Cortical oscillations and speech processing: emerging computational principles and operations

Neural entrainment to pitch changes of auditory targets in noise

Neural entrainment to certain acoustic features can predict speech-in-noise perception, but these features are difficult to separate. We measured neural responses to both natural speech-in-noise and stimuli (auditory figure-ground) that simulate speech-in-noise without any acoustic or linguistic confounds such as stress contour and semantics. The figure-ground stimulus is formed by multiple temporally coherent pure-tone components embedded in a random tone cloud. Previous work has shown that discrimination of dynamic figure-ground based on the fundamental frequency (F0) of natural speech predicts speech-in-noise recognition independent of hearing and age. In this study, we compared the brain substrate for the figure-ground analysis based on the F0 contour and a statistically similar '1/f' contour with speech-in-noise. We used the temporal response function to predict the electroencephalography responses to the frequency trajectories of the auditory targets. We demonstrate that the brain significantly tracked the pitch changes in both AFG conditions (F0 and 1/F tracking) and a sentence-in-noise condition (F0 tracking) at similar latencies, but at similar magnitudes only when tracking the F0 contour. The pitch-tracking accuracy was consistently high across the delta and theta bands for the AFG condition but not for speech. Sensor-space analysis revealed that speech-in-noise performance correlated with the positive peak amplitude of the F0 figure-ground at 100 ms. Source-space analysis revealed bilateral temporal lobe and hippocampal generators, and strong tracking in the superior parietal lobe for auditory figures and natural speech. In conclusion, our findings demonstrate that the human brain reliably tracks the F0 trajectory of both speech and a non-linguistic figure in noise, with speech tracking showing reduced accuracy in the theta band compared to figure-ground tracking. Despite the difference in prediction accuracy, we reveal striking similarities in neural entrainment patterns and source locations between the two paradigms. These results suggest that neural entrainment engages high-level cortical mechanisms independent of linguistic content. Furthermore, we show that TRF peak amplitude serves as a potential biomarker for speech-in-noise ability, highlighting possible clinical applications.

EEG-based neurophysiological indicators in pronoun resolution using feature analysis

BACKGROUND

Pronoun resolution is a crucial aspect of language comprehension, yet its underlying neural mechanisms remain poorly understood. While previous studies have explored individual linguistic factors, a systematic analysis of Electroencephalography (EEG)-based neurophysiological indicators across different resolution cues (gender, verb bias, and discourse focus) remains unexplored, limiting our understanding of neural-cognitive processes.

NEW METHOD

We developed an approach combining ReliefF feature selection and Linear Discriminant Analysis (LDA) to analyze EEG data from twenty participants during pronoun resolution tasks. The method examined neural indicators focusing on power spectral density (PSD) and time-domain features, including Zero-Crossing Rate and Peak-to-Peak amplitude.

RESULTS

We identified crucial neural indicators across 14 channels and 4 frequency bands, highlighting PSD features in specific channels (AF3, AF4, FC6, F4, T7, T8, and O2) across theta, beta, and gamma bands. Gender-cue processing exhibited enhanced neural responses in prefrontal and temporal regions with shorter reaction times (748.77 ms) compared to verb bias (903.20 ms) and discourse focus (948.92 ms).

COMPARISON WITH EXISTING METHODS

Unlike previous studies examining individual linguistic factors, our approach simultaneously analyzed multiple resolution cues. The method achieved significant above-chance classification accuracy (49.08 % vs. 33.33 %) across three linguistic factors. This multi-factor analysis provides a more nuanced understanding of pronoun resolution processes than traditional single-factor studies.

CONCLUSIONS

Our findings suggest a more efficient, feature-driven processing mechanism for gender-cue resolution, contrasting with more complex, reasoning-dependent processing of verb semantics and discourse cues. These insights have implications for developing computational models of language processing and potential clinical applications for language disorders.

Directed Weighted EEG Connectogram Insights of One-to-One Causality for Identifying Developmental Dyslexia

Developmental dyslexia (DD) affects approximately 5-12% of learners, posing persistent challenges in reading and writing. This study presents a novel electroencephalography (EEG)-based methodology for identifying DD using two auditory stimuli modulated at 4.8[Formula: see text]Hz (prosodic) and 40[Formula: see text]Hz (phonemic). EEG signals were processed to estimate one-to-one Granger causality, yielding directed and weighted connectivity matrices. A novel Mutually Informed Correlation Coefficient (MICC) feature selection method was employed to identify the most relevant causal links, which were visualized using connectograms. Under the 4.8[Formula: see text]Hz stimulus, altered theta-band connectivity between frontal and occipital regions indicated compensatory frontal activation for prosodic processing and visual-auditory integration difficulties, while gamma-band anomalies between occipital and temporal regions suggested impaired visual-prosodic integration. Classification analysis under the 4.8[Formula: see text]Hz stimulus yielded area under the ROC curve (AUC) values of 0.92 (theta) and 0.91 (gamma band). Under the 40[Formula: see text]Hz stimulus, theta abnormalities reflected dysfunctions in integrating auditory phoneme signals with executive and motor regions, and gamma alterations indicated difficulties coordinating visual and auditory inputs for phonological decoding, with AUC values of 0.84 (theta) and 0.89 (gamma). These results support both the Temporal Sampling Framework and the Phonological Core Deficit Hypothesis. Future research should extend the range of stimuli frequencies and include more diverse cohorts to further validate these potential biomarkers.

Changes in Early Aperiodic EEG Activity Are Linked to Autism Diagnosis and Language Development in Infants With Family History of Autism

Delays in language often co-occur among toddlers diagnosed with autism. Despite the high prevalence of language delays, the neurobiology underlying such language challenges remains unclear. Prior research has shown reduced EEG power across multiple frequency bands in 3-to-6-month-old infants with an autistic sibling, followed by accelerated increases in power with age. In this study, we decompose the power spectra into aperiodic (broad band neural firing) and periodic (oscillations) activity to explore possible links between aperiodic changes in the first year of life and later language outcomes. Combining EEG data across two longitudinal studies of infants with and without autistic siblings, we assessed whether infants with an elevated familial likelihood (EFL) exhibit altered changes in both periodic and aperiodic EEG activity at 3 and 12 months of age, compared to those with a low likelihood (LL), and whether developmental change in activity is associated with language development. At 3 months of age (n = LL 59, EFL 57), we observed that EFL infants have significantly lower aperiodic activity from 6.7 to 55 Hz (p < 0.05). However, change in aperiodic activity from 3 to 12 months was significantly increased in infants with a later diagnosis of autism, compared to EFL infants without an autism diagnosis (n = LL-NoASD 41, EFL-noASD 16, EFL-ASD 16). In addition, greater increases in aperiodic offset and slope from 3 to 12 months were associated with worse language development measured at 18 months (n = 24). Findings suggest that early age-dependent changes in EEG aperiodic power may serve as potential indicators of autism and language development in infants with a family history of autism.

Neocortical and Hippocampal Theta Oscillations Track Audiovisual Integration and Replay of Speech Memories

"Are you talkin' to me?!" If you ever watched the masterpiece "Taxi Driver" directed by Martin Scorsese, you certainly recall the monologue during which Travis Bickle rehearses an imaginary confrontation in front of a mirror. While remembering this scene, you recollect a myriad of speech features across visual and auditory senses with a smooth sensation of unified memory. The aim of this study was to investigate how the fine-grained synchrony between coinciding visual and auditory features impacts brain oscillations when forming multisensory speech memories. We developed a memory task presenting participants with short synchronous or asynchronous movie clips focused on the face of speakers in real interviews, all the while undergoing magnetoencephalography recording. In the synchronous condition, the natural alignment between visual and auditory onsets was kept intact. In the asynchronous condition, auditory onsets were delayed to present lip movements and speech sounds in antiphase specifically with respect to the theta oscillation synchronizing them in the original movie. Our results first showed that theta oscillations in the neocortex and hippocampus were modulated by the level of synchrony between lip movements and syllables during audiovisual speech perception. Second, theta asynchrony between the lip movements and auditory envelope during audiovisual speech perception reduced the accuracy of subsequent theta oscillation reinstatement during memory recollection. We conclude that neural theta oscillations play a pivotal role in both audiovisual integration and memory replay of speech.

