just what the automaton does; other models do other things and we must hurry on to another wing of the gallery to see them.  That the manner of representation is a choice made by someone operating within a local epistemological scheme of ideas supporting multiple representational possibilities—that actors might choose to produce an atlas in a different way, but do not do so, because, for their purposes, it would make no sense—never seems to be taken seriously.

Daston and Galison do not present these portrayals of representational practice as a possible reading—they are didactically presented as manifestations of real ideas of objectivity.  If we have the courage to question the sense of certainty surrounding this reading of their Gallery, we might be willing to conclude that this is simply one new interpretation to be added on top of all the others in our ever-proliferating Grand Gallery of Practices.  Or, bolder still, we might conclude that for their characterizations of practice and ideas and their periodizations to cohere, Daston and Galison have amalgamated their notion of historical objectivity with the historiographical idea of the automaton-scientist and have superimposed this system against the record.  If we were to replace the overarching history of virtue with the diverse, fine-grained, complex systems of ideas hidden between the lines in the portraits presented here, rather more satisfying explanations of the representational practices depicted in the gallery might be uncovered.