Computer Models and Philosophy

The use of computers to model complex phenomena – particularly evolutionary, social, and psychological phenomena – is opening important new areas in the Philosophy of Science, as well as impacting on the existing theory of scientific models. Such models also provide new and exciting opportunities for exploring deep and fascinating topics that are of philosophical interest in themselves, through Computer-Enhanced Thought Experiments. Related to this is the use of computer models to facilitate understanding, or to act as "Intuition Pumps". Even where computer models are of doubtful realism, they can still serve a potentially valuable purpose as Existence Proofs, or counterexamples to general claims. All these uses are relatively uncontroversial, but at the other end of the spectrum are speculations about the possibility of Artificial Life.

Computer Models in the Physical Sciences

Computer models are used routinely in the Physical Sciences, necessitated by the sheer volume of calculation required, and the often complex connections between theory and observation. Traditionally the computer has tended to be a number-cruncher, solving sets of equations that model the phenomena in question at a high level of abstraction (as for example in statistical mechanics). More recently, however, the incredible growth of computer power has allowed implementation of models which represent large numbers of entities individually rather than just collectively, as in The Millennium Simulation Project which modelled more than 10 billion particles. This enables the modelling of phenomena which cannot adequately be treated statistically, because the behaviour of individual entities – even within a large collective – has a disproportionate impact. A nice human-scale example is given by the modelling of traffic flow, as in this Traffic Simulator Applet by Cay Horstmann (San Jose State University).

Agent-Based Models

A further move in the same direction is to model individual entities that have the potential for their own distinctive behaviour. Each entity or "agent" is stored as an "object" within the program, each with its own individual "state" (typically specified by the values given to a range of integer variables). Again this has become possible only thanks to the rapid development of computers, which are now sufficiently large and fast to enable models of this kind to be played out with thousands of agents, each interacting with others sequentially, possibly millions of times (e.g. either randomly or based on their geographical location and movements). Use of object-oriented programming languages (such as C++, Delphi, and Java) has greatly simplified their implementation. And more recently, such Agent-Based Modelling has become practicable for non-specialist programmers, through the development of tools such as NetLogo, which enable them relatively easily to develop models of a power that would have been unthinkable even for expert programmers a couple of decades ago.

Computer Models in the Biological Sciences

The Biological Sciences are characterised by complex interactions between systems which are themselves highly complex. Some of these interactions – between proteins, cells, organs etc. – occur within individual animals or plants, their study falling into the fast-developing field of Systems Biology, which has been transformed by the use of computer models. Meanwhile interactions between organisms – as in the study of Evolution or Behavioural Ecology – are now widely investigated using Agent-Based Models. One of the most famous of these, Craig Reynolds "Boids", is pictured above. It served to demonstrate that "flocking" behaviour could perfectly well be explained on the hypothesis that birds are acting without any central coordination, according to three simple and locally-applied rules. As such, it provided an important existence proof of a potential type of explanation, and opened a new line of empirical investigation.

Computer Models in the Behavioural Sciences

Agent-Based Models likewise provide a very natural medium for investigation of the large-scale interactions between humans, as in the Social Sciences and Economics. It has been prominently suggested that such investigations amount to a new paradigm of Generative Science, whereby models are confirmed not through successful prediction, but instead by successful generation of already-observed phenomena from simple (and plausible) foundations.

An interdisciplinary field of particular interest to philosophers is the Evolution of Cooperation, which has the prospect of major implications for Moral Philosophy as well as all the behavioural sciences.

Boids screenshot

Screenshot of Craig Reynolds' Boids software, which simulates birds' flocking behaviour