Statistically driven spectral analysis requires dynamically presented stochastic tone streams

Evidence

The present paradigm builds on research from the spectral and the temporal domain, e.g. profile analysis (Green and Berg 1991 ) or some paradigms in mismatch negativity. For example in profile analysis subjects assess the spectral shape of stimuli. In this context it has also been demonstrated that subjects can compare spectra, however, the results were not interpretable in a statistical sense, due to several reasons. First, the time scale of presentation is much longer in the present paradigm than typically in profile analysis, where the stimulus is presented for multiple seconds rather than subsecond presentations. Second, in the present context the stimulus is composed of randomly timed occurrences of tones in certain frequency bins (which defines the spectral profile over time), rather than static profiles from constantly presented tones or filtered noise. Third, the comparison between different profiles is performed dynamically, rather than in a two stimulus comparison task. As described above this allowed us to study the dynamic integration of stochastic information, and enables the collection of reaction times as another indicator of the underlying processes. — Dijk, Pim van, Physiology, Psychoacoustics and…, 3340

Categorization

• QUESTION: Is there converging evidence that reaction times (not just accuracy) index how the auditory system accumulates spectral statistics?