| Abstract #225, Date 2/15/99, Session H, Podium , 2:00p |
| A sum-of-exponentials model for estimating the component processes contributing to temporal masking functions |
| *C. Formby, J.C. Rutledge, L.P. Sherlock (University of Maryland) |
At last year's meeting, we (Formby et al, 1998; ARO Abstract #687) presented temporal masking functions that were characterized by complex patterns of temporal undershoot (enhanced detection) and overshoot (diminished detection) at onset and offset of a gated narrowband noise masker. These complex patterns may serve perceptually as cues to enhance the temporal onset and offset of the gated masker and to accentuate the temporal edges of the temporal masking function. The purpose of this presentation is to present a strategy for modeling the component processes that contribute to the general features of the temporal masking function, including the complex temporal undershoot and overshoot patterns. Our modeling strategy is similar in principle to that used in the study of vestibular adaptation (Formby et al, 1998; ARO Abstract # 226), and is well suited for evaluating responses to step-like stimulation of the kind used in electrophysiological studies of peripheral neural adaptation. The sum-of-exponentials model uses linear combinations of exponential functions to represent the unadapted excitation response to the masker and the opposing effects of adaptation. Three exponential components are represented in the model, corresponding generally to "very rapid", "rapid", and "short-term" adaptation processes described in animal studies of peripheral neural adaptation. Each adaptation component is subtractive and partially negates the unadapted excitation response. This subtractive strategy allows for quantification of the relative amplitude, time constant, and delay of each adaptation component with respect to the unadapted excitation component. The amplitude of the unadapted excitation component grows nonlinearly with masker level, and mirrors the compressive input-output velocity response of the basilar membrane. The model also includes a virtual start coefficient that allows each component to begin prior to the actual physical onset or offset of the masker. The masker onset and offset responses are modeled independently in this scheme and are approximately inverse operations. |