Book Club: Objectivity, Pt. 1

Lorraine Daston and Peter Galison argue in their recent book Objectivity that “objectivity” is not a transcendental concept: it “has a history”.  It therefore must have come into existence at a certain point in time.  They further argue that the character of objectivity has both transformed and fractured with time, and that the transformation, fracturing, and collection of different notions of objectivity can be periodized.  To demonstrate their case, they analyze scientific imagery and its production through time, as surveyed by assembling a gallery of scientific “atlases” (collections of images meant to convey a body of scientific knowledge).

Objectivity is, I think, an important work of historiography that addresses head-on problems associated with the historiography of the gallery of practices: how to write analytical long-term (“mesoscopic”) histories of scientific practice, how to write histories that do not depend on a strict narrative of causality, and how to write about science as a cultural and intellectual activity without questioning the ultimate validity of scientific work.  In addition to their scattered commentary toward these points throughout the book, see a distillation of their position last year in Victorian Studies.

Reviews suggest that the book will be a widely-read touchstone.  I believe this presents a danger to a historiography that is not sufficiently reflective.  Rather than critically reflect on the book’s strategies, there is a danger that historians will now be inspired to “go fishing” for instances of “truth-to-nature” or “mechanical objectivity” in the histories that they themselves study.  Those who do so may deploy Daston and Galison’s methodological insights directly, or may question or problematize their periodizations by finding contrary examples.  But I believe a more productive exercise is questioning the methodology in toto as an approach to the historiographical problems the book explicitly seeks to address.

The book is certainly broadly-researched and its presentation is extremely erudite, and it contains useful insights.  But I disagree with Daston and Galison as to what these insights are.  I believe the presentation in Objectivity is grounded on a few interrelated assumptions.  First, the practice of representation constitutes a fundamental act of science, a presentation of what scientists believe the scientific object must look like as seen through the scientific eye.  Second, there is, therefore, a one-to-one-to-one correlation between the act of representing an object, the concept of objectivity being deployed, and the characteristics of the “scientific self”.  Third, given this 1:1:1 correlation, practices and ideas  exist in tandem relationships, i.e. proper “practice”, in this case representation, is governed by an associated “idea” (in their terminology, “epistemic virtue”), in this case objectivity.

For my part, I do not believe objectivity “has a history”, though I agree that objectivity is not transcendental.  In their book, representation is not a coherently defined practice; it is a variety of practices that serve a variety of socio-epistemological functions.  These functional practices exist among other functional practices, which, together, constitute a matrix of practices; the functional deployment of these practices is governed by a set of ideas.  Thus, practices—even superficially similar practices—serving different functions (e.g. representation as presentation of evidence vs. representation as pedagogy) within historical sets of ideas cannot be analyzed to reveal varying historical attitudes toward the same idea (i.e. objectivity).  Thus Objectivity is not a history of objectivity; it is a gallery of partially characterized representational strategies that exist in history, but that have not been given a clear historical-philosophical relationship to each other.

To be more concrete and clear, I would like to try and define at least part of the matrix of practices presented within the book’s gallery.

First, the problem of “objectivity”, here defined as the effort to control or escape the scientific self to represent the world, is actually at least three separate (albeit interrelated) problems:

1. Avoiding mental prejudice in the selection between possible interpretations of evidence.
2. Overcoming physiological or cognitive limitations in observing and representing the world.
3. Representing abstract concepts in a way that allows others to perceive that concept.

The problems are interrelated, for example, in that physiological limitations in the ability to see clearly allows space for mental prejudice to intrude in a selection between interpretations of what is being seen (if Mars is fuzzy in your telescope, you might pick out canals on its surface).  Nevertheless, these issues are distinct, as it is possible for one to be relevant without the other being so.

The object of representation is likewise divided into interrelated concepts.  The image can be of:

1. What has been seen
2. What has been learned
3. What one is arguing
4. The phenomenon that one wants others to see
5. Phenomena that might prevent others from seeing that phenomenon.

So, I might create a different image of a plant depending on if I want you to see (1) the plant that is in my greenhouse, (2) what is typical of all plants of a certain species, (3) what distinguishes one species from another species, (4) pathologies that sometimes appear in plants of a certain species, or (5) incidental marks you might mistake as that pathology.  If I want to use a photograph, a likelife drawing, an enhanced photograph, or a diagram, and to decide on what text and data must accompany the image, or if I can’t make an image and I need you to actually visit my greenhouse (say if a plant is distinguished by a strange smell) will all depend on what I need to communicate and what skills and technology are available to me.

