Peter Galison: Objectivity

“Objectivity: the Limits of Scientific Sight”

Peter Galison
Pellegrino University Professor
Harvard University

When scientific objectivity became a goal in the early 19th century it was by no means obviously something to be desired.  Natural philosophers had to invert the old epistemic virtues that involved finding ideal forms that lay behind the variations of this or that individual.  Where genius was, plain-sight observation came to dominate.  I will here track how the images and image-making technologies of scientific atlases helped define the modern scientific category of mechanical objectivity–and the new quieted and transparent scientific self that accompanied it.  The fate of objectivity kept turning: twentieth century scientists questioned image-based, mechanical objectivity; they demanded more interpretation and modification of images than mechanical objectivity ever allowed.  With that shift toward a “trained eye” came a new view of the right scientific self, one that explicitly made use of intuition, expertise, and the unconscious.  Now, in the early twenty-first century new kinds of scientific pictures (such as nanomanipulated images) are demanding quite unexpected ways of being– scientist-selves perched uneasily between scientific, engineering, and entrepreneurial forms of sight. Representation begins to cede its place to presentation.

Discussing Galison: regarder les sciences par les vertus épistémiques

Margherita introduced some of the issues that are discussed in Peter Galison’s book, Objectivity.

Wilson Bentley's snowflake (1902)

To download her slides (in French), click here.

Third invited speaker: Barbara Tversky

2 April 2009

Third special session: Barbara Tversky (Stanford University)

Morning session (in collaboration with the general seminar): Telling Stories in Pictures

Abstract:

Visual communication is ancient, preceding written language.  A selection of historic,  developmental, and experimental examples show that graphics use space and the elements in it to convey meanings.  Notably, elements such as lines, boxes, blobs, and arrows have context-dependent meanings that are readily produced and readily understood.  Visual narratives need to determine spatial perspective and scale, to cut up time, to link successive frames, and to convey visually-perceptible and more abstract meanings.  Research on cognitive principles for diagram design and on analyses of comics and graphic novels will illuminate how visual narratives are constructed.

Afternoon informal session: more on Comics

For her papers on diagrams, see here.

Questions for Barbara Tversky

The group came up with the following questions for a discussion, in the following 7 topics:

1. Graphs, Concepts and reasoning

2. The relationship between conceptual system and perception

3. Terminological clarification

4. Questions at the interface between research on graphs and theories of concepts and  reasoning

5/6. Constructive perception and animations

7. Recency of diagrams

Preparing for Tversky 2: Semantics, Syntax and Pragmatics of Graphics

Discussion on Tversky’s paper Semantics, Syntax and Pragmatics of Graphics.

I discussed Tversky’s approach. In my interpretation, Tversky succeeded in ‘dissolving’ some of the traditional oppositions that have been an obstacle to discuss the nature of diagrams, such as the oppositions vision vs. language, geometric/Gestalt properties vs. interpretation, internal vs. external representations, objects in the world vs. (mental and public) representations.

Nevertheless, she is maybe dissolving too much of what a theory of diagrams could be, as Sandro discussed last time. In fact, several issued remain obscure. Again, it is not clear what kind of theory her theory of concepts could be, if, as she seems to suggest, concepts can be ‘depicted’. Moreover, the use of the term ‘natural’ is not completely clear: if a skill is ‘natural’, would it mean that it is spontaneous, very easy to apply, or what else? Some examples are problematic. In the map case, her idea is that there is variability in them but the same underlying structure. It is a matter of study to determine what this underlying structure is. Also the example of icons and the reference to resemblance is problematic, since resemblance has been discussed (and most of the time refuted) in the literature as a notion which is able to explain icons and the way they work.

Preparing for Tversky: Spatial schemas in depiction

Discussion on Tversky’s paper Spatial schemas in depiction.

Sandro criticized Tversky’s approach, since according to him it is not clear what her general framework is.

