Seth Bullock

Made to Measure: Ecological Rationality in Structured Environments

A working assumption that processes of natural and cultural evolution have tailored the mind to fit the demands and structure of its environment begs the question: how are we to characterize the structure of cognitive environments? Decision problems faced by real organisms are not like simple multiple-choice examination papers. For example, some individual problems may occur much more frequently than others, whilst some may carry much more weight than others. Such considerations are not taken into account when (i) the performance of candidate cognitive mechanisms is assessed by employing a simple accuracy metric that is insensitive to the structure of the decision-makers environment, and (ii) reason is defined as the adherence to internalist prescriptions of classical rationality. Here we explore the impact of frequency and significance structure on the performance of a range of candidate decision-making mechanisms. We show that the character of this impact is complex, since structured environments demand that decision-makers trade off general performance against performance on important subsets of test items. As a result, environment structure obviates internalist criteria of rationality. Failing to appreciate the role of environment structure in shaping cognition can lead to mischaracterising adaptive behavior as irrational.

Environment quality predicts parental provisioning decisions

Although avian parents appear to exhibit a variety of feeding strategies in nature, there currently exist no models or theories that account for this range of diversity. Here we present the results of a computer simulation designed to model inter–dependent parental decisions, where investment is meted out in small doses, and must be distributed over time to maximize return on investment at the end of the parental–care period. With this technique we show that the success of various simple, observed, parental rules of thumb varies with environmental resource level, and that increasing the complexity of parental decision rules does not necessarily result in increased fitness.

Navigating the perfect storm: Research strategies for socialecological systems in a rapidly evolving world

The ‘Perfect Storm’ metaphor describes a combination of events that causes a surprising or dramatic impact. It lends an evolutionary perspective to how social-ecological interactions change. Thus, we argue that an improved understanding of how social-ecological systems have evolved up to the present is necessary for the modelling, understanding and anticipation of current and future social-ecological systems. Here we consider the implications of an evolutionary perspective for designing research approaches. One desirable approach is the creation of multi-decadal records produced by integrating palaeoenvironmental, instrument and documentary sources at multiple spatial scales. We also consider the potential for improved analytical and modelling approaches by developing system dynamical, cellular and agent-based models, observing complex behaviour in social-ecological systems against which to test systems dynamical theory, and drawing better lessons from history. Alongside these is the need to find more appropriate ways to communicate complex systems, risk and uncertainty to the public and to policy-makers.

Combating Coevolutionary Disengagement by Reducing Parasite Virulence

While standard evolutionary algorithms employ a static, absolute fitness metric, co-evolutionary algorithms assess individuals by their performance relative to populations of opponents that are themselves evolving. Although this arrangement offers the possibility of avoiding long-standing difficulties such as premature convergence, it suffers from its own unique problems, cycling, over-focusing and disengagement. Here, we introduce a novel technique for dealing with the third and least explored of these problems. Inspired by studies of natural host-parasite systems, we show that disengagement can be avoided by selecting for individuals that exhibit reduced levels of virulence, rather than maximum ability to defeat coevolutionary adversaries. Experiments in both simple and complex domains are used to explain how this counterintuitive approach may be used to improve the success of coevolutionary algorithms.

The View From Elsewhere: Perspectives on ALife Modeling

Many artificial life researchers stress the interdisciplinary character of the field. Against such a backdrop, this report reviews and discusses artificial life, as it is depicted in, and as it interfaces with, adjacent disciplines (in particular, philosophy, biology, and linguistics), and in the light of a specific historical example of interdisciplinary research (namely cybernetics) with which artificial life shares many features. This report grew out of a workshop held at the Sixth European Conference on Artificial Life in Prague and features individual contributions from the workshops eight speakers, plus a section designed to reflect the debates that took place during the workshops discussion sessions. The major theme that emerged during these sessions was the identity and status of artificial life as a scientific endeavor.

