This study has done both quantitative and qualitative analyses on the diagrams of mechanistic models for cell phenomena, circa 1970s–2005. The results from examining over 3,500 papers across eight journals show that mechanism diagrams play a crucial role in the practice of developing mechanistic explanations for cell biology. This study argues that mechanism diagrams serve as both representation and practice in model-development. Such a role transcends the dichotomy between static visual representations and dynamic research processes. This study borrows the phrase “state of becoming” from cartography to characterise the mechanism diagrams in cell biology. Below introduces the two aspects of this state.
The first aspect is the historicity embedded in biological diagrams. The components of mechanism diagrams not only represent the knowns – including entities, activities, and their relationships – but also represent how these components have come to be the knowns. The process includes continual data-gathering, model-building, and error-correcting etc. Moreover, it involves the inter-field and inter-level interactions between different perspectives for the same phenomena. This is due to the complex nature of biological mechanisms. This study treats mechanism diagrams as a communicative device that acts in the research dynamics. The communication takes place not only between different researchers in the field but also between the same researchers at different stages of model developing. Meanwhile, the development of the representational signs has its history, too. This study imports art-historical method to examine the visual elements in biological diagrams, showing that the visual conventions in mechanism diagrams are not given but have evolutionary processes. Neither the represented nor the representation is fixed. They have been developed and are still open to evolution. Cell biologists are both the author and the viewer with trained interpretation. Such interpretation decodes the meanings of conventional signs, and simultaneously decodes the historicity embedded in the making of these conventions.
The second aspect is the deep involvement of mechanism diagrams in both the defining of research arenas and the intervention in the mechanisms. Both activities extend into the future and cannot be separated from the aforementioned dynamics of communication. Mechanism diagrams engage the user through bringing about new problems and activating the future research. Bechtel (2013) has demonstrated this with diagrams that contain question marks. This survey also includes many examples that visualise the underdetermined parts of the models. The diagrams integrate the knowns and the unknowns for visual reasoning. The unknowns are visualised so that they can become actual in the future. My analysis on visual elements and compositions of mechanism diagrams suggest that cell biologists reason with diagrams while developing the visual languages. Through such reasoning, the mechanisms of interest are constantly defined and redefined, greatly according to the pragmatic purpose to control the mechanisms.
In sum, mechanism diagrams, while appearing as static representations, are constantly in the state of becoming. This is because they embody the research dynamics, and that the making of them is an important part of knowing and controlling what they represent.
Digital data processing has many particularly interesting applications in scientific imaging. Today, as most imaging instruments have become digital — they produce data in numeric form, as lists or matrices of numbers – images can be mathematically reconstructed in 3D from 2D projections, or blur and various artifacts can be corrected for by algorithms. These practices have become so common in scientific imaging that it is now more appropriate to talk about imaging systems, which combine an instrument and a computer, than about imaging instruments that are rarely used in an autonomous way.
The analysis of the practices of image processing has led to very little philosophical investigations. So far, most contributions have dealt separately with instruments (e.g. (Hacking 1981) on the microscope) or with computer models and simulations (Winsberg 2001, Humphreys 2009). In the literature on instruments, there is no reference to the possibility to add computer processing to the raw data produced by the instrument, either because many of the most famous contributions on the topic precede the digital era, or because this possibility has been ignored since. The contributions on models and simulations, which are more recent in general, have focused on the use of computers alone, and particularly on the experimental character of results that have been obtained through the exploration of computer models of phenomena. In this communication, I will consider jointly instruments and computers to study the role that image processing plays in the empirical investigation. More precisely, I will take the point of view of agents (the investigators) and ask in what respect image processing is beneficial for them. Why, for instance, is it desirable to use images that have been obtained less directly and that have been altered by algorithms?
The main thesis regarding the interest to perform image processing is that it renders images less demanding for the investigator who is in charge of interpreting them. In spite of a further sophistication in the production of images, which could lead to more elements to take into account into the interpretation, algorithms of data processing are aimed to facilitate interpretation. In fact, I will argue that they realize a kind of pre-interpretation, because they perform certain tasks that would otherwise be under the investigator's responsibility: subtracting noise or removing artifacts are done mentally during the interpretation when images haven't been processed. As a result, there is an economy of skills and knowledge brought by image processing. The rest of the communication will discuss potential dangers of these new practices, especially with regards to the objectivity of processed images.
In an essay about the atomistic iconography Christoph Lüthy's presents globules as visual symbols for distinct theories of matter. In a similar vein we want to put up for discussion diagrams of cell lineages and the notational systems of modern computer music (e.g., Ligeti, Lutoslawski, Xenakis). Our point of departure is to challenge the conviction that the eye is the most important tool for recognizing patterns and forms in nature. For, the ear also identifies spatial patterns, or the gestalt of a sonification that the composer depicts in a specific notational system, creating audible sounds by inaudible structures. To foster our argument, we will compare cell lineage diagrams around 1900 to the graphic notational systems of music in the 1960s. For ordering and classifying dividing cells, the cytologists created ‘data displays’ that resemble the graphic systems of modern music. We will survey Carl Naegeli's approach to figure cell-formation of peat mosses, Wilhelm Hofmeister's attempt to include geometric figures into Naegeli's arithmetic series and Maupas' diagram of Paramecium. When comparing their diagrams, it looks as if the biologists had developed sophisticated graphic methods for representing emergent shapes nearly 100 years before the mid-20th century appearance of graphical music notation. The hypothesis therefore is that the cell diagrams (e.g., Hofmeister, Maupas) resonate with the computational notational systems of algorithmic music composition. For, both notational systems, or data displays, provide a structurally similar form of reasoning, regardless of whether we observe an object (cell) or perceive a sense data (sound). For example, processes of cell divisions partly resemble the theoretical approaches of computer musicians tracing the sound as processes of fraction, interval and spinning turns of semi-development. Moreover, the combination of these notational practices also reminds of Helmholtz' work on the ‘acoustic image’ in the 1860s and his transdisciplinary approach of ‘inclusive research’. A preliminary argument we want to explore here is that the shape of sound in sensu Helmholtz represents the acoustic equivalent of a dynamic motion that is fixed in space by notation and coding (Boulez). Our objectives are (1) to entangle sounds (sonification) and cells, and (2) to disclose the epistemic, aesthetic, and methodic similarity, or difference between these gestalten, which come into being either by software, or by the narratives of wet experimenting in the laboratory.