What is complex systems science (CSS)? In a nutshell, it is the study of systems that are considered fundamentally complex: those in which there are many parts involved, and often relationships between parts are just as important as the parts in determining system properties. Complex systems are set apart from those that are “merely” complicated via the tendency to display emergent properties over scale, self organisation over time, involving feedbacks and non linearity. Common examples include understanding collective behaviour, in particular under contexts of natural resource management or institutional decision making (socio-ecological systems), as well as natural phenomena such as ecosystems and climate.
Modelling and simulation of these systems is employed to understand their dynamics, optimise strategies, and predict responses. CSS has a natural base in network theory, but also integrates agent based modelling, systems theory, game theory, and artificial intelligence, among other things. One of the things that really interested me about CSS was that it seems to ignore traditional disciplinary boundaries – complex systems are everywhere – and therefore there is a fantastic opportunity for the transfer of concepts, tools and methodologies across disciplines.
Some highlights of the workshop included:
Michael Breakspear (Queensland Institute of Medical Research) enlightened us on how CSS is being used to understand how our brains work: how neuroscience is combining information on the brains anatomical networks (the physical “wiring”), with functional dynamics (relationships between activity in different parts of the brain), to shed light upon “effective” networks (causal effects). Turns out some of the most connected parts of the brain is used most not when we do complex tasks, but rather when we daydream…
Richard Fuller from (CSIRO Ecosystem Sciences/UQ) discussed presentation of science, and in particular how sometimes presentation of research regarding CSS may not be easy. Media outlets may be great for research that has a clear, succinct, and non-controversial message, but can be minefields for more challenging messages.
Markus Brede (CSIRO Marine and Atmospheric Research) discussed how cooperation can evolve (in game theory) with allowances for learning and building of trust (both in direct repeated interactions, and observation of the behaviour of others), voluntary participation, altruism (kin selection), and structured populations. In some networks cooperation could be attributable to key well connected “leaders”, to the opportunity to avoid interactions with players who were not fair, and also to bias who we want to learn from (aspiration bias). Integration of some of these rules could create areas of cooperation, no matter how strong the game (how tempting it was to default).
Kirsty Kitto (QUT) gave us a preview of her recent work using quantum formalism to provide a geometric framework to model how agents make decisions when these are fundamentally contextualised. This framework can also formalise communication between agents (including when and where communication might lead to a change in agent position). I’m not going to pretend I understood it all, but it looked really cool, and definitely Kitto is one to look out for in the near future as she and her group develop this framework further to describe aggregate behaviour of multiple agents, and eventually use it for modelling in CSS.
Ryan McAllister (CSIRO Ecosystem Sciences) discussed some of his recent work using experimental economics to explore economic behaviour in the face of variability – and how trust can be a key driver of success. Look out for more from him regarding how social institutions may be enhanced to deal with the uncertainty and variability predicted under climate change!
It was great to see the various methodological approaches of CSS, and really interesting to see how different researchers used them: while Michael used them to understand the system, and Markus to explore key drivers in the system, Ryan was more focussed on using the models in applied situations to develop testable hypotheses. Then in stepped the delightful Pascal Perez (University of Wollongong), who uses CSS tools (participatory agent based modelling) not only to try to understand and explore natural resource based systems, but also with a large focus on understanding how humans interacted with natural systems, and for providing a platform for managers and stakeholders to effectively communicate.
With so many different disciplines, methods, and objectives, complex systems science is certainly a little confusing at first. But the enthusiasm for taking complicated things and reducing them to the “merely” complex really cuts at these traditional boundaries. So thanks again to CSIRO CSS for this really interesting workshop – and for those of you interested, keep an eye out, it’s run every year.