Adding “foveal vision” to Wilson's animat

Different animals employ different strategies for sampling sensory data. In animals that can see, differences in sampling strategy manifest themselves as differences in field of view and in spatially variant sampling (so-called foveal vision). In analyzing adaptive behavior in animals, or attempting to design autonomous robots, mechanisms for exploring variations in sensory sampling strategy will be required. This article describes our work exploring a minimal system for investigating the effects of variations in patterns of sensory sampling. We have reimplemented Wilsons animat (Wilson, 1985b) and then experimented with altering its sensory sampling pattern (i.e., its sensory field). Empirical results are presented which demonstrate that alterations in the sensory field pattern can have a significant effect on the animats observable behavior. Analysis of our results involves characterizing the interaction between the animats sensory field and the environment within which the animat resides. We found that the animats observed behavior can, at least in part, be explained by the animat cautiously moving in a manner that attempts to maximize the generation of new information from the environment over time. We demonstrate that similar explanations can be offered for behavioral patterns in real animals. The article concludes with a discussion of the generality of the results and reflections on the prospects for further work.

Learning lessons from the common cold: How reducing parasite virulence improves coevolutionary optimization

Inspired by the virulence of natural parasites, a novel approach is developed to tackle disengagement, a detrimental phenomenon coevolutionary systems sometimes experience. After demonstrating beneficial results in a simple model, minimum comparison sorting networks are coevolved, with results suggesting that moderating parasite virulence can help in practical problem domains.

Stability in flux: community structure in dynamic networks.

The structure of many biological, social and technological systems can usefully be described in terms of complex networks. Although often portrayed as fixed in time, such networks are inherently dynamic, as the edges that join nodes are cut and rewired, and nodes themselves update their states. Understanding the structure of these networks requires us to understand the dynamic processes that create, maintain and modify them. Here, we build upon existing models of coevolving networks to characterize how dynamic behaviour at the level of individual nodes generates stable aggregate behaviours. We focus particularly on the dynamics of groups of nodes formed endogenously by nodes that share similar properties (represented as node state) and demonstrate that, under certain conditions, network modularity based on state compares well with network modularity based on topology. We show that if nodes rewire their edges based on fixed node states, the network modularity reaches a stable equilibrium which we quantify analytically. Furthermore, if node state is not fixed, but can be adopted from neighbouring nodes, the distribution of group sizes reaches a dynamic equilibrium, which remains stable even as the composition and identity of the groups change. These results show that dynamic networks can maintain the stable community structure that has been observed in many social and biological systems.

Logistic constraints on 3D termite construction

The building behaviour of termites has previously been modelled mathematically in two dimensions. However, physical and logistic constraints were not taken into account in these models. Here, we develop and test a three-dimensional agent-based model of this process that places realistic constraints on the diffusion of pheromones, the movement of termites, and the integrity of the architecture that they construct. The following scenarios are modelled: the use of a pheromone template in the construction of a simple royal chamber, the effect of wind on this process, and the construction of covered pathways. We consider the role of the third dimension and the effect of logistic constraints on termite behaviour and, reciprocally, the structures that they create. For instance, when agents find it difficult to reach some elevated or exterior areas of the growing structure, building proceeds at a reduced rate in these areas, ultimately influencing the range of termite-buildable architectures.

Understanding the role of recruitment in collective robot foraging

When is it profitable for robots to forage collectively? Here we compare the ability of swarms of simulated bio-inspired robots to forage either collectively or individually. The conditions under which recruitment (where one robot alerts another to the location of a resource) is profitable are characterised, and explained in terms of the impact of three types of interference between robots (physical, environmental, and informational). Key factors determining swarm performance include resource abundance, the reliability of shared informa- tion, time limits on foraging, and the ability of robots to cope with congestion around discovered resources and around the base location. Additional experiments introducing odometry noise indicate that collective foragers are more susceptible to odometry error.