The alert reader will further note that even though similar drawings or photographs of the same “plant” are being shown, the relevant object of the representation is different in each case: (1 and 2) the plant, (3) distinguishing features of the plant, (4) the pathology, (5) the false pathology.  Actually, the relevant aspect of a representation might also be categorized:

1. Typical examples of a type
2. Distinguishing or relevant features of a type
3. Typical appearances of a type in situ
4. Typical variations on a type
5. The appearance of a single instance of a possible type (the tissue under my microscope)
6. Contested features of a single instance of a possible type (the cellular structure of the tissue under my microscope)
7. Different appearances of a single object (the planet Mars in my telescope vs. another telescope)
8. Unexpected features or appearances of a type (e.g. unexpected asymmetries of snowflakes)

Concordant with relevant aspects is the question of intended audience:

1. Those benefiting from a consolidation of taxonomical gains (amateurs, or others requiring observational guidance to give names to what will be seen in field or lab)
2. Students requiring ersatz experience to identify species in situ (guides to appearance or variations on type).
3. Possible opponents requiring evidence concerning contested points.
4. Other experts to whom an appeal for explanation of unexpected or difficult-to-interpret observations might be made.

I do not mean to provide here an exhaustive socio-epistemological problematic of representation.  Nevertheless, it seems likely that many of these factors ought, in some sense, to be taken into account when providing a representation, and that, therefore, any given instance of representation cannot be taken to be reflective of what it means to depict something scientifically or objectively, but that, rather, a specific representational strategy has been chosen for a specific purpose as part of a larger socio-epistemological project.

These strategies may be expected to vary from science to science or even sub-issue to sub-issue within a science.  And, of course, these practices may differ between different eras in the same science or sub-issue.  For example, changes in the technology of a science should serve to move functions inside and outside the realm of published representation.  Where once it might have been necessary to bring students to a central museum or experimental apparatus; reduced publishing and book costs, or the advent of photographic imagery could begin to serve as a replacement for direct experience, where it could not before.  This would mean, for example, that “trained judgment” was always a notion of objectivity in many sciences, but that it only moved into “atlases” only once the technology could bear the burden.

What is really strange to me is that Daston and Galison absolutely recognize the fact that different atlases serve different functions with different audiences; that “trained judgment” serves a pedagogical purpose, where “mechanical objectivity” serves as direct evidence of what is seen.  Yet, they seem to take this to mean that pedagogy is constitutive of the scientific self not only in some sciences, but in some time periods across different sciences, where the trained eye is apparently altogether non-constitutive of what it means to practice science in not only individual sciences, but entire time periods adhering to the virtue of mechanical objectivity.

Their argument only makes sense to me by accepting that the act of representation adequately reflects a coherent portrait of what it means to portray scientifically.  While Daston and Galison allow that objectivity is but one epistemic virtue among many that can apparently be read from other kinds of source material, this seems to imply that if one were to make a study of these other kinds of source material, it would be possible to garner equally self-contained impressions of what it means to be scientific.  The notion that epistemic virtues as tandem practices-ideas can be abstracted and identified flowing through history renders a tacit epistemology that assumes epistemic virtues are normalized vectors that combine linearly (to use mathematical terminology) to produce knowledge.

I disagree fundamentally.  I believe that practices exist in a complex matrix of practices, and that the deployment of these practices is governed by an intertwined set of socio-epistemological ideas, no section of which can be abstracted into a concept with its own coherent and self-standing history.  For the act of representation to be understood adequately, associated practices and argumentative ideas must be taken into account, which requires the historian to look beyond the act of representation to the particulars of the socio-intellecutal problems that the image is intended to resolve.  That Daston and Galison can arrive at a coherent reading of a gallery of practices across sciences and within large blocks of time absent such an account suggests to me an unduly biased reading of the historical record.

I am not inclined to accept the historical reality of the notions of objectivity they argue for, though I believe these notions might be put to other purposes as socio-philosophical characterizations of different representational strategies.  More on this possibility in the future.

Go to Part 2a.