This is a new project that I am becoming very interested in.
The agent based modelling (ABM) of human decisions is becoming increasingly widespread as large scale attempts to understand and predict social behaviour become more common. Within such models, it is necessary to incorporate a sophisticated understanding of context, and the manner in which different contextual factors might have a strong influence upon an agents choices and decisions. While notions such as historical context (where the past history of an agent might effect its later actions) and situational context (where the agent will choose a different action in a different situation) abound in these modelling scenarios, such approaches are not always sufficient for generating an understanding of all important social effects. Indeed, it is often the case that people might ``know too differently'', and frequently results in a set of competing, and sometimes even incompatible framings of the information within a system.
For example, consider the problem of finding an optimal water management plan that can be followed in a changing climate, where a variety of different stakeholders exist, each with widely competing interests. In a situation of water shortage a number of different framings can be provided, resulting in the attribution of different meanings to the situation, and each potentially requiring different responses. A farmer will be concerned with insufficient supply, while environmentalists might approach the water system thinking that the problem is one of excessive consumption. Context can thus serve as a frame for two very different positions, and these two contexts appear to be incompatible in that they require different actions. Farmers will clamour for more water to be released, while environmentalists will generally want environmental flows to be maintained. Such scenarios become more complex when we consider the manner in which the framing that an agent adopts for one scenario (such as water management), might influence their decisions about other scenarios (such as judgements about climate change or politics). However, these highly interdependent scenarios are not represented in an adaptive manner by the standard approaches to agent based modelling. In particular, the manner in which a novel context might evoke a new and highly adaptive set of responses is yet to be captured in the modelling literature.
A modelling technology which could account for the effect of these different problem representations would benefit regulatory bodies which need to navigate between multiple interests and concerns, and I am currently searching for viable avenues here. Many of them are inspired by quantum theory, but the models developed are more geometric, than Hilbert space inspired.
I am currently seeking postgraduate students to work on this project. I am particularly interested in recruiting students with a strong background in any of: physics; mathematics; computing science. If you are interested then you should email me your CV, as well as a 3 page project proposal which discusses:
Much of my current work focusses upon this. I am working with Professor Peter Bruza to develop what would fundamentally be a new quantum theory, one that uses the contextuality of quantum measurement to good effect to model the contextual dependency that words exhibit. Consider for example the word "boxer"... are you thinking of a dog? or of someone who punches people? Depending upon the context in which this word is used you will be biased towards assuming one meaning for it. In Is there something quantum-like about the human mental lexicon? this theory is developed in some detail, but the basic idea is illustrated below.
Here, a word, w, is represented with respect to two different bases, or contexts. In these different contexts it has a different probability of recall, which means that the quantum formalism provides a very natural way in which to incorporate the exceptionally common ambiguity of meaning we see occuring in natural language. For more details see the paper.
One of my big pet projects at the moment involves the interpretation of QT. In my Thesis, my paper High End Complexity and the paper Why Quantum Theory? I have started to sketch out my approach to QT, namely that it is a theory of contextually complex systems. This theory falls into the area of Quantum Interaction
While simple behaviour is relatively straightforward to define there appears to be a wide range of complex behaviour. It seems possible to identify a complexity scale, which moves from simple systems described by simple theories which make reductive or separable assumptions (e.g. the Newtonian understanding of projectile motion), through those theories that describe more complicated behaviour consisting of, for example, a large number of components but which are still essentially separable (e.g. statistical mechanics), and then into the broad class of systems often identified as complex. Within this class of complex systems we can place a number of those analytic approaches traditionally associated with complexity, but which apply reductive assumptions, at the low end of the `complex behaviour' spectrum before we move into the behaviour that is more poorly modelled by such theories. At the high end of such a scale we would expect to see processes such as biological development, the evolution of mind, language and societies etc., processes which exhibit significant contextual dependency and are not well modelled by our current theories. The above figure illustrates how such a scale might look.
For more information about this work see my paper High End Complexity
One of my very long term interests involves developing a theory of the dynamical formation of hierarchical structures. In particular, I would like to develop dynamical models of the process of biological development. The basic idea of this model was sketched out in my paper High End Complexity, and would involve the development of a quantum field theoretic model of differentiation coupled with a set of morphogenic equations describing the growth of cells once they have differentiated. This is a project of immense interest to me, it would involve establishing a number of new symmetry groups and making use of dynamical symmetry breaking in order to generate novelty... if I ever get time and/or funding for it I will return to this (but it is always in the back of my mind).
I have in the course of my research worked with a number of stochastic iteration equations. I am particularly interested in finding classes of such equations that can generate complex emergent behaviour. I think it is likely that such equations must have a highly non-linear term, and a driving noise term, and interestingly, the ones that appear to generate the most interesting behaviour generally represent relational structures using a matrix form. I would very much like to formalise these ideas though...
My PhD research at Flinders University centred upon Process Physics and was carried out under the supervision of Professor Reg Cahill. The apparent success of Process Physics in generating interesting, and possibly complex, emergent behaviour led to many of my current research interests. Rather than the static 4-dimensional modelling of present day (non-process) physics, Process Physics is providing a dynamic model where space and matter are seen to emerge from a fundamentally random but self-organising system. The key insight is that to adequately model reality we must move on from the traditional non-process syntactical information modelling to a process semantic information modelling; such information is "internally meaningful". See the Process Physics homepage for more details.