Spectral-temporal EEG dynamics of speech discrimination processing in infants during sleep

Abstract

BackgroundOddball paradigms are frequently used to study auditory discrimination by comparing event-related potential (ERP) responses from a standard, high probability sound and to a deviant, low probability sound. Previous research has established that such paradigms, such as the mismatch response or mismatch negativity, are useful for examining auditory processes in young children and infants across various sleep and attention states. The extent to which oddball ERP responses may reflect subtle discrimination effects, such as speech discrimination, is largely unknown, especially in infants that have not yet acquired speech and language.ResultsMismatch responses for three contrasts (non-speech, vowel, and consonant) were computed as a spectral-temporal probability function in 24 infants, and analyzed at the group level by a modified multidimensional scaling. Immediately following an onset gamma response (30–50 Hz), the emergence of a beta oscillation (12–30 Hz) was temporally coupled with a lower frequency theta oscillation (2–8 Hz). The spectral-temporal probability of this coupling effect relative to a subsequent theta modulation corresponds with discrimination difficulty for non-speech, vowel, and consonant contrast features.DiscussionThe theta modulation effect suggests that unexpected sounds are encoded as a probabilistic measure of surprise. These results support the notion that auditory discrimination is driven by the development of brain networks for predictive processing, and can be measured in infants during sleep. The results presented here have implications for the interpretation of discrimination as a probabilistic process, and may provide a basis for the development of single-subject and single-trial classification in a clinically useful context.ConclusionAn infant’s brain is processing information about the environment and performing computations, even during sleep. These computations reflect subtle differences in acoustic feature processing that are necessary for language-learning. Results from this study suggest that brain responses to deviant sounds in an oddball paradigm follow a cascade of oscillatory modulations. This cascade begins with a gamma response that later emerges as a beta synchronization, which is temporally coupled with a theta modulation, and followed by a second, subsequent theta modulation. The difference in frequency and timing of the theta modulations appears to reflect a measure of surprise. These insights into the neurophysiological mechanisms of auditory discrimination provide a basis for exploring the clinically utility of the MMR TF and other auditory oddball responses.

The relationship between mismatch response and the acoustic change complex in normal hearing infants

“…In addition to the MMR response, there is an additional peak following the MMR peak; it is unclear what this third peak represents. One possibility might be that the initiation of a response from any given stimulus sets up a spatial-temporal oscillation containing information about the stimulus properties, and that the absence of a stimulus that would follow causes a rapid dampening of that oscillation resulting in what appears to be an off-set peak (Freeman and Vitiello, 2006; Gilley et al, 2017). This is further supported by the findings in the ACC results in which a fourth peak appears after the offset of a periodic stimulus sequence (as described above).…”

Mismatched response predicts behavioral speech discrimination outcomes in infants with hearing loss and normal hearing

“…For the purposes of segment alignment, we considered the onset of each vowel stimulus to be the point at which voicing begins (i.e., voice onset) and the onset of each CV stimulus to be the point at which the consonant burst begins. Time‐frequency coherence (TFC) maps were computed for the standard and deviant conditions in each MMR SE block using a bootstrapped averaging procedure (Gilley et al, 2017; Uhler et al, 2018). First, data were pooled into two sets of trials defined by whether a trial was standard or deviant and with the constraint that included trials must be both preceded by and followed by a standard trial.…”

“…This approach allows us to examine and define global response features that contribute to outcomes predictions while accounting for individual differences and other sources of variability. Using this method, we compared three different stimulus contrasts (/a‐i/, /ba‐da/, and non‐speech stimuli) to identify the optimal contrast for feature extraction (Gilley et al, 2017). We examined TF‐MMR SE in 44 sleeping, normal hearing (NH) infants aged 1.8–4.6 months, the age range in which infants with HL typically receive hearing aids.…”

“…Our recent work, described above, focused on improving ERP analyses in order to reliably capture and analyze individual MMR SE results from infants with NH and with HL. To this end, we have examined the application of time‐frequency (TF) transformations and latent variable (i.e., variables that are not directly observed but rather inferred from observed variables) analyses for extracting experimentally relevant signals in the MMR (Gilley et al, 2017) and have systematically examined the effects of different time and frequency windows for identifying MMR signals (Uhler et al, 2018). While such methods have been successful for extracting MMR signals, they have also highlighted the individual variability inherent in infant ERPs.…”

NMDA Receptor Antagonist Effects on Speech-Related Mismatch Negativity and Its Underlying Oscillatory and Source Activity in Healthy Humans

“…Although gamma band oscillations have also been associated with depression and fast-acting antidepressants like ketamine (Fitzgerald and Watson, 2018), they have been frequently associated with the pathophysiology of schizophrenia and associated auditory evoked response deficits (Curic et al, 2019). Moreover, relative to standard sounds, deviant sounds elicit a cascade of oscillatory modulations (beginning with gamma and followed by beta, both of which are coupled with theta) that are altered by acute ketamine administration (Gilley et al, 2017; Schuelert et al, 2018). Further, EEG source localization lacks the spatial sensitivity of other imaging methodologies such as fMRI and was limited to specific regions of interest in the temporal and inferior frontal cortex, thus, it did not include examination of a possible role for other circuits (e.g., thalamocortical and limbic) involved in sensory processing.…”