Title：Evolutionary mechanics: degeneracy and the emergence of flexibility in a dynamic and uncertain world

Speaker：DR. James Whitacre, University of Birmingham

Time：2:00-3:00pm, April 25th, 2011

Venue: Meeing Room 337, Level 3,  Building No. 5

Abstract:

Engineered systems are designed to deftly operate under predetermined conditions yet are notoriously fragile when unexpected perturbations arise. In contrast, biological systems operate in a highly flexible manner; quickly learning adequate responses to novel conditions and evolving new routines/traits to remain competitive under persistent environmental change.

The basic principles by which biological systems can be robust to widely different classes of perturbations yet display a propensity to adapt under novel conditions remains unresolved yet is essential to the application of Darwinian and complex systems principles in fields such as artificial intelligence (evolutionary algorithms, artificial life, machine learning), organization science, and other newly emerging fields such as complex systems engineering.

In this brief talk, I will outline a basic set of concepts that are often overlooked in both our understanding of biological systems and the application of biological principles but yet are potentially fundamental to the relationships between robustness, complexity, and evolvability throughout biology.  In particular, I will outline arguments that functional plasticity (proteins/agents/devices that can perform several distinct functions but whose function changes depending on its context, i.e. one-to-many mapping), functional redundancy (structurally distinct proteins/agents/devices that at times perform similar functions and are thus compensatory in certain contexts, i.e. many-to-one mapping), and the architectural relationships between these two properties are often primary determinants of a system’s propensity to maintain/conserve core functions while also evolving new ones. 

These non-trivial Genotype:Phenotype mapping features could be relevant to our understanding of certain biological processes (e.g. the mechanistic/molecular basis of adaptive immunity, the robustness of some disease states such as cancer) the analysis of biological data (modelling polygenic quantitative traits in genome wide association studies) and the resilience of certain socio-technical systems. At present, my colleagues and I have accumulated some evidence that appears to support the generality of the arguments that will be outlined in this talk, e.g. in the context of evolvable problem representations for dynamic optimization problems, neutral evolution theory in genome:proteome models, and resilience to uncertainty in strategic planning problems.

In consideration of the diversity of academic disciplines that have informed this research, in this talk I will ground these ideas using abstract and conceptually intuitive explanations as to why functional plasticity and functional redundancy can confer advantages in the robustness and flexibility of an abstract multi-agent system that exists within a volatile environment.

Short bio:

James Whitacre is a research fellow with the Computational Intelligence group at the University of Birmingham.  Before coming to Birmingham, James was a post-doc at the Artificial Life and Adaptive Robotics lab at ADFA and he received his PhD from University of New South Wales where he studied the relationship between adaptation and self-organization in evolutionary algorithms, and his BS from Princeton University. 

His current research looks broadly at the relationships between robustness, evolvability, and complexity and how these relationships change between engineered and biological systems.  Applications of this research have looked at evolvable G:P mappings for evolutionary algorithms applied to dynamic optimization problems, group behaviors that improve evolvability in ensemble learning, design concepts for developing robust and adaptable transportation vehicle fleets, and design concepts for enhancing reconfiguration flexibility in manufacturing assembly systems.  Current theoretical studies are also investigating biologically plausible causes of criticality in gene regulatory networks, specific properties of biological phenotypes that influence ecosystem resilience, and the causes and therapeutic implications of cancer evolution.