DIVA Meets EEG: Model Validation Using Formant-Shift Reflex

Abstract

The neurocomputational model 'Directions into Velocities of Articulators' (DIVA) was developed to account for various aspects of normal and disordered speech production and acquisition. The neural substrates of DIVA were established through functional magnetic resonance imaging (fMRI), providing physiological validation of the model. This study introduces DIVA_EEG an extension of DIVA that utilizes electroencephalography (EEG) to leverage the high temporal resolution and broad availability of EEG over fMRI. For the development of DIVA_EEG, EEG-like signals were derived from original equations describing the activity of the different DIVA maps. Synthetic EEG associated with the utterance of syllables was generated when both unperturbed and perturbed auditory feedback (first formant perturbations) were simulated. The cortical activation maps derived from synthetic EEG closely resembled those of the original DIVA model. To validate DIVA_EEG, the EEG of individuals with typical voices (N = 30) was acquired during an altered auditory feedback paradigm. The resulting empirical brain activity maps significantly overlapped with those predicted by DIVA_EEG. In conjunction with other recent model extensions, DIVA_EEG lays the foundations for constructing a complete neurocomputational framework to tackle vocal and speech disorders, which can guide model-driven personalized interventions.

Keywords: DIVA model; EEG; auditory feedback; feedback perturbation; vocal compensation.

Figure 1.

DIVA model scheme. vMC, ventral motor cortex; vPMC, ventral premotor cortex; vSC, ventral somatosensory cortex; pAC, posterior auditory cortex.

Figure 2.

Block diagram illustrating the methodology proposed for the construction of DIVA_EEG. Both the DIVA model Simulation and the Experimental Phase of the study are presented.

Figure 3.

Schematic of the apparatus for applying formant perturbations. Participants produced monosyllabic words containing the vowel /e/ while their auditory feedback was perturbed toward the participant-specific vowel /a/ (e.g., participants produced /mes/ but heard a word that sounded like /mas/).

Figure 4.

Simulations of the brain cortical activity associated with the different DIVA maps during the vocalization of the phoneme /e/ with undisturbed feedback: (A) Time course of activity of DIVA cortical maps. t: target, s: state, e: error (B) Topographic representations of cortical activity for time t = 0, 10, 25, 250, 510, 550 ms relative to the onset of the vocalization. top panel: cortical seeds. middle panel: simulated EEG. bottom panel: source space representation of the synthetic EEG.

Figure 5.

Simulations of the brain cortical activity associated with the different DIVA maps elicited by auditory feedback perturbations (F1 shifts) during the vocalization of the phoneme /e/. (A) Time course of activity of the DIVA cortical maps whose activity varied in response to feedback perturbations. Activities in undisturbed, downshifted, and upshifted conditions are presented. The shaded area represents the N1-P2 interval of the ERP. t: target, s: state, e: error (B) Scalp topography and source space representation of the synthetic EEG estimated in the time interval that corresponds to the generation of the N1-P2 complex. (C) Synthetic EEG (N1-P2 interval) contrasted across conditions.

Figure 6.

Acoustic and electrophysiological parameters describing the monitoring of one’s own vocalization. (A) Examples of vocal compensations elicited by F1 perturbations in the auditory feedback. In the left panel, an oscillogram representative of the phoneme /mes/ is illustrated. Likewise, the direction of the perturbation is indicated at the top of each chart. The mean F1 values of vocalizations produced in unperturbed acoustic conditions and those of vocal compensations to perturbed auditory feedback are presented in the right panel, along with the corresponding sample distributions. (B) Event-related potential (ERP) elicited by actively monitoring the auditory feedback of one’s own vocalizations. In the left panel, the grand average of the ERP elicited by both unperturbed and F1-shifted auditory feedback are presented. The shaded area indicates the N1-P2 complex. Scalp topography of the N1-P2 complex is illustrated in the middle panel. The mean amplitude of the N1-P2 complex elicited by unperturbed and perturbed auditory feedback are presented in the right panel, along with the corresponding sample distribution. (C) Current density maps illustrating the brain generators of the N1-P2 complex in the different conditions (unperturbed and perturbated auditory feedback). (D) Differences in the cortical activity obtained in response to unperturbed and perturbated auditory feedbacks. The difference between the current density maps elicited by F1 perturbations of equal magnitude and opposite directions is presented in the right panel. (E) Cortical sources of the N1-P2 complex elicited in response to F1 perturbations in the auditory feedback of one’s own vocalizations that are predicted by the DIVA model. They are illustrated both areas and voxels for which the activity predicted by the model overlapped that estimated from the real EEG. Statistically significant differences between groups are represented by *.