Quantitative Model of Human Dynamic Attention Allocation

The objective of this proposal is to demonstrate the feasibility of a quantitative dynamical model of human attention and perception. Recent experimental findings have demonstrated the importance of event timing for perceiving and attending to complex sequences of events: 1) The auditory system uses active temporal mechanisms for analysis of sound from the earliest stages of sensation through the development of integrated high-level percepts; 2) Attention exploits temporal stimulus structure to coordinate perception, cognition and action in a way that is optimal for a given task environment. In both cases, the nervous system exploits nonlinear oscillation to encode and interact with complex stimuli that unfold over time, albeit over different time scales. Analysis and interaction with temporal stimuli via networks of nonlinear oscillators is referred to here as nonlinear time-frequency transformation. This proposal is to develop a nonlinear time-frequency transformation software toolbox, and to evaluate its utility as a model of human attention and perception. Evaluation will include 1) analyses of temporally complex data streams 2) analysis of data streams from multiple sources (spatial locations or perceptual modalities) and 3) production of attentional expenditure estimates for a human confronted with such a data stream in a complex task environment.