This work addresses the problem of difficulties in classical interpretation of combination tones as non- linear distortions. One of the basic problems of such an interpretation is to point out the sources of these distortions. Besides, these kinds of distortions have numerous “anomalies” which are difficult to explain on the grounds of physics or physiology. The aim of the model presented in this paper is to show that combination tones phenomenon can be explained as an effect of central mechanisms. Most of existing theories of pitch perception focus mainly on virtual pitch perception and do not take into account combination tones as an element of the same mechanism. The proposed model of central auditory processing for pitch perception allows one to interpret in a coherent way both virtual pitches and combination tones phenomena. This model is of a demonstrative nature and gives an introduction to more advanced model. It belongs to the class of spectral models and it will be shown that such a model can be in a simple way extended to spectral - time model which is partially consistent with autocorrelation models.
The present study consisted of two experiments. The goal of the first experiment was to establish the just noticeable differences for the fundamental frequency of the vowel /u/ by using the 2AFC method. We obtained the threshold value for 27 cents. This value is larger than the motor reaction values which had been observed in previous experiments (e.g. 9 or 19 cents). The second experiment was intended to provide neurophysiological confirmation of the detection of shifts in a frequency, using event-related potentials (ERPs). We concentrated on the mismatch negativity (MMN) - the component elicited by the change in the pattern of stimuli. Its occurrence is correlated with the discrimination threshold. In our study, MMN was observed for changes greater than 27 cents - shifts of ±50 and 100 cents (effect size - Cohen’s d = 2.259). MMN did not appear for changes of ±10 and 20 cents. The results showed that the values for which motor responses can be observed are indeed lower than those for perceptual thresholds.