1,721,061 research outputs found
Simulating bounded rationality: Optimality modelling without an optimality commitment
No full textIf you accept that evolved agents will be only boundedly rational, what are the consequences for the tools we use to make models of and construct theories about behaviour? In particular, consider methods like agent-based modelling -- with its roots in game theory and incorporating a notion of optimization or maximizing across alternatives -- do such methods remain viable? Or can cognitive science only deal with contingent, historical accounts of behaviour? I will argue that optimality models can continue to be used, because when used properly they were never wedded to a global notion of optimization or rationality in the first place. Such models are best viewed as ways of finding out which of a pre-specified set of behavioural alternatives is likely to dominate in a specific environment. As such, they are important tools for a program of research into bounded rationality. The argument will be illustrated with examples from modelling work on social learning in rats and intentional communication in monkeys
Modelling academic research funding as a resource allocation problem
No full textAcademic research funding is allocated through a competitive bidding process that may lead to inefficiency as excessive time is spent on proposal writing. We develop a simple agent-based model of the process and find that current systems are indeed likely to be inefficient. Alternative allocation schemes involving either a cap on individual effort or appraisal from the centre are indicated as improvements
Commentary: Evolutionary simulation modelling clarifies interactions between parallel adaptive processes
No full textHomeostatic plasticity improves signal propagation in continuous time recurrent neural networks
No full textContinuous-time recurrent neural networks (CTRNNs) are potentially an excellent substrate for the generation of adaptive behaviour in artificial autonomous agents. However, node saturation effects in these networks can leave them insensitive to input and stop signals from propagating. Node saturation is related to the problems of hyper-excitation and quiescence in biological nervous systems, which are thought to be avoided through the existence of homeostatic plastic mechanisms. Analogous mechanisms are here implemented in a variety of CTRNN architectures and are shown to increase node sensitivity and improve signal propagation, with implications for robotics. These results lend support to the view that homeostatic plasticity may prevent quiescence and hyper-excitation in biological nervous systems
The explanatory value of some post-connectionist models
No full textThought displays a systematicity that cannot be explained by the connections between simple neuron-like units. This is the gist of Fodor and Pylyshyn’s (F & P) challenge to connectionism. Furthermore, they assume thought to be representational and there is no available method to detect representations among the mere relations between neurons and the like. In this talk we would like to question the fairness of the challenge and the correctness of the assumption, but also the need for connectionism to accept the challenge at face value. F & P seem to force us to choose between representationalism and some form of eliminativism with respect to systematicity. However, we will argue that this is a false dilemma. An explanatory pluralism grounded on the idea that thought is a property of the relation between an agent and its environment (and peers, if any) is sufficient to decline F & P’s invitation to answer the challenge by embracing connectionist explanatory fascism. This idea follows a long tradition in theoretical biology. The possibility of such a pluralism will be explored by means of recent examples from artificial life. We will finish by wondering whether, once that we abandon representationalism, it makes sense to distinguish between explanatory and ontological pluralism
Task allocation in networks of satellites with Keplerian dynamics
No full textThe management of distributed satellite systems requires the coordination of a large number of heterogeneous spacecraft. Task allocation in such a system is complicated by limited communication and individual satellite dynamics. Previous work has shown that task allocation using a market-based mechanism can provide scalable and efficient management of static networks; in this paper we extend this work to determine the impact of dynamic topologies. We develop a Keplerian mobility model to describe the topology of the communication network over time. This movement model is then used in simulation to show that the task allocation mechanism does not show a significant decrease in effectiveness from the static case, reflecting the suitability of distributed market-based control to the highly dynamic environment
Distributed and Centralized Task Allocation: When and Where to Use Them
No full textSelf-organisation is frequently advocated as the solution for managing large, dynamic systems. Distributed algorithms are implicitly designed for infinitely large problems, while small systems are regarded as being controllable using traditional, centralised approaches. Many real-world systems, however, do not fit conveniently into these "small" or "large" categories, resulting in a range of cases where the optimal solution is ambiguous. This difficulty is exacerbated by enthusiasts of either approach constructing problems that suit their preferred control architecture. We address this ambiguity by building an abstract model of task allocation in a community of specialised agents. We are inspired by the problem of work distribution in distributed satellite systems, but the model is also relevant to the resource allocation problems in distributed robotics, autonomic computing and wireless sensor networks. We compare the behaviour of a self-organising, market-based task allocation strategy to a classical approach that uses a central controller with global knowledge. The objective is not to prove one mechanism inherently superior to the other; instead we are interested in the regions of problem space where each of them dominates. Simulation is used to explore the trade-off between energy consumption and robustness in a system of intermediate size, with fixed communication costs and varying rates of component failure. We identify boundaries between regions in the parameter space where one or the other architecture will be favoured. This allows us to derive guidelines for system designers, thus contributing to the development of a disciplined approach to controlling distributed systems using self-organising mechanisms
- …
