The opening-paragraph of Kulvicki (2010) summarises many philosophers’ interests in the topic as follows:

There are two problems with understanding how scientists use representations. On the one hand, what makes graphs, diagrams, pictures, images, descriptions, lists, and so on about other things? What is it about our intentions toward them, their intrinsic features, and the relations between them that conspires to make these artefacts represent other things, and what exactly do they represent when they succeed in doing so? (…) On the other hand, we have the problem of what makes certain kinds of scientific representations different from one another and thus well suited to this or that end. (Kulvicki 2010: 295)

As this comment highlights, the primary philosophical problem posed by scientific representations turns on the question “what turns something into a scientific representation of something else?” (Frigg & Nguyen 2016), which is often read as a demand to state necessary and sufficient conditions for something to counts as a representation. Such approaches can help us to identify what it takes for a visual artefact to be epistemically successful (e.g. Bolinska 2016), but can hardly be used to inform the formulation of concrete design principles.

Howard Wainer’s foreword to Jacques Bertin’s “Semiology of Graphics”, in which he compares Bertin’s contribution to the earlier achievements of William Playfair, draws our attention to a similar contrast:

Playfair understood the power of his invention to convey information; or, if poorly done, to misinform: “As the purpose of the following Charts is to convey information in a distinct and easy manner, (…), the chief merit is that they can have is truth and accuracy (…)” In the course of his great work Playfair illustrates good graphic practice but does not explain why the specific structures of his graphic forms and formats work; nor does he provide any theory to guide future constructors of graphics. (Bertin 1983: ix)

Inferential conceptions of visualisation, as informed by similar proposals to analyse scientific representation in terms of surrogative inference (Suarez 2004), can be extended to make room for the design-problem of visualisation by distinguishing between the object-level and the meta-level problem of visualisation. I propose to characterise this second problem as:

The meta-level problem of choosing the visual representation from a given range of options that makes answering a given range of object-level questions efficient and effective (correct and complete).

I will argue in favour of this proposal along two lines. First, I will relate it to Keith Stenning’s proposal that people reason by finding a representation in which the problem is trivial to solve (Stenning 2002), that is: by engaging in meta-level reasoning. Second, I will show that several important theoretical practices within the visualisation sciences, like the development of taxonomies (Card & Mackinlay 1997) or the design of specification-languages (Wilkinson 2005), can be understood as forms of meta-level reasoning. The final conclusion of this presentation is that the meta-level problem of visualisation is an ampliative inference problem that is best understood as a characterisation-problem. The goal is to find, within a given (but possibly incomplete) design-space, the best representation of a given collection of data-objects.


  • Bertin, J. (1983). Semiology of Graphics. Diagrams, Networks, Maps. Madison, Wisconsin: University of Wisconsin press.
  • Bolinska, A. (2016). Successful visual epistemic representation. Studies in History and Philosophy of Science Part A, 56, 153–160.
  • Frigg, R., & Nguyen, J. (2016). Scientific Representation. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Winter 201). Metaphysics Research Lab, Stanford University.
  • Kulvicki, J. (2010). Knowing with Images: Medium and Message*. Philosophy of Science, 77(2), 295–313.
  • Card, S. K., & Mackinlay, J. (1997). The structure of the information visualization design space. In Information Visualization, 1997. Proceedings., IEEE Symposium on (pp. 92–99).
  • Stenning, K. (2002). Seeing reason: Image and language in learning to think. Oxford University Press.
  • Suárez, M. (2004). An inferential conception of scientific representation. Philosophy of Science, 71(5), 767–779.
  • Wilkinson, L. (2005). The Grammar of Graphics. Dordrecht: Springer-Verlag New York Inc.