Recent Research Studies
A. Somatosensory goals in speech
What is the sensory basis of intelligible speech among post-lingually deaf adults? How is that the deaf speakers remain intelligible long after hearing loss? To address these questions we tested cochlear implant recipients (at David Ostry’s lab) who we studied with their implants turned off. Our participants were required to repeat words from a computer screen while a robotic device applied a load to the jaw. As in adults with normal hearing, we found that even without explicit training implant subjects progressively adapted to the mechanical perturbation. The compensations we observed were for movement deviations that were rather small, pointing to the deaf speakers extraordinary sensitivity to make accurate movements. These findings point to non-auditory somatosensory information as a basis for intelligible speech in the deaf (Figure 3).
Sagittal plane jaw movement paths (left panel). For implant subjects, jaw paths were straight in the absence of load (gold: implant turned on; cyan: implant turned off and remained so until the end of the experiment). The jaw was deflected in the protrusion direction when the load came on (red). After training, movement curvature decreased (black). When the load was switched off unexpectedly at the end of training, there was a small after-effect (gray). Scatter plot showing learning for a representative implant subject (right panel).
B. Production-perception link
In another study (at David Ostry’s lab), we tested the idea that motor learning modifies sensory function and, in particular, that speech motor learning affects a speaker’s auditory map. We demonstrated that in the course of force-field learning speakers show systematic changes in their perceptual classification of speech sounds. The data fit well with the idea that speech perception is closely linked to somatosensory function and is modified by speech motor learning (Figure 4).
Example of auditory recalibration following force-field learning. The psychometric function depicting identification probability for had before (cyan) and after (red) force-field learning. A perceptual shift towards head was observed following learning.
C. Neural phase coherence in speech motor learning
The neural mechanisms underlying precise spatiotemporal coordination of distributed brain regions supporting speech remain elusive. The oscillatory dynamics inherent in the brain’s electrical activity provide a suitable means to study the neural communication within brain networks at timescales relevant for speech. We have provided evidence that brain oscillation patterns change as a result of speech motor adaptation (Figure 6). Changes in phase coherence were observed at distinct regions over the scalp that significantly correlated with the amount of adaptation, increasing over the central region while decreasing over the bilateral fronto-temporal areas revealing an interacting speech motor network that supports learning.
Coherence is a predictor of speech motor adaptation. A, Changes in phase coherence using bootstrap analysis at early and late training relative to baseline. Shown here are the three representative electrodes Cz, Fc5 and Fc6. Relative to baseline, significant changes were observed only at late training, and and during speech planning as shown by black rectangles. B, Correlation between the amount of adaptation and phase coherence at late training. While the amount of adaptation was positively correlated at Cz, it was negative at Fc5. No correlation was observed at Fc6.
D. Variability in speech motor learning: Psychophysics and neural Bases
What are the learnability factors of speech? As children learn to talk many factors, such as their ability to process speech sounds, or how well they can repeat what they heard, may contribute to their learning progression. An important component in speech learning seems to be the detection of errors received through sensory feedback, and use that information to rectify mistakes in utterances produced subsequently for maintaining speech goals. We provide evidence that a better learner is more able to correct for the errors received through feedback, and thus shows better control in the use of feedback in updating subsequent production.
Variable outcomes in speech motor learning (left-panel). While both learners and non-learners start at the similar baseline level (blue vs. cyan) learners oppose the downward auditory perturbation by shifting production upward (red) and the non-learners seem to follow the perturbation in their production (magenta). Both groups similarly have different after-effect profiles (black vs. gray). One-lag covariance computed using formant trajectories is a predictor of learning (right-panel). The amount of learning correlates well with one-lag covariance; the worse a non-learner is the higher the covariance is and vice versa.
It can be shown that learning differences are captured and predicted by distinctive EEG spectral profiles.
A. Motor speech learning differences are predicted by EEG frequency bands. Linear classification was used to assess information contained in different EEG frequency bands about predicting outcome differences in the speech motor task. Only the theta, beta and gamma band activities at late training were found to be reliable predictors of motor speech learning. The significant electrode locations are marked in red. B. The power levels at early and late training for two representative electrodes are shown in gamma and beta bands. Note that although the adapters and non-adapters did not differ in their power levels at early training, differences between them emerged by late-training.
E. Link between orosensation and swallow kinematics
The observed link between production and perception is not unique to speech. There is a growing body of evidence that points to a closer association between sensory and motor systems in many goal directed tasks such as human arm movement. It can then be reasonably asked if there is a link between swallow motor function and oral sensation. This will highlight a potential sensory origin of dysphagia and understanding of such relationships will help situate swallowing in the broader parlance of human motor control.
A. A measures of oral sensation of taste. Tendency to identify sour sensation in a two-alternate forced-choice identification test. The ordinate depicts the probability identifying a stimulus as sour. Stimuli were drawn from a sweet-sour continuum with mixtures of sweet and sour tastes at varying proportions. Vertical bar represents ±SE. B. Relationship between oral sensation of taste and pharyngeal swallow events. Greater tendency to identify the sour taste is a significantly correlated with shorter pharyngeal delay time for paste.