Rhythm training improves word-reading in children with dyslexia

Specific learning disorder with reading deficit (SLD-reading), or dyslexia, is one of the most common neurodevelopmental disorders. Intensive training with reading specialists is recommended, but delayed access to care is common. In this study, we tested an innovative user-friendly and accessible intervention, the medical device Mila-Learn, which is a video game based on cognitive and rhythmic training to improve phonological and reading ability. This randomized placebo-controlled trial (ClinicalTrials.gov NCT05154721) included children aged 7 to 11 y/o with SLD-reading. The children were in 2nd to 5th grade and had been in school for more than 3 years. The exclusion criteria were reading or writing remediation during the past 12 months, previous use of Mila-Learn, and severe chronic illness. The patients, who were all native French speakers, were recruited from throughout France and were randomly assigned to Mila-Learn or a matched-placebo game for an 8-week training. The primary variable was nonword decoding. The secondary variables included phonological skills, 2-min word-reading accuracy and speed, grapheme-to-phoneme conversion skills, and self-esteem. Between September 2021 and April 2023, 151 children were assigned to Mila-Learn (n = 75; male = 36; female = 39) or the placebo (n = 76; male = 42; female = 34). We registered 39 adverse events; only one was due to the protocol and was in the placebo group. We found no differences between the groups in nonword decoding in the intention-to-treat (N = 151; p = 0.39) or per-protocol analysis (N = 93; p = 0.21). However, the per-protocol analysis revealed the superiority of Mila-Learn over the placebo by 5.05 score points (95% CI 0.21; 10.3, p < 0.05) for 2-minute word-reading accuracy and by 5.44 score points (95% CI 0.57; 10.99, p < 0.05) for 2-min word-reading speed. We found no other significant effects. Mila-Learn is safe for children with SLD-reading who might benefit from this medical device.Study registration: ClinicalTrials.gov NCT05154721-13/12/2021.

Frontal-auditory cortical interactions and sensory prediction during vocal production in marmoset monkeys

The control of speech and vocal production involves the calculation of error between the intended vocal output and the resulting auditory feedback. This model has been supported by evidence that the auditory cortex (AC) is suppressed immediately before and during vocal production yet remains sensitive to differences between vocal output and altered auditory feedback. This suppression has been suggested to be the result of top-down signals about the intended vocal output, potentially originating from frontal cortical (FC) areas. However, whether FC is the source of suppressive and predictive signaling to AC during vocalization remains unknown. Here, we simultaneously recorded neural activity from both AC and FC of marmoset monkeys during self-initiated vocalizations. We found increases in neural activity in both brain areas from 1 to 0.5 s before vocal production (early pre-vocal period), specifically changes in both multi-unit activity and theta-band power. Connectivity analysis using Granger causality demonstrated that FC sends directed signaling to AC during this early pre-vocal period. Importantly, early pre-vocal activity correlated with both vocalization-induced suppression in AC as well as the structure and acoustics of subsequent calls, such as fundamental frequency. Furthermore, bidirectional auditory-frontal interactions emerged during experimentally altered vocal feedback and predicted subsequent compensatory vocal behavior. These results suggest that FC communicates with AC during vocal production, with frontal-to-auditory signals that may reflect the transmission of sensory prediction information before vocalization and bidirectional signaling during vocalization suggestive of error detection that could drive feedback-dependent vocal control.

Binaural Temporal Fine Structure Sensitivity for Children With Developmental Dyslexia

PURPOSE

Atypical temporal processing is thought to be involved in the phonological difficulties that characterize children with developmental dyslexia (DYS). The temporal sampling (TS) theory of dyslexia posits that the processing of low-frequency envelope modulations is impaired, but the processing of binaural temporal fine structure (TFS) is preserved in children with DYS.

METHOD

Binaural TFS sensitivity was assessed for children with DYS utilizing the methods developed by Flanagan et al. for typically developing (TD) children. New results for 58 children with DYS (ages 7-9.6 years) were compared with those for 30 age-matched controls (chronological age-matched [CA]) reported in Flanagan et al. Threshold frequency, that is, the highest frequency at which an interaural phase difference (IPD) of 30° or 180° could be distinguished from an IPD of 0° was determined using a two-interval forced-choice task in which the frequency was adaptively varied, with stimuli presented via headphones.

RESULTS

For those who were able to perform the task above chance, the median TFS180 thresholds were: DYS = 886 Hz; CA = 999Hz. For TFS30 thresholds: DYS = 388 Hz; CA = 442 Hz. A linear mixed-effects model with dependent variable threshold frequency and fixed effects of group (CA and DYS) and phase (180° and 30°) showed no significant difference between groups (p > .05) and no significant interaction between group and phase. Both groups performed more poorly than young typically hearing adults (p < .001) for both phases.

CONCLUSIONS

Binaural TFS sensitivity does not differ significantly for children with DYS and TD children. For both groups, the development of binaural TFS sensitivity is protracted. The results are consistent with TS theory.

EEG responses to onset-edge and steady-state segments of continuous speech under selective auditory attention modulation

Electroencephalography (EEG) signals provide valuable insights into the neural mechanisms of speech perception. However, it still remains unclear how neural responses dynamically align with continuous speech under selective attention modulation in the complex auditory environments. This study examined the evoked and induced EEG responses, their correlations with speech features, and cortical distributions for the target and ignored speech in two-talker competing scenarios. Results showed that selective attention increased the evoked EEG power for the target speech compared to the ignored speech. In contrast, the induced power indicated no significant differences. Low-frequency EEG power and phase responses reliably indexed the target speech identification amid competing streams. Cortical distribution analyses revealed that the evoked power differences between the target and ignored speech were concentrated in the central and parietal cortices. Significant induced power differences between the target and ignored speech presented only at the onset-edge segments in the left temporal cortex. Comparisons between onset-edge and steady-state segments showed the evoked power differences in the right central and temporal cortices and the induced power differences in the frontal cortex for the ignored speech. No significant differences of the cortical distribution were observed between the onset-edge and steady-state segments for the target speech. These findings underscore the distinct contributions of the evoked and induced neural activities and their cortical distributions to selective auditory attention and segment-specific speech perception.

Linear phase property of speech envelope tracking response in Heschl's gyrus and superior temporal gyrus

Understanding how the brain tracks speech during listening remains a challenge. The phase resetting hypothesis proposes that the envelope-tracking response is generated by resetting the phase of intrinsic nonlinear neural oscillations, whereas the evoked response hypothesis proposes that the envelope-tracking response is the linear superposition of transient responses evoked by a sequence of acoustic events in speech. Recent studies have demonstrated a linear phase property of the envelope-tracking response, supporting the evoked response hypothesis. However, the cortical regions aligning with the evoked response hypothesis remain unclear. To address this question, we directly recorded from the cortex using stereo-electroencephalography (SEEG) in nineteen epilepsy patients as they listened to natural speech, and we investigated whether the phase lag between the speech envelope and neural activity linearly changes across frequency. We found that the linear phase property of low-frequency (LF) (.5-40 Hz) envelope tracking was widely observed in Heschl's gyrus (HG) and superior temporal gyrus (STG), with additional sparser distribution in insula, postcentral gyrus, and precentral gyrus. Furthermore, the latency of LF envelope-tracking responses derived from phase-frequency curve exhibited an increase gradient along HG and in the posterior-to-anterior direction in STG. Our findings suggest that auditory cortex can track speech envelope in line with the evoked response hypothesis.