In fact, according to Sandro, any explanation of the functioning of graphs is committed with a theory of concepts and a theory of reasoning. This commitment should be explicit. I.e. the researcher working on graphs should explicitly aim at using graphs to produce arguments, testable hypothesis, case studies, etc. in favour of one or another theory of concepts and reasoning.

Moreover, a work of terminological clarification is needed. Terms such as resemblance, analogies, natural correspondence, mapping, … have a strong intuitive impact, but we do not know if there are natural kinds behind them. Graphs could be a good study case to work on technical definitions of these terms.

For the complete power point, click here.

Second invited speaker: Patrick Maynard

6 March 2009

Second special session: Patrick Maynard (University of Western Ontario), the author of Drawing Distinctions, The varieties of graphic expression (2005) (Cornell University Press).

Morning session: Diagram Imagination or Darwin’s diagrams

Patrick’s paper

Afternoon informal session

Preparing for Maynard: session TWO

Sandro Pignocchi, ORGANIZING THE TOOL BOX

Sandro discussed Maynard’s idea of a tool-box of different skills, concepts and disciplines to talk about drawings. He moved from the two following hypotheses:

Hyp 1 The perceptual efficiency of each tool comes from the way it activates some modules of our perceptual system.

Hyp 2 A modular approach helps to organize and operationalize the tool-box approach.

He proposes to distinguish among 4 different sets of tools, in order to organize the tool box.

1. Pre-recognition tools
(Tools which operate before semantic categorization)

2. Recognition tools
(Tools which operate during semantic categorization)

3. Motor tools
(Tools which provide a perceptual access to the gesture of the draftsman and, more generally, to the intentional process of production of the drawing)

4. Pseudo-conventional tools?

To download Sandro’s power point, click here.

Preparing for Maynard: session ONE

The session was dedicated to the presentations of the topics discussed by Patrick Maynard in the first three chapters of his Drawing distinctions.

More in general, the subject of the book is drawing, from the perspective of an inclusive conception of art. Drawing is art ‘at full stretch’, as poetry is language at full stretch. Moreover, drawing is a ‘worldwide’ activity not a technological invention: some of its basic techniques have changed so little since prehistory. It employs not only different means (there are many ways to draw), but it also exploits different meanings. To study drawing, the book assembles tools from a variety of contexts.

The key features of the first three chapters of the book are the discussion of its methodology (tool-kit approach and dialectics), the definition of constructional drawings, the topological vs. perspectival reading of a design, the human appeal to patterns, which are most of the times no rapresentational, such as symmetry.

To download (Valeria’s) presentation of the first three chapters of the book, click here.

First invited speaker: John Kulvicki

7 November 2008

First special session: John Kulvicki (Dartmouth College)

Morning session (in collaboration with the general seminar at the Institut Nicod): Heavenly sight and the nature of seeing-in

Abstract:

Richard Wollheim famously understood pictures to be distinct from other kinds of representation in virtue of eliciting a special kind of experience: seeing-in.  What a picture depicts is determined in large part by what appropriate observers can see in it.  Many agree that pictures often evoke experiences of seeing something in a marked surface, even if they do not assign seeing-in such a central place in their theories of depiction or characterize it in exactly the way Wollheim does.  This paper proposes a new way to understand seeing-in that is motivated by a curious renaissance discussion of vision after death: heavenly sight (Baxandall 1988, 104-105).

The first section introduces seeing-in as Wollheim understands it, while the second section introduces heavenly sight.  Heavenly sight is impossible, and it might seem unimaginable were it not for seeing-in.  In fact, seeing-in is best understood as an approximation to heavenly sight.  This conflicts with the way in which Wollheim explicated seeing-in, however, so the rest of the paper is devoted to arguing in favor of the present view.  Section three unpacks and then rejects the reasons for which Wollheim would resist the heavenly sight proposal.  Section four argues that experiences akin to heavenly sight are indeed possible.  And section five concludes by showing that this proposal does an excellent job accounting for seeing-in.

