In this post I foolishly avoided tackling the meaning of the word ‘representation’, a Philosopher of Science’s favourite, and consequently neglected to acknowledge that all scientific images are representative. For a more thorough discussion of representation see Teaching Representation.
It doesn’t happen often, but every now and again I find I want to talk about something, an object or a concept, which either has no name, or else is difficult to give a name to. The feeling can be both unsettling and invigorating. Often within research groups, scientists develop their own words for things that they discuss frequently. These might be abbreviations, where a verbal interpretation of the abbreviation is found to roll of the tongue more easily, such as mir-na rather than M I R N A or micro R N A. A common type of project in biochemistry is the identification of an unknown molecule or substance. The unknown might be called ‘Factor X’ locally until its identity is established. During my PhD I discovered that my lipid transfer protein, RdgBβ, bound and transferred a lipid that didn’t share characteristics with any of the lipids known to be transferred by our family of proteins (the phosphatidylinositol transfer proteins (PITPs), which transfer phosphatidylinositol (PI) and phosphatidylcholine (PC)). The lipid bound by RdgBβ was called the ‘mystery lipid’ for a long time in our lab until we identified that it was phosphatidic acid (PA). In rare cases, a scientist might discover a new factor entirely; recently, one of my colleagues at the University of Bristol discovered a new microRNA, and so had the pleasure (or burden) of naming it for the wider scientific community.
Molecular biologists write a lot, but we also draw for a variety of purposes and take photographs (using ‘photograph’ in the broadest sense of the word – see Cell Microscopy Images; Abstract Art, and below). These types of image can be further defined as follows:
1. Representative: photographs and microscope images of cells or tissues (both ‘photographs’, although the latter might use electrons, as in electron microscopy, or other wavelengths in the electromagnetic spectrum other than visible light, to image biological specimens). This category could also include diffraction patterns generated during the process of X-ray crystallography, and possibly anatomical drawings, and images of DNA gels and western blots. This category includes the raw data, and could/should be described as ‘objective’: it has not been processed or manipulated. I’m not sure whether to draw a line before numerical data – tables of numbers. If you hold the chart far enough from your eyes, the numbers become nothing more than marks on a page, arranged in space – an image.
2. Graphical: (mathematical) graphs and other forms of analysing and displaying data. We now use computers to generate graphical forms of data, although at one time this work would have been carried out by hand, on graph paper using a pencil and ruler. Graphical images also include those derived from molecular visualisation programs, such as an image of an X-ray crystal structure of a protein.
3. Abstract, illustrative or diagrammatic (I regularly refer to these types of images as ‘schematic‘): this level of image responds to the previous type of image. It likely combines the knowledge gleaned from multiple experiments, often by many researchers. This type of image is commonly mechanistic in molecular biology, describing the interaction between, and function of, multiple cellular components. Drawings of this type include those describing a series of enzymatic reactions, such as the MAP kinase cascade, glycolysis or the Krebs (citric acid or tricarboxylic acid (TCA)) cycle.
Laura Perini is an Assistant Professor of Philosophy at Pomona College in California, and has written extensively about the forms of representation used by scientists. She also has grouped images produced by biologists into three categories, although whereas my categorisation is derived from the source of the information, hers derives from the amount of information contained within the image or diagram, described as an image’s ‘repleteness’, or how well-filled with information an image is. Perini’s classifications are given below. It should be noted that this analysis chiefly refers to figures in biology textbooks.
1. Replete pictorial representations: these include ‘naturalistic’ drawings or paintings, as well as photographs and other images produced through detection mechanisms, such as electron microscopy (broadly the same as my ‘Representative’, category 1, above). These types of images are packed with information, and show specimens as close to reality as possible.
2. Schematic diagrams: these diagrams contain a low relative repleteness and a maximum of two levels of organisation, e.g. whole organism and tissue levels, or tissue and cellular levels. Schematic diagrams include drawings of the structure of a cell or organelle. They are generic in that they do not give precise details or the spatial representation of component parts. Perini gives the example of the rough endoplasmic reticulum, where this type of diagram shows ribosomes as black dots and provides information as to their approximate distribution across the organelle rather than their exact position.
3. Compositional diagrams: include chemical diagrams, electrical circuit diagrams, and some diagrams of biological models, such as the Krebs cycle. These types of diagram are composed of discrete, visible elements that function as labels: atomic characters, arrows, lines, and other shapes. For Perini, compositional diagrams describe the relationship between system components in space. (I would argue that diagrams of the Krebs cycle (see below) combine spatial and temporal relationships, and in terms of the spatial distribution, are actually less precise than schematic diagrams of cell ultrastructure, for example.)
Interestingly, graphs of biological data have been omitted completely from this classification of biological images. Although textbooks rarely reproduce results from the original experiments, they do give idealised graphs, for example those describing rates of enzyme activity.
The third category from my classification system could be applied to both Perini’s schematic and compositional diagrams categories, yet it also differs from the examples given for these categories. In a sense, they are simplified or generic, often using a single example of a component in a system, where in reality there would be many of that particular component, protein, enzyme, compound, etc. However, the diagrams to which I refer do give relative spatial information. The diagram below describes a possible function of RdgBβ. This diagram contains a low relative repleteness and generalised distribution of forms, yet also carries discrete, visible elements including arrows, lines, and other shapes. Whereas Perini has argued that the shapes of elements in compositional diagrams are not representative of the shapes of the molecules themselves, those in the diagram below are: the RdgBβ element describes the structure of this protein, containing the location of the lipid in the lipid-binding pocket and long C-terminal tail (discussed in the Terry Winters: Process and Generative Art post); the AT1 receptor is a seven transmembrane receptor, and here it is shown with seven transmembrane domains. Further examples of these types of diagrams, which I will now refer to as ‘Mechanistic Diagrams’, exhibit more than two levels of organisation, from the atomic through to the cellular, or sometimes tissue, level. This was the name I was looking for.
Perini, L., ‘Form and Function: A Semiotic Analysis of Figures in Biology Textbooks.’ Book chapter in, Anderson, N.A., and Dietrich, M.R, ‘The Educated Eye: Visual Culture and Pedagogy in the Life Sciences,’ Dartmouth College Press, 2012, Chapter 10.