Dichotic training modulates phonetic feature processing in late English learners: Evidence from categorical perception and the mismatch negativity

In English, voiceless stop consonants (i.e., /p-t-k/) have long voice onset times (VOT), and voiced stop consonants (i.e., /b-d-g/) have short VOTs. In French, voiceless stop consonants have short VOTs while voiced stop consonants have pre-lag VOTs, which can make late second language (L2) learning of English difficult. The Asymmetric Sampling in Time (AST) model proposes that slow and fast cerebral rhythms dominate in the auditory cortex of the right hemisphere and left hemisphere (LH), respectively, and preferentially support processing of long and short speech events. Our study extends the application of this model to sub-phonemic levels, where short VOTs were presented to the right ear (stimulating the LH), and long VOTs were presented to the left ear (dichotic stimulation). Pre-/post-tests assessed categorical perception performance in adults trained dichotically, binaurally, or without training. Consonant perception improved in dichotic and binaural groups, with generalization of training only in the dichotic group (experiment 1). Following dichotic training, the Mismatch Negativity amplitude increased for voiced English deviants, and hemispheric dominance shifted appropriately for voiceless deviants (experiment 2). These findings demonstrate that perceptual training grounded in an AST framework can induce phonological perception changes in late English learners, suggesting promising directions for L2 perceptual training.

Vocal and articulatory speech control deficits in individuals with post-stroke aphasia

Individuals with post-stroke aphasia exhibit deficits in regulating vocal (i.e., laryngeal) pitch control during speech vowel production; however, it is not determined whether such deficits also exist when they control their supra-laryngeal speech articulators during word production. To address this question, 19 subjects with post-stroke aphasia and 20 controls were tested under an altered auditory feedback paradigm in which they received + 30% shifts in their vowel first-formant frequency during word production. In addition, 17 aphasia subjects and 19 controls from the same groups also completed steady vowel vocalizations while receiving randomized pitch shifts at ± 100 cents. Consistent with previous findings, our data showed that the magnitude of compensatory vocal responses to pitch-shifted vowel productions was significantly reduced in individuals with aphasia vs. controls. In addition, we also found that the magnitude of compensatory articulatory responses to formant-shifted vowels during word production was significantly diminished in the aphasia group compared with controls. However, no significant correlation was found between the vocal and articulatory compensatory responses to pitch and formant alterations. These findings suggest that vocal and articulatory motor speech control are regulated via independent mechanisms, and stroke-induced damage to left-hemispheric brain networks can selectively impair them in stroke survivors with aphasia.

Impact of age-related hearing loss on decompensation of left DLPFC during speech perception in noise: a combined EEG-fNIRS study

Understanding speech-in-noise is a significant challenge for individuals with age-related hearing loss (ARHL). Evidence suggests that increased activity in the frontal cortex compensates for impaired speech perception in healthy aging older adults. However, whether older adults with ARHL still show preserved compensatory function and the specific neural regulatory mechanisms underlying such compensation remains largely unclear. Here, by utilizing a synchronized EEG-fNIRS test, we investigated the neural oscillatory characteristics of the theta band and synchronous hemodynamic changes in the frontal cortex during a speech recognition task in noise. The study included healthy older adults (n = 26, aged 65.4 ± 2.8), those with mild hearing loss (n = 26, aged 66.3 ± 3.8), and those with moderate to severe hearing loss (n = 26, aged 67.5 ± 3.7). Results showed that, relative to healthy older adults, older adults with ARHL exhibited lower activation and weakened theta band neural oscillations in the left dorsolateral prefrontal cortex (DLPFC) under noisy conditions, and this decreased activity correlated with high-frequency hearing loss. Meanwhile, we found that the connectivity of the frontoparietal network was significantly reduced, which might depress the top-down articulatory prediction function affecting speech recognition performance in ARHL older adults. The results suggested that healthy aging older adults might exhibit compensatory attentional resource recruitment through a top-down auditory-motor integration mechanism. In comparison, older adults with ARHL reflected decompensation of the left DLPFC involving the frontoparietal integration network during speech recognition tasks in noise.

Asymmetric Sampling in Time: Evidence and perspectives

Auditory and speech signals are undisputedly processed in both left and right hemispheres, but this bilateral allocation is likely unequal. The Asymmetric Sampling in Time (AST) hypothesis proposed a division of labor that has its neuroanatomical basis in the distribution of neuronal ensembles with differing temporal integration constants: left auditory areas house a larger proportion of ensembles with shorter temporal integration windows (tens of milliseconds), suited to process rapidly changing signals; right auditory areas host a larger proportion with longer time constants (∼150-300 ms), ideal for slowly changing signals. Here we evaluate the large body of findings that clarifies this relationship between auditory temporal structure and functional lateralization. In this reappraisal, we unpack whether this relationship is influenced by stimulus type (speech/nonspeech), stimulus temporal extent (long/short), task engagement (high/low), or (imaging) modality (hemodynamic/electrophysiology/behavior). We find that the right hemisphere displays a clear preference for slowly changing signals whereas the left-hemispheric preference for rapidly changing signals is highly dependent on the experimental design. We consider neuroanatomical properties potentially linked to functional lateralization, contextualize the results in an evolutionary perspective, and highlight future directions.

Moving rhythmically can facilitate naturalistic speech perception in a noisy environment

The motor system is known to process temporal information, and moving rhythmically while listening to a melody can improve auditory processing. In three interrelated behavioural experiments, we demonstrate that this effect translates to speech processing. Motor priming improves the efficiency of subsequent naturalistic speech-in-noise processing under specific conditions. (i) Moving rhythmically at the lexical rate (~1.8 Hz) significantly improves subsequent speech processing compared to moving at other rates, such as the phrasal or syllabic rates. (ii) The impact of such rhythmic motor priming is not influenced by whether it is self-generated or triggered by an auditory beat. (iii) Overt lexical vocalization, regardless of its semantic content, also enhances the efficiency of subsequent speech processing. These findings provide evidence for the functional role of the motor system in processing the temporal dynamics of naturalistic speech.

Temporal Structure of Music Improves the Cortical Encoding of Speech

Long- and short-term musical training has been proposed to improve the efficiency of cortical tracking of speech, which refers to the synchronization of brain oscillations and the acoustic temporal structure of external stimuli. Here, we study how musical sequences with different rhythm structures can guide the temporal dynamics of auditory oscillations synchronized with the speech envelope. For this purpose, we investigated the effects of prior exposure to rhythmically structured musical sequences on cortical tracking of speech in Basque-Spanish bilingual adults (Experiment 1; N = 33, 22 female, Mean age = 25 years). We presented participants with sentences in Basque and Spanish preceded by musical sequences that differed in their rhythmical structure. The rhythmical structure of the musical sequences was created to (1) reflect and match the syllabic structure of the sentences, (2) reflect a regular rhythm but not match the syllabic structure of the sentences, and (3) follow an irregular rhythm. Participants' brain responses were recorded using electroencephalography, and speech-brain coherence in the delta and theta bands was calculated. Results showed stronger speech-brain coherence in the delta band in the first condition, but only for Spanish stimuli. A follow-up experiment including a subset of the initial sample (Experiment 2; N = 20) was conducted to investigate whether language-specific stimuli properties influenced the Basque results. Similar to Experiment 1, we found stronger speech-brain coherence in the delta and theta bands when the sentences were preceded by musical sequences that matched their syllabic structure. These results suggest that not only the regularity in music is crucial for influencing cortical tracking of speech, but so is adjusting this regularity to optimally reflect the rhythmic characteristics of listeners' native language(s). Despite finding some language-specific differences across frequencies, we showed that rhythm, inherent in musical signals, guides the adaptation of brain oscillations, by adapting the temporal dynamics of the oscillatory activity to the rhythmic scaffolding of the musical signal.