Wollheim correctly characterized the phenomenology of seeing-in, but he overlooked a promising way of explaining the phenomenon in detail.  Most discuss seeing-in as a tool for understanding what makes depiction a distinctive kind of representation.  The focus here, by contrast, is on the nature of this special kind of experience.  The exact relation such experiences bear to a correct account of pictorial representation is not of central concern.

Useful references:

Baxandall, M. 1988. Painting and Experience in Fifteenth Century Italy, 2nd ed. Oxford: Oxford UP.

Wollheim, R. 1974. On Art and the Mind. Cambridge, MA: Harvard UP.
——. 1980. Art and its Objects, 2nd ed. Cambridge: Cambridge UP.
——. 1987. Painting as an Art. Princeton: Princeton UP.
——. 1998. On pictorial representation. Journal of Aesthetics and Art Criticism 56(3): 217-226.
——. 2003a. In defense of seeing-in. In Hecht, et al.: 3-16.
——. 2003b. What makes representational painting truly visual? Proc. of the Aristotelian Society, Supp.: 131-147.

Afternoon session: discussion on Pictorial Diversity

Abstract:

There are undeniably many ways of depicting things, but it is unclear what this pictorial diversity is.  What is a way of depicting something, and how many ways of depicting things are there?  There are many individual pictures, of course, and many of them differ from one another.  They have different surface features and they represent different things, but all of this is unsurprising.  No one thinks that all the world’s pictures depict one and the same thing and no one denies that pictures depicting different things can differ quite a bit in their surface features.  Pictorial diversity starts to seem interesting and confusing when one tries to fix a picture’s content while varying its surface features, or vice versa.  Could this picture, ostensibly one of van Gogh’s bedroom portraits, depict instead Gauguin’s aunt?  Could Fra Angelico’s Annunciation and Bosch’s Ship of Fools depict exactly the same thing?  Sensible individuals, as well as most philosophers, would answer each of the above questions with a resounding “No!”  But it is obscure whether such denials are directed at the particular examples or at the more general claims that these questions interrogate.  Most would agree that some pictures admit of multiple interpretations.  And it will become clear that many agree there are pictures with rather distinct surface features that nevertheless have exactly the same contents.  This paper seeks to explain what is at issue in discussing pictorial diversity and to introduce two tools for understanding it.  The value of these tools will become clear as they are used.

After section 1 clears up some preliminary issues, section 2 argues that while there are various means of making pictures, viewers do not use a complementary variety of schemes for interpreting them.  So, interpretative diversity is outmatched by the myriad means for making pictures.  Section 3 shows that the distinct interpretative schemes employed by observers meet an interesting constraint: they compete neither syntactically nor semantically with one another.  Interpretations of non-pictorial representations, like diagrams and linguistic representations, do not obey this constraint.  Section 4 considers an objection to the previous section’s claims based on Ernst Gombrich’s discussion of landscape painting, and section 5 considers a related objection based on Heinrich Wölfflin’s distinction between linear and painterly styles.  Section 6 discusses related worries about the depiction of space.  The overall conclusion is that people have a very powerful tendency to avoid competing interpretative schemes for pictures.  Sections 7 and 8 consider how one might explain this surprising pattern and suggest that while no accounts of pictorial representation predict or preclude such a pattern, some make room for it more naturally than others.

The following does not take any particular account of depiction as its starting point.  There are indefinitely many ways in which one could pair representations with contents.  Some systematic ways of doing so are pictorial and some are not; some are diagrammatic, some linguistic, and so on.  In order to avoid committing to a particular account of depiction from the outset, it will be important to focus on the practice of interpreting representations that, intuitively at least, are pictorial, and sometimes comparing these to representations that are not.  Just as an account of depiction overall must begin in part with intuitions concerning which representations are pictures, so this discussion begins by focusing on the respects in which, at least intuitively, systems of representation can differ from one another and nevertheless all be pictorial.  The upshot of this investigation will be an understanding of pictorial diversity that any account of depiction must respect, and if possible, explain.