Speech Understanding: Future Predicts the Past

Speech perception is determined by surrounding context. However, this is not necessarily a serial process where the context precedes the speech signal. A study by Gwilliams et al (2018) in the Journal of Neuroscience has demonstrated that the auditory system has postdictive processing capability, in other words subsequent context can bias the perception of the preceding signal.

In the study, Gwillams and colleagues investigated how phoneme perception was influenced by subsequent context. To do so, they recorded whole-head magnetoencephalography (MEG) while participants listened to phonemes that varied in ambiguity prior to the onset of syllables or words.

What they discovered was that the auditory cortex actively maintained the acoustic signal, while concurrently making guesses about the identity of the incoming signal. Further, this occurred without conscious awareness of the disambiguity. This allowed the auditory system to use contextual information available later than the sensory input.  The findings suggest the brain re-evaluates the interpretation of speech the moment that each subsequent speech sound is heard to update its interpretation.

Reference

Gwilliams, Linzen, Poeppel, Marantz (2018). In spoken word recognition the future predicts the past. J Neurosci, prepub.