Prosodic Differences in Women with the FMR1 Premutation: Subtle Expression of Autism-Related Phenotypes Through Speech

Evidence suggests that carriers of FMR1 mutations (e.g., fragile X syndrome and the FMR1 premutation) may demonstrate specific phenotypic patterns shared with autism (AU), particularly in the domain of pragmatic language, which involves the use of language in social contexts. Such evidence may implicate FMR1, a high-confidence gene associated with AU, in components of the AU phenotype. Prosody (i.e., using intonation and rhythm in speech to express meaning) is a pragmatic feature widely impacted in AU. Prosodic differences have also been observed in unaffected relatives of autistic individuals and in those with fragile X syndrome, although prosody has not been extensively studied among FMR1 premutation carriers. This study investigated how FMR1 variability may specifically influence prosody by examining the prosodic characteristics and related neural processing of prosodic features in women carrying the FMR1 premutation (PM). In Study 1, acoustic measures of prosody (i.e., in intonation and rhythm) were examined in speech samples elicited from a semi-structured narrative task. Study 2 examined the neural frequency following response (FFR) as an index of speech prosodic processing. Findings revealed differences in the production of intonation and rhythm in PM carriers relative to controls, with patterns that parallel differences identified in parents of autistic individuals. No differences in neural processing of prosodic cues were found. Post hoc analyses further revealed associations between speech rhythm and FMR1 variation (number of CGG repeats) among PM carriers. Together, the results suggest that FMR1 may play a role in speech prosodic phenotypes, at least in speech production, contributing to a deeper understanding of AU-related speech and language phenotypes among FMR1 mutation carriers.

Hierarchical dynamic coding coordinates speech comprehension in the human brain

Speech comprehension involves transforming an acoustic waveform into meaning. To do so, the human brain generates a hierarchy of features that converts the sensory input into increasingly abstract language properties. However, little is known about how rapid incoming sequences of hierarchical features are continuously coordinated. Here, we propose that each language feature is supported by a dynamic neural code, which represents the sequence history of hierarchical features in parallel. To test this 'Hierarchical Dynamic Coding' (HDC) hypothesis, we use time-resolved decoding of brain activity to track the construction, maintenance, and update of a comprehensive hierarchy of language features spanning phonetic, word form, lexical-syntactic, syntactic and semantic representations. For this, we recorded 21 native English participants with magnetoencephalography (MEG), while they listened to two hours of short stories in English. Our analyses reveal three main findings. First, the brain represents and simultaneously maintains a sequence of hierarchical features. Second, the duration of these representations depends on their level in the language hierarchy. Third, each representation is maintained by a dynamic neural code, which evolves at a speed commensurate with its corresponding linguistic level. This HDC preserves the maintenance of information over time while limiting destructive interference between successive features. Overall, HDC reveals how the human brain maintains and updates the continuously unfolding language hierarchy during natural speech comprehension, thereby anchoring linguistic theories to their biological implementations.

Mouth rhythm as a "packaging mechanism" of information in speech: A proof of concept

This paper postulated and tested the possibility that the mouth rhythm functions as a "packaging mechanism" of information in speech. Cross-spectral analysis between two time series of mouth aperture size [parameterized as sample-by-sample interlip distances, i.e., o(t)] and information variations in speech [parameterized as frame-by-frame spectral entropy values, i.e., h(t)] was employed to reveal their underlying spectro-temporal relationship. Using a corpus containing more than 1000 utterances produced by a typical British English speaker, it was observed that both signals share slow recurring rates corresponding to the stress and syllable, with a slight phase lag of h(t) behind o(t) in the vicinity of 5 Hz.

Optimal Frequency for Cranial Electromagnetic Field Stimulation

Background Communication among neurons generates electromagnetic fields (EMFs) that can be measured through a noninvasive, portable helmet equipped with 20 sensors. The EMF data reveal a variety of EMF patterns that have yet to be elucidated. Understanding a propagated frequency from the brain and its subunits can assist with diagnosing the brain and its subunits' function and treatment. Here, the authors provide an interpretation of the EMF patterns with an emphasis on frequency. Methods From January 2025 to February 2025, a prospective clinical study was conducted to enroll patients greater than 18 years old diagnosed with atraumatic and traumatic brain injury whose EMFs, which were collected using a helmet equipped with 20 sensors, were obtained within 24 hours of presentation. EMF data were collected using DAQami software (DATAQ Instruments Inc., Akron, Ohio, United States) and analyzed using fast Fourier transformation with Igor Pro 8 software (WaveMetrics Inc., Lake Oswego, Oregon, United States). Based on each patient's clinical presentations and/or radiographic findings, the sensors of interest, their opposing sensors, and frequencies of interest (FOIs) were selected. Results A total of 10 patients were enrolled with a mean age of 47.1 years. Mechanisms of injury included spontaneous hypertensive intracranial hemorrhage (one patient) and head trauma after a motor vehicle collision, dirt bike accident, or ground-level fall (nine patients). Radiographic findings included spontaneous basal ganglia hemorrhage (one patient), isolated traumatic subdural hematoma (one patient), traumatic subarachnoid hemorrhage (one patient), and no intracranial abnormalities (seven patients). The following targeted FOIs were found: 5.2 Hz, 7.3 Hz, 7.6 Hz, 7.7 Hz, 7.9 Hz, 8.3 Hz, 8.6 Hz, 8.7 Hz, 9.5 Hz, and 10.4 Hz. Conclusions EMF of the human brain reveals changes in neuronal activities in atraumatic and traumatic brain injury patients. This information allows for the localization of sites of brain injuries and the selection of frequencies that can be used for understanding the EMF frequency and function on the macroscopic level as well as at the cellular level. This specific information can then be utilized for stimulation to modulate the changes in neuronal, circuit, and brain function activities. Our frequency selection technique enables more precise, tailored, and potentially more effective treatment aiming to restore EMF activity.

Localization of Brain Injuries Using Cranial Electromagnetic Fields

BACKGROUND

Atraumatic brain injury and traumatic brain injury (TBI) have been demonstrated to be associated with changes in brain electromagnetic field (EMF) activity due to alterations in the structure and function of neural circuitry. Modulation of abnormal EMF activity through EMF stimulation may promote neural regeneration and may be more beneficial when the specific change in EMF frequency that is correlated with either computed tomography (CT) imaging changes, neurological changes, or both can be precisely localized. The authors investigate the efficacy and feasibility of a noninvasive portable helmet with sensors and built-in signal generators to measure and localize specific changes with frequency and amplitude from brain EMF for both atraumatic and TBI patients.

METHODS

This prospective clinical study was conducted from January 2025 to February 2025 and enrolled patients greater than 18 years old diagnosed with atraumatic and TBI, including negative image concussion. Baseline EMF activity was recorded using a helmet equipped with 20 sensor stimulators. Localization of EMF activity was determined based on sensor activity corresponding to the patient's neurological deficits on exam and/or structural lesion(s) on CT. EMF data were collected using the DAQami software (Dataq Instruments, Akron, OH) and analyzed using fast Fourier transformation with the Igor Pro 8 software (Wavemetrics Inc., Lake Oswego, OR) to localize normal and abnormal brain EMF signals by comparing opposing and adjacent sensors.

RESULTS

Ten patients were enrolled in this study with a mean age of 47.1 years. Mechanisms of injury included spontaneous hypertensive intracranial hemorrhage (one patient) and head trauma after motor vehicle collision (auto vs. auto; auto vs. motorcycle; and auto vs. pedestrian), dirt bike accident, and ground-level fall (nine patients). Radiographic findings included spontaneous basal ganglia hemorrhage (one patient), isolated traumatic subdural hematoma (one patient), traumatic subarachnoid hemorrhage (one patient), and no intracranial abnormalities (seven patients). Abnormal EMF activity was recorded and correlated with neurological deficits on exam, CT findings, or both, demonstrating the usefulness of EMF in localizing brain injuries.

CONCLUSIONS

These results demonstrate the efficacy and feasibility of utilizing a noninvasive portable helmet for real-time EMF recording and localization of brain abnormalities in atraumatic and TBI patients, including image-negative concussions. EMF measurements may aid in monitoring recovery after atraumatic brain injury and TBI and enable the clinician to tailor treatment plans based on the patient's unique brain EMF patterns.

Sequence chunking through neural encoding of ordinal positions

Grouping sensory events into chunks is an efficient strategy to integrate information across long sequences such as speech, music, and complex movements. Although chunks can be constructed based on diverse cues (e.g., sensory features, statistical patterns, internal knowledge) recent studies have consistently demonstrated that the chunks constructed by different cues are all tracked by low-frequency neural dynamics. Here, I review evidence that chunking cues drive low-frequency activity in modality-dependent networks, which interact to generate chunk-tracking activity in broad brain areas. Functionally, this work suggests that a core computation underlying sequence chunking may assign each event its ordinal position within a chunk and that this computation is causally implemented by chunk-tracking neural activity during predictive sequence chunking.

