Computational techniques and methods are ubiquitous in biology. One may observe that this situation is common to most contemporary scientific fields. Still, the relation between biology and computation in somehow tighter than the relation between computation and other sciences (with the possible exception of cognitive sciences and AI). As a matter of fact, it is not just the case that virtually all biological systems and processes (genetic, developmental, evolutionary, ecological etc.) can be understood, analysed and simulated computationally. Also, the other way round, computational techniques are frequently inspired in biological knowledge (i.e., genetic algorithms, evolutionary computation, etc.). Far from being working tools just for computer scientists, many biologically-inspired algorithms have been employed, in their turn, to model and understand biological phenomena. Finally, and perhaps most notably, computation in biology is not only conceived as a mean to predict the behaviour of a system: biological systems themselves are often interpreted as capable of performing computations in their own right. In this sense, computation provides (similarly, again, to what happens in the disciplines which study mind and cognition) a fundamental tool for theoretical projects aimed to reproduce and control living processes (like Artificial Life or synthetic biology).
My goal in this paper is to offer a broad classification of different approaches to computation in biology and highlight some conceptual issues related to each of them. Some of these issues have already being calling the attention of biologists and philosophers for a long time. Others have passed somehow unnoticed. At any rate, the discussion over these issues have been customarily held within the boundaries of specific philosophical fields (like the philosophy of biology, the general philosophy of science or the philosophy of computation). What is still missing is a broader and systematic account of the relation between biology and computation. I believe that there is a potentially rewarding pay-off, in terms of the intelligibility of certain conceptual problems, in any attempt to rethink traditional debates within new frameworks. It is in this spirit — that is, with the hope of helping to develop further debates in a more focused way — that I try to outline a sort of common agenda for what we may call a philosophy of biological computation (or computing).
In order to attain this goal, I shall proceed as follow. First, I shall introduce two general distinctions between different uses of computation in science. The first distinction concerns the choice of employing computation as a mean to model phenomena or, alternatively, to explain them. In some cases, computational methods are just tools to solve problems that could not be solved analytically with more traditional mathematical techniques. However, occasionally, scientists conceive a natural system as able to perform computations or behave algorithmically and, accordingly, they try to explain it as a computational system. The second distinction is about the aspects of computational science that are considered as pertinent to, respectively, model or explain natural phenomena. On the one hand, we have research programmes that rely on a computational approach because computer sciences provide them with algorithmic resources and specific mathematical tools. On the other hand, there are areas of natural sciences that interact with computer sciences because these are, at least in part, engineering sciences. In this sense, they offer an important source of ideas to think and build system architectures.
Neither distinction should be considered as completely neat since, in practice, many current theoretical projects adopt hybrids positions. Still, I think that they are helpful insofar as they allow to provide a classification of different approaches to computation in biology. From the four possible combinations between the two distinctions, we obtain four broad categories: mathematical modelling, mathematical explanations, engineering explanations, and engineering modelling. Each category denotes a more or less definite class of research programmes in biology, aimed to stress specific aspects of the interaction between biology and computation. Thus, mathematical modelling includes those theoretical projects (like bioinformatics) that employ computational concepts instrumentally, to build reliable models of biological processes and to test hypotheses statistically. The category “mathematical explanations” includes those proposals, within Artificial Life and system biology, aimed to explain biological processes as abstract computational processes. The approaches which intend to provide engineering explanations of biological systems (like many research programmes within synthetic biology) also seek a computational understanding of biological systems but, instead of focusing on the algorithmic features of living processes, insist on the architectural similarities between organisms and computers. Finally, the label “engineering modelling” denotes those approaches to biology (such as bio-robotics and parts of synthetic biology) which aim to manipulate and modify biological systems and processes through computer engineering.
To conclude my talk, I shall present a short list of conceptual challenges related to the theoretical projects previously presented. More specifically, I shall stress the importance of clarifying the relation between abstract and concrete properties in computational systems, in order to make sense of the multiple realisability of biological properties on non-biological supports. As I shall argue, both mathematical and engineering explanations vitally depend on the multiple realisability of biological properties. Without clear criteria about when a biological property is realised, many claims of the supporters of Artificial Life or synthetic biology remains metaphorical. Secondly, I shall emphasise the importance of clarifying some general ontological commitments in the debate over what we can reasonably infer, from the results obtained through the employment of computational techniques, about the nature of biological processes and systems.