Gamma and theta/alpha-band oscillations in the electroencephalogram distinguish the content of inner speech.

Abstract

Inner speech refers to the silent production of language in one's mind. As a purely mental action without obvious physical manifestations, inner speech has been notoriously difficult to quantify. Inner speech is thought to be closely related to overt speech. Overt speech has been consistently shown to elicit reduced auditory-evoked potentials compared to externally-generated speech (a phenomenon known as speaking-induced suppression), as well as changes in oscillatory activity in the gamma and theta frequency bands. Given the functional similarities between inner and overt speech, the present study used a novel experimental protocol to investigate whether these metrics can be used to distinguish the content of a person's inner speech. Healthy control participants (n = 129) produced an inner syllable at a precisely specified time. An audible syllable was concurrently presented which either matched or mismatched the content of the inner syllable. The N1 component of the auditory evoked potential was suppressed in the Match condition while the P2 component was suppressed in the Mismatch condition, replicating our previous findings. A novel finding was that the Match and Mismatch conditions could be differentiated on the basis of their evoked gamma and theta / alpha oscillations. There was a single gamma-band oscillation in the vicinity of the P2 that differed in spectral power between the Match and Mismatch conditions, which provides support for the idea that 'late' gamma-band activity may index consciously-perceived expectancy violations, or cognitive prediction errors. The Match and Mismatch conditions also differed in terms of their evoked power in a temporally-extended cluster which spanned the theta and alpha bands. The results of this study provide support for the phenomenon of 'inner speaking-induced suppression', and demonstrate that inner syllables can be differentiated using scalp electroencephalography (evoked potentials and oscillations) based on whether their content matches that of a simultaneously-presented audible syllable.Significance statement Inner speech refers to the silent production of language in one's mind. As a purely mental action without obvious physical manifestations, inner speech has been notoriously difficult to quantify empirically. The aim of the present study was to develop an objective, electroencephalography-based marker of inner speech that was sensitive to its content. The results of this study demonstrate that it is possible to distinguish between two syllables produced in inner speech on the basis of their auditory-evoked activity and evoked oscillations in the gamma and theta/alpha bands in the electroencephalogram. The gamma-band oscillation may represent a 'cognitive' prediction error. The ability to determine the content of a person's inner speech on the basis of a non-invasive biometric signal could have significant commercial, industrial, and clinical applications.