Learning to operate an imagined speech Brain-Computer Interface involves the spatial and frequency tuning of neural activity

Brain-Computer Interfaces (BCI) will revolutionize the way people with severe impairment of speech production can communicate. While current efforts focus on training classifiers on vast amounts of neurophysiological signals to decode imagined speech, much less attention has been given to users' ability to adapt their neural activity to improve BCI-control. To address whether BCI-control improves with training and characterize the underlying neural dynamics, we trained 15 healthy participants to operate a binary BCI system based on electroencephalography (EEG) signals through syllable imagery for five consecutive days. Despite considerable interindividual variability in performance and learning, a significant improvement in BCI-control was globally observed. Using a control experiment, we show that a continuous feedback about the decoded activity is necessary for learning to occur. Performance improvement was associated with a broad EEG power increase in frontal theta activity and focal enhancement in temporal low-gamma activity, showing that learning to operate an imagined-speech BCI involves dynamic changes in neural features at different spectral scales. These findings demonstrate that combining machine and human learning is a successful strategy to enhance BCI controllability.

Resting-State Electroencephalogram and Speech Perception in Young Children with Developmental Language Disorder

BACKGROUND/OBJECTIVES

Endogenous oscillations reflect the spontaneous activity of brain networks involved in cognitive processes. In adults, endogenous activity across different bands correlates with, and can even predict, language and speech perception processing. However, it remains unclear how this activity develops in children with typical and atypical development.

METHODS

We investigated differences in resting-state EEG between preschoolers with developmental language disorder (DLD), their age-matched controls with typical language development (TLD), and a group of adults.

RESULTS

We observed significantly lower oscillatory power in adults than in children (p < 0.001 for all frequency bands), but no differences between the groups of children in power or hemispheric lateralisation, suggesting that oscillatory activity reflects differences in age, but not in language development. The only measure that differed between the children's groups was theta/alpha band ratio (p = 0.004), which was significantly smaller in TLD than in DLD children, although this was an incidental finding. Behavioural results also did not fully align with previous research, as TLD children performed better in the filtered speech test (p = 0.01), but not in the speech-in-babble one, and behavioural test scores did not correlate with high-frequency oscillations, lateralisation indices, or band ratio measures.

CONCLUSIONS

We discuss the suitability of these resting-state EEG measures to capture group-level differences between TLD/DLD preschoolers and the relevance of our findings for future studies investigating neural markers of typical and atypical language development.

Atypical audio-visual neural synchrony and speech processing in early autism

BACKGROUND

Children with Autism Spectrum disorder (ASD) often exhibit communication difficulties that may stem from basic auditory temporal integration impairment but also be aggravated by an audio-visual integration deficit, resulting in a lack of interest in face-to-face communication. This study addresses whether speech processing anomalies in young autistic children (mean age 3.09-year-old) are associated with alterations of audio-visual temporal integration.

METHODS

We used high-density electroencephalography (HD-EEG) and eye tracking to record brain activity and gaze patterns in 31 children with ASD (6 females) and 33 typically developing (TD) children (11 females), while they watched cartoon videos. Neural responses to temporal audio-visual stimuli were analyzed using Temporal Response Functions model and phase analyses for audiovisual temporal coordination.

RESULTS

The reconstructability of speech signals from auditory responses was reduced in children with ASD compared to TD, but despite more restricted gaze patterns in ASD it was similar for visual responses in both groups. Speech reception was most strongly affected when visual speech information was also present, an interference that was not seen in TD children. These differences were associated with a broader phase angle distribution (exceeding pi/2) in the EEG theta range in children with ASD, signaling reduced reliability of audio-visual temporal alignment.

CONCLUSION

These findings show that speech processing anomalies in ASD do not stand alone and that they are associated already at a very early development stage with audio-visual imbalance with poor auditory response encoding and disrupted audio-visual temporal coordination.

Statistical learning beyond words in human neonates

Interest in statistical learning in developmental studies stems from the observation that 8-month-olds were able to extract words from a monotone speech stream solely using the transition probabilities (TP) between syllables (Saffran et al., 1996). A simple mechanism was thus part of the human infant's toolbox for discovering regularities in language. Since this seminal study, observations on statistical learning capabilities have multiplied across domains and species, challenging the hypothesis of a dedicated mechanism for language acquisition. Here, we leverage the two dimensions conveyed by speech -speaker identity and phonemes- to examine (1) whether neonates can compute TPs on one dimension despite irrelevant variation on the other and (2) whether the linguistic dimension enjoys an advantage over the voice dimension. In two experiments, we exposed neonates to artificial speech streams constructed by concatenating syllables while recording EEG. The sequence had a statistical structure based either on the phonetic content, while the voices varied randomly (Experiment 1) or on voices with random phonetic content (Experiment 2). After familiarisation, neonates heard isolated duplets adhering, or not, to the structure they were familiarised with. In both experiments, we observed neural entrainment at the frequency of the regularity and distinct Event-Related Potentials (ERP) to correct and incorrect duplets, highlighting the universality of statistical learning mechanisms and suggesting it operates on virtually any dimension the input is factorised. However, only linguistic duplets elicited a specific ERP component, potentially an N400 precursor, suggesting a lexical stage triggered by phonetic regularities already at birth. These results show that, from birth, multiple input regularities can be processed in parallel and feed different higher-order networks.

An efficient deep learning approach for automatic speech recognition using EEG signals

Electroencephalogram (EEG) signals enhance human-machine interaction but pose challenges in speech recognition due to noise and complexity. This study proposes an Efficient Deep Learning Approach (EDLA) integrating the Gannet Optimization Algorithm (GOA) and Elman Recurrent Neural Network (ERNN) for speaker identification. EEG data is preprocessed using a Savitzky-Golay filter, and key features are selected via recursive feature elimination. Evaluated on the Kara One dataset, EDLA achieves 95.2% accuracy, outperforming baseline methods. This framework advances EEG based speech recognition aiding brain-computer interfaces and assistive technologies for individuals with speech disorders.

Dynamic changes in large-scale functional connectivity prior to stimulation determine performance in a multisensory task

Complex behavior and task execution require fast changes of local activity and functional connectivity in cortical networks at multiple scales. The roles that changes of power and connectivity play during these processes are still not well understood. Here, we study how fluctuations of functional cortical coupling across different brain areas determine performance in an audiovisual, lateralized detection task in the ferret. We hypothesized that dynamic variations in the network's state determine the animals' performance. We evaluated these by quantifying changes of local power and of phase coupling across visual, auditory and parietal regions. While power for hit and miss trials showed significant differences only during stimulus and response onset, phase coupling already differed before stimulus onset. An analysis of principal components in coupling at the single-trial level during this period allowed us to reveal the subnetworks that most strongly determined performance. Whereas higher global phase coupling of visual and auditory regions to parietal cortex was predictive of task performance, a second component revealed a reduction in coupling between subnetworks of different sensory modalities, probably to allow a better detection of the unimodal signals. Furthermore, we observed that long-range coupling became more predominant during the task period compared to the pre-stimulus baseline. Taken together, our results show that fluctuations in the network state, as reflected in large-scale coupling, are key determinants of the animals' behavior.

Simulating the impact of white matter connectivity on processing time scales using brain network models

The capacity of the brain to process input across temporal scales is exemplified in human narrative, which requires integration of information ranging from words, over sentences to long paragraphs. It has been shown that this processing is distributed in a hierarchy across multiple areas in the brain with areas close to the sensory cortex, processing on a faster time scale than areas in associative cortex. In this study we used reservoir computing with human derived connectivity to investigate the effect of the structural connectivity on time scales across brain regions during a narrative task paradigm. We systematically tested the effect of removal of selected fibre bundles (IFO, ILF, MLF, SLF I/II/III, UF, AF) on the processing time scales across brain regions. We show that long distance pathways such as the IFO provide a form of shortcut whereby input driven activation in the visual cortex can directly impact distant frontal areas. To validate our model we demonstrated significant correlation of our predicted time scale ordering with empirical results from the intact/scrambled narrative fMRI task paradigm. This study emphasizes structural connectivity's role in brain temporal processing hierarchies, providing a framework for future research on structure and neural dynamics across cognitive tasks.

