Emergent transducing computing for evolving interactive musical agents

  • P. E. Rawbone

    Student thesis: Doctoral Thesis

    Abstract

    This dissertation reports on interdisciplinary research concerning interaction in the fields of computing, Artificial Intelligence (AI), and computer music. It claims that computer programs prescribed in advance of the computing often cannot reasonably facilitate interactive music and AI, whereas interactive computing is more suitable. Interactive computing is an alternative approach, emplacing emphasis on real-time emergent interaction with the environment.

    A new agent model developed during this research extends the current conception of interactive computing but is ill-suited to simulation on a modern computer. In this respect the model is theoretical, employing an interactive transducing strategy which points toward prospective generative self-modifiable future computers. The model facilitates general environmental problem solving and potential solutions to the Symbol Grounding Problem and Frame Problem, and enables the extrapolation of a meta-problem, original to this dissertation, termed the interaction problem. These features support an extension of computationalism (the thesis that cognition is computable) with a proposal that cognition is interactively computable (interactive computationalism), offering a potential solution to the interaction problem.

    In consideration of the relevance of interactive computing in computer music, the new agent model is applied to the topic of virtual societies of interactive music agents, which includes a comparison with an existing computer music model. This aids an analysis and a presentation of exploratory ideas in interactive music, which is portrayed as comprising of real-time, parallel, unpredictable, and modifiable processes, and not a prescribed or static act. These ideas aim to assist the ongoing musical progression and growing sophistication of interactive music systems, which in the future might be restricted to prescriptive computing strategies.

    The relationship between the prescription and emergence of processes is a key issue for discussion in the respective topics that this dissertation reports on, and provides a point of debate around computer music composition and the predictability of agent computing in interaction with the environment. The subject of predictability results in a number of core ideas in this dissertation, enabling a new approach to the Frame Problem and Symbol Grounding Problem in AI and exploratory discussions of interactive computer music.

    Broadly, the examination provides an interactive framework incorporating interaction as computation in computer science, interaction as a thesis in AI, and interaction as a computational compositional conception in computer music. The dissertation closes in consideration of the challenges in a future implementation of the new agent model.

    Date of Award2011
    Original languageEnglish

    Cite this

    '