A simple method for generating arbitrary Gaussian noise and compensating filter properties of a stimulator to measure the frequency response function of a mechanosensory sensillum

Abstract

Gaussian noise is one of the principal inputs used for studying biological systems, especially sensory and neuronal systems. Because of the inherent nonlinearities of these systems, the properties of the noise strongly influence the outcome of the analysis. We show for the case of an insect filiform sensillum, a simple mechanoreceptor with a single sensory cell, that changes in noise amplitude and spectral properties affect the linear transfer function of the sensillum. We then explain how to use the inverse fast Fourier transform to generate Gaussian noise with an arbitrary frequency spectrum, including band-limited white noise with a perfectly sharp roll-off edge. Finally, we demonstrate a straightforward procedure for compensating filter properties of a stimulator in order to produce a proper band-limited Gaussian white noise stimulus. Using this approach one can generate well defined Gaussian noise stimulus that can be adapted to any particular application.