Relational neuroscience: Insights from hyperscanning research

Humans are highly social, typically without this ability requiring noticeable efforts. Yet, such social fluency poses challenges both for the human brain to compute and for scientists to study. Over the last few decades, neuroscientific research of human sociality has witnessed a shift in focus from single-brain analysis to complex dynamics occurring across several brains, posing questions about what these dynamics mean and how they relate to multifaceted behavioural models. We propose the term 'Relational Neuroscience' to collate the interdisciplinary research field devoted to modelling the inter-brain dynamics subserving human connections, spanning from real-time joint experiences to long-term social bonds. Hyperscanning, i.e., simultaneously measuring brain activity from multiple individuals, has proven to be a highly promising technique to investigate inter-brain dynamics. Here, we discuss how hyperscanning can help investigate questions within the field of Relational Neuroscience, considering a variety of subfields, including cooperative interactions in dyads and groups, empathy, attachment and bonding, and developmental neuroscience. While presenting Relational Neuroscience in the light of hyperscanning, our discussion also takes into account behaviour, physiology and endocrinology to properly interpret inter-brain dynamics within social contexts. We consider the strengths but also the limitations and caveats of hyperscanning to answer questions about interacting people. The aim is to provide an integrative framework for future work to build better theories across a variety of contexts and research subfields to model human sociality.

Adults With Dyslexia Use Internalised Beat Cues Less Than Controls When Estimating Interval Length

Difficulties in both duration and beat-based time perception are common in individuals with dyslexia (DD). It is also known that internalised beat cues may aid in duration processing. This study investigated whether the difficulties in duration processing among DD stem from their inability to utilise internal beat cues. Participants with and without dyslexia estimated intervals ranging from 500 ms to 10 s. In the beat cue condition, participants listened to a sequence of 500 ms beats before the interval, and in the no beat cue condition, they were exposed to silence while EEG was recorded. Interestingly, the two groups did not differ in duration estimation performance, but they did differ in their utilisation of beat cues, with DD showing less sensitivity to these, whether the impact was negative (cues before shorter intervals) or positive (before longer intervals). Brainwave entrainment to the target frequency was significantly higher compared with entrainment to a non-target frequency, and cross-group differences were null. Our findings suggest that DD may have difficulties either in retaining the beat when it is no longer audible, or in using the internalised beat for duration estimation. Nevertheless, they can achieve comparable accuracy to neurotypical adults, possibly through compensatory strategies.

The Impact of Selective Attention and Musical Training on the Cortical Speech Tracking in the Delta and Theta Frequency Bands

Oral communication regularly takes place amidst background noise, requiring the ability to selectively attend to a target speech stream. Musical training has been shown to be beneficial for this task. Regarding the underlying neural mechanisms, recent studies showed that the speech envelope is tracked by neural activity in auditory cortex, which plays a role in the neural processing of speech, including speech in noise. The neural tracking occurs predominantly in two frequency bands, the delta and the theta bands. However, much regarding the specifics of these neural responses, as well as their modulation through musical training, still remain unclear. Here, we investigated the delta- and theta-band cortical tracking of the speech envelope of target and distractor speech using magnetoencephalography (MEG) recordings. We thereby assessed both musicians and nonmusicians to explore potential differences between these groups. The cortical speech tracking was quantified through source-reconstructing the MEG data and subsequently relating the speech envelope in a certain frequency band to the MEG data using linear models. We thereby found the theta-band tracking to be dominated by early responses with comparable magnitudes for target and distractor speech, whereas the delta band tracking exhibited both earlier and later responses that were modulated by selective attention. Almost no significant differences emerged in the neural responses between musicians and nonmusicians. Our findings show that only the speech tracking in the delta but not in the theta band contributes to selective attention, but that this mechanism is essentially unaffected by musical training.

Beta oscillations predict the envelope sharpness in a rhythmic beat sequence

Periodic sensory inputs entrain oscillatory brain activity, reflecting a neural mechanism that might be fundamental to temporal prediction and perception. Most environmental rhythms and patterns in human behavior, such as walking, dancing, and speech do not, however, display strict isochrony but are instead quasi-periodic. Research has shown that neural tracking of speech is driven by modulations of the amplitude envelope, especially via sharp acoustic edges, which serve as prominent temporal landmarks. In the same vein, research on rhythm processing in music supports the notion that perceptual timing precision varies systematically with the sharpness of acoustic onset edges, conceptualized in the beat bin hypothesis. Increased envelope sharpness induces increased precision in localizing a sound in time. Despite this tight relationship between envelope shape and temporal processing, it is currently unknown how the brain uses predictive information about envelope features to optimize temporal perception. With the current EEG study, we show that the predicted sharpness of the amplitude envelope is encoded by pre-target neural activity in the beta band (15-25 Hz), and has an impact on the temporal perception of target sounds. We used probabilistic sound cues in a timing judgment task to inform participants about the sharpness of the amplitude envelope of an upcoming target sound embedded in a beat sequence. The predictive information about the envelope shape modulated task performance and pre-target beta power. Interestingly, these conditional beta-power modulations correlated positively with behavioral performance in the timing judgment task and with perceptual temporal precision in a click-alignment task. This study provides new insight into the neural processes underlying prediction of the sharpness of the amplitude envelope during beat perception, which modulate the temporal perception of sounds. This finding could reflect a process that is involved in temporal prediction, exerting top-down control on neural entrainment via the prediction of acoustic edges in the auditory stream.

The functional role of oscillatory dynamics in neocortical circuits: A computational perspective

The dynamics of neuronal systems are characterized by hallmark features such as oscillations and synchrony. However, it has remained unclear whether these characteristics are epiphenomena or are exploited for computation. Due to the challenge of selectively interfering with oscillatory network dynamics in neuronal systems, we simulated recurrent networks of damped harmonic oscillators in which oscillatory activity is enforced in each node, a choice well supported by experimental findings. When trained on standard pattern recognition tasks, these harmonic oscillator recurrent networks (HORNs) outperformed nonoscillatory architectures with respect to learning speed, noise tolerance, and parameter efficiency. HORNs also reproduced a many characteristic features of neuronal systems, such as the cerebral cortex and the hippocampus. In trained HORNs, stimulus-induced interference patterns holistically represent the result of comparing sensory evidence with priors stored in recurrent connection weights, and learning-induced weight changes are compatible with Hebbian principles. Implementing additional features characteristic of natural networks, such as heterogeneous oscillation frequencies, inhomogeneous conduction delays, and network modularity, further enhanced HORN performance without requiring additional parameters. Taken together, our model allows us to give plausible a posteriori explanations for features of natural networks whose computational role has remained elusive. We conclude that neuronal systems are likely to exploit the unique dynamics of recurrent oscillator networks whose computational superiority critically depends on the oscillatory patterning of their nodal dynamics. Implementing the proposed computational principles in analog hardware is expected to enable the design of highly energy-efficient and self-adapting devices that could ideally complement existing digital technologies.

