Abstract
This paper describes research to inform the production of spatial audio that consolidates knowledge from several disparate fields. A perceptual model is proposed, based on contemporary theories of perception, as the basis for new approaches to both audio spatial understanding and the generation of reproduced sound fields using loudspeakers. A fine-grain, modular model of perception is suggested that will allow audio attributes to have perceptual significance with respect to their causal trajectories. This represents an evolution towards the construction of believable sound fields which will result in improved apprehension of spatial information which is tuned to our perceptual systems, over traditional geometric based approaches to sound localisation with a sound field.