New Paper: Neural modelling of the encoding of fast frequency modulation

The human brain is the best speech-recognising system to date. However, the computational mechanisms that the human brain uses for excelling at speech recognition are far from understood. Here we investigate how formant transitions, short frequency sweeps that characterise consonants preceding a vowel, are processed in the brain. Our main hypothesis is that frequency sweeps are processed using a predictive strategy that reinforces expected representations at lower levels. We used computational modelling to predict how such mechanism would affect the perception of a series of frequency sweeps within the frequency range and duration of formant transitions. To validate the model, we conducted a behavioural experiment designed to test whether the frequency sweeps were perceived as predicted by our model. Our results indicate that predictive processing is an essential mechanism for speech processing even at early stages of the processing hierarchy.

Tabas A. & von Kriegstein, K. (2021) Neural modelling of the encoding of fast frequency modulation. PLOS Computational Biology 17(3): e1008787. https://doi.org/10.1371/journal.pcbi.1008787