Attentional Inhibition Ability Predicts Neural Representation During Challenging Auditory Streaming

Focusing on a single source within a complex auditory scene is challenging. M/EEG-based auditory attention detection (AAD) allows to detect which stream an individual is attending to within a set of multiple concurrent streams. The high interindividual variability in the auditory attention detection performance often is attributed to physiological factors and signal-to-noise ratio of neural data. We hypothesize that executive functions-in particular sustained attention, working memory, and attentional inhibition-may partly explain the variability in auditory attention detection performance, because they support the cognitive processes required when listening to complex auditory scenes. We chose a particularly challenging auditory scene by presenting dichotically polyphonic classical piano excerpts that lasted 1 min each. Two different excerpts were presented simultaneously, one in each ear. Forty-one participants, with different degrees of musical expertise, listened to these complex auditory scenes focusing on one ear while we recorded the EEG. Participants also completed several tasks assessing executive functions. As expected, EEG-based auditory attention detection was greater for attended than unattended stimuli. Importantly, attentional inhibition ability did explain 6% of the reconstruction accuracy and 8% of the classification accuracy. No other executive function was a significant predictor of reconstruction or classification accuracies. No clear effect of musical expertise was found on reconstruction and classification performance. In conclusion, cognitive factors seem to impact the robustness of the neural auditory representation and hence the performance of EEG-based decoding approaches. Taking advantage of this relation could be useful to improve next-generation hearing aids.

Investigating the Neural Basis of the Loud-first Principle of the Iambic-Trochaic Law

The perception of rhythmic patterns is crucial for the recognition of words in spoken languages, yet it remains unclear how these patterns are represented in the brain. Here, we tested the hypothesis that rhythmic patterns are encoded by neural activity phase-locked to the temporal modulation of these patterns in the speech signal. To test this hypothesis, we analyzed EEGs evoked with long sequences of alternating syllables acoustically manipulated to be perceived as a series of different rhythmic groupings in English. We found that the magnitude of the EEG at the syllable and grouping rates of each sequence was significantly higher than the noise baseline, indicating that the neural parsing of syllables and rhythmic groupings operates at different timescales. Distributional differences between the scalp topographies associated with each timescale suggests a further mechanistic dissociation between the neural segmentation of syllables and groupings. In addition, we observed that the neural tracking of louder syllables, which in trochaic languages like English are associated with the beginning of rhythmic groupings, was more robust than the neural tracking of softer syllables. The results of further bootstrapping and brain-behavior analyses indicate that the perception of rhythmic patterns is modulated by the magnitude of grouping alternations in the neural signal. These findings suggest that the temporal coding of rhythmic patterns in stress-based languages like English is supported by temporal regularities that are linguistically relevant in the speech signal.

[The contribution of language biomarkers in the clinical evaluation of the serious game Poppins]

Dyslexia is a neurodevelopmental disorder that can be diagnosed as early as the end of first grade, thanks to behavioral assessments. However, neurophysiological markers (biomarkers) could enable early detection of children at risk of developing this specific learning disorder. These various EEG biomarkers, once adjusted to clinical constraints, could constitute reliable tools to assist healthcare professionals in the early detection of dyslexia or other neurodevelopmental disorders. They also open up new prospects in the development of digital tools for early diagnosis and remediation.

Exploring Relevant Features for EEG-Based Investigation of Sound Perception in Naturalistic Soundscapes

A comprehensive analysis of everyday sound perception can be achieved using electroencephalography (EEG) with the concurrent acquisition of information about the environment. While extensive research has been dedicated to speech perception, the complexities of auditory perception within everyday environments, specifically the types of information and the key features to extract, remain less explored. Our study aims to systematically investigate the relevance of different feature categories: discrete sound-identity markers, general cognitive state information, and acoustic representations, including discrete sound onset, the envelope, and mel-spectrogram. Using continuous data analysis, we contrast different features in terms of their predictive power for unseen data and thus their distinct contributions to explaining neural data. For this, we analyze data from a complex audio-visual motor task using a naturalistic soundscape. The results demonstrated that the feature sets that explain the most neural variability were a combination of highly detailed acoustic features with a comprehensive description of specific sound onsets. Furthermore, it showed that established features can be applied to complex soundscapes. Crucially, the outcome hinged on excluding periods devoid of sound onsets in the analysis in the case of the discrete features. Our study highlights the importance to comprehensively describe the soundscape, using acoustic and non-acoustic aspects, to fully understand the dynamics of sound perception in complex situations. This approach can serve as a foundation for future studies aiming to investigate sound perception in natural settings.

The impact of speaker accent on discourse processing: A frequency investigation

Previous studies indicate differences in native and foreign speech processing (Lev-Ari, 2018), with mixed evidence for differences between dialectal and foreign accent processing (Adank, Evans, Stuart-Smith, & Scott, 2009; Floccia et al., 2006, 2009; Girard, Floccia, & Goslin, 2008). Two theories have been proposed: The Perceptual Distance Hypothesis suggests that dialectal accent processing is an attenuated version of foreign accent processing (Clarke & Garrett, 2004), while the Different Processes Hypothesis argues that foreign and dialectal accents are processed via distinct mechanisms (Floccia, Butler, Girard, & Goslin, 2009). A recent single-word ERP study suggested flexibility in these mechanisms (Thomas, Martin, & Caffarra, 2022). The present study deepens this investigation by investigating differences in native, dialectal, and foreign accent processing across frequency bands during extended speech. Electroencephalographic data was recorded from 30 participants who listened to dialogues of approximately six minutes spoken in native, dialectal and foreign accents. Power spectral density estimation (1-35 Hz) was performed. Linear mixed models were done in frequency windows of particular relevance to discourse processing. Frequency bands associated with phoneme [gamma], syllable [theta], and prosody [delta] were considered along with those of general cognitive mechanisms [alpha and beta]. Results show power differences in the Gamma frequency range. While in higher frequency ranges foreign accent processing is differentiated from power amplitudes of native and dialectal accent processing, in low frequencies we do not see any accent-related power amplitude modulations. This suggests that there may be a difference in phoneme processing for native accent types and foreign accent, while we speculate that top-down mechanisms during discourse processing may mitigate the effects observed with short units of speech.

Frequency-specific cortico-subcortical interaction in continuous speaking and listening

Speech production and perception involve complex neural dynamics in the human brain. Using magnetoencephalography, our study explores the interaction between cortico-cortical and cortico-subcortical connectivities during these processes. Our connectivity findings during speaking revealed a significant connection from the right cerebellum to the left temporal areas in low frequencies, which displayed an opposite trend in high frequencies. Notably, high-frequency connectivity was absent during the listening condition. These findings underscore the vital roles of cortico-cortical and cortico-subcortical connections within the speech production and perception network. The results of our new study enhance our understanding of the complex dynamics of brain connectivity during speech processes, emphasizing the distinct frequency-based interactions between various brain regions.

The human auditory cortex concurrently tracks syllabic and phonemic timescales via acoustic spectral flux

Dynamical theories of speech processing propose that the auditory cortex parses acoustic information in parallel at the syllabic and phonemic timescales. We developed a paradigm to independently manipulate both linguistic timescales, and acquired intracranial recordings from 11 patients who are epileptic listening to French sentences. Our results indicate that (i) syllabic and phonemic timescales are both reflected in the acoustic spectral flux; (ii) during comprehension, the auditory cortex tracks the syllabic timescale in the theta range, while neural activity in the alpha-beta range phase locks to the phonemic timescale; (iii) these neural dynamics occur simultaneously and share a joint spatial location; (iv) the spectral flux embeds two timescales-in the theta and low-beta ranges-across 17 natural languages. These findings help us understand how the human brain extracts acoustic information from the continuous speech signal at multiple timescales simultaneously, a prerequisite for subsequent linguistic processing.

Hearing and cognitive decline in aging differentially impact neural tracking of context-supported versus random speech across linguistic timescales

Cognitive decline and hearing loss are common in older adults and often co-occur while investigated separately, affecting the neural processing of speech. This study investigated the interaction between cognitive decline, hearing loss, and contextual cues in speech processing. Participants aged 60 years and older were assessed for cognitive decline using the Montreal Cognitive Assessment and for hearing ability using a four-frequency pure tone average. They listened to in-house-designed matrix-style sentences that either provided supportive context or were random, while we recorded their electroencephalography. Neurophysiological responses were analyzed through auditory evoked potentials and speech tracking at different linguistic timescales (i.e., phrase, word, syllable and phoneme rate) using phase-locking values. The results showed that cognitive decline was associated with decreased response accuracy in a speech recognition task. Cognitive decline significantly impacted the P2 component of auditory evoked potentials, while hearing loss influenced speech tracking at the word and phoneme rates, but not at the phrase or syllable rates. Contextual cues enhanced speech tracking at the syllable rate. These findings suggest that cognitive decline and hearing loss differentially affect the neural mechanisms underlying speech processing, with contextual cues playing a significant role in enhancing syllable rate tracking. This study emphasises the importance of considering both cognitive and auditory factors when studying speech processing in older people and highlights the need for further research to investigate the interplay between cognitive decline, hearing loss and contextual cues in speech processing.

Infant low-frequency EEG cortical power, cortical tracking and phase-amplitude coupling predicts language a year later

Cortical signals have been shown to track acoustic and linguistic properties of continuous speech. This phenomenon has been measured in both children and adults, reflecting speech understanding by adults as well as cognitive functions such as attention and prediction. Furthermore, atypical low-frequency cortical tracking of speech is found in children with phonological difficulties (developmental dyslexia). Accordingly, low-frequency cortical signals may play a critical role in language acquisition. A recent investigation with infants Attaheri et al., 2022 [1] probed cortical tracking mechanisms at the ages of 4, 7 and 11 months as participants listened to sung speech. Results from temporal response function (TRF), phase-amplitude coupling (PAC) and dynamic theta-delta power (PSD) analyses indicated speech envelope tracking and stimulus-related power (PSD) for delta and theta neural signals. Furthermore, delta- and theta-driven PAC was found at all ages, with theta phases displaying stronger PAC with high-frequency amplitudes than delta. The present study tests whether these previous findings replicate in the second half of the full cohort of infants (N = 122) who were participating in this longitudinal study (first half: N = 61, (1); second half: N = 61). In addition to demonstrating good replication, we investigate whether cortical tracking in the first year of life predicts later language acquisition for the full cohort (122 infants recruited, 113 retained) using both infant-led and parent-estimated measures and multivariate and univariate analyses. Increased delta cortical tracking in the univariate analyses, increased ~2Hz PSD power and stronger theta-gamma PAC in both multivariate and univariate analyses were related to better language outcomes using both infant-led and parent-estimated measures. By contrast, increased ~4Hz PSD power in the multi-variate analyses, increased delta-beta PAC and a higher theta/delta power ratio in the multi-variate analyses were related to worse language outcomes. The data are interpreted within a "Temporal Sampling" framework for developmental language trajectories.

Concurrent processing of the prosodic hierarchy is supported by cortical entrainment and phase-amplitude coupling

Models of phonology posit a hierarchy of prosodic units that is relatively independent from syntactic structure, requiring its own parsing. It remains unexplored how this prosodic hierarchy is represented in the brain. We investigated this foundational question by means of an electroencephalography (EEG) study. Thirty young adults listened to German sentences containing manipulations at different levels of the prosodic hierarchy. Evaluating speech-to-brain cortical entrainment and phase-amplitude coupling revealed that prosody's hierarchical structure is maintained at the neural level during spoken language comprehension. The faithfulness of this tracking varied as a function of the hierarchy's degree of intactness as well as systematic interindividual differences in audio-motor synchronization abilities. The results underscore the role of complex oscillatory mechanisms in configuring the continuous and hierarchical nature of the speech signal and situate prosody as a structure indispensable from theoretical perspectives on spoken language comprehension in the brain.

Unraveling Brain Synchronisation Dynamics by Explainable Neural Networks using EEG Signals: Application to Dyslexia Diagnosis

The electrical activity of the neural processes involved in cognitive functions is captured in EEG signals, allowing the exploration of the integration and coordination of neuronal oscillations across multiple spatiotemporal scales. We have proposed a novel approach that combines the transformation of EEG signal into image sequences, considering cross-frequency phase synchronisation (CFS) dynamics involved in low-level auditory processing, with the development of a two-stage deep learning model for the detection of developmental dyslexia (DD). This deep learning model exploits spatial and temporal information preserved in the image sequences to find discriminative patterns of phase synchronisation over time achieving a balanced accuracy of up to 83%. This result supports the existence of differential brain synchronisation dynamics between typical and dyslexic seven-year-old readers. Furthermore, we have obtained interpretable representations using a novel feature mask to link the most relevant regions during classification with the cognitive processes attributed to normal reading and those corresponding to compensatory mechanisms found in dyslexia.

Early- and Late-Stage Auditory Processing of Speech Versus Non-Speech Sounds in Children With Autism Spectrum Disorder: An ERP and Oscillatory Activity Study

Individuals with autism spectrum disorder (ASD) often exhibit greater sensitivity to non-speech sounds, reduced sensitivity to speech, and increased variability in cortical activity during auditory speech processing. We assessed differences in cortical responses and variability in early and later processing stages of auditory speech versus non-speech sounds in typically developing (TD) children and children with ASD. Twenty-eight 4- to 9-year-old children (14 ASDs) listened to speech and non-speech sounds during an electroencephalography session. We measured peak amplitudes for early (P2) and later (P3a) stages of auditory processing and inter-trial theta phase coherence as a marker of cortical variability. TD children were more sensitive to speech sounds during early and later processing stages than ASD children, reflected in larger P2 and P3a amplitudes. Individually, twice as many TD children showed reliable differentiation between speech and non-speech sounds compared to children with ASD. Children with ASD showed greater intra-individual variability in theta responses to speech sounds during early and later processing stages. Children with ASD show atypical auditory processing of fundamental speech sounds, perhaps due to reduced and more variable cortical activation. These atypicalities in the consistency of cortical responses to fundamental speech features may impact the development of cortical networks and have downstream effects on more complex forms of language processing.

A Bio-Inspired Spiking Attentional Neural Network for Attentional Selection in the Listening Brain

Humans show a remarkable ability in solving the cocktail party problem. Decoding auditory attention from the brain signals is a major step toward the development of bionic ears emulating human capabilities. Electroencephalography (EEG)-based auditory attention detection (AAD) has attracted considerable interest recently. Despite much progress, the performance of traditional AAD decoders remains to be improved, especially in low-latency settings. State-of-the-art AAD decoders based on deep neural networks generally lack the intrinsic temporal coding ability in biological networks. In this study, we first propose a bio-inspired spiking attentional neural network, denoted as BSAnet, for decoding auditory attention. BSAnet is capable of exploiting the temporal dynamics of EEG signals using biologically plausible neurons and an attentional mechanism. Experiments on two publicly available datasets confirm the superior performance of BSAnet over other state-of-the-art systems across various evaluation conditions. Moreover, BSAnet imitates realistic brain-like information processing, through which we show the advantage of brain-inspired computational models.

Neural encoding of melodic expectations in music across EEG frequency bands

The human brain tracks regularities in the environment and extrapolates these to predict future events. Prior work on music cognition suggests that low-frequency (1-8 Hz) brain activity encodes melodic predictions beyond the stimulus acoustics. Building on this work, we aimed to disentangle the frequency-specific neural dynamics linked to melodic prediction uncertainty (modelled as entropy) and prediction error (modelled as surprisal) for temporal (note onset) and content (note pitch) information. By using multivariate temporal response function (TRF) models, we re-analysed the electroencephalogram (EEG) from 20 subjects (10 musicians) who listened to Western tonal music. Our results show that melodic expectation metrics improve the EEG reconstruction accuracy in all frequency bands below the gamma range (< 30 Hz). Crucially, we found that entropy contributed more strongly to the reconstruction accuracy enhancement compared to surprisal in all frequency bands. Additionally, we found that the encoding of temporal, but not content, information metrics was not limited to low frequencies, rather it extended to higher frequencies (> 8 Hz). An analysis of the TRF weights revealed that the temporal predictability of a note (entropy of note onset) may be encoded in the delta- (1-4 Hz) and beta-band (12-30 Hz) brain activity prior to the stimulus, suggesting that these frequency bands associate with temporal predictions. Strikingly, we also revealed that melodic expectations selectively enhanced EEG reconstruction accuracy in the beta band for musicians, and in the alpha band (8-12 Hz) for non-musicians, suggesting that musical expertise influences the neural dynamics underlying predictive processing in music cognition.