Abnormal vocal behavior predicts executive and memory deficits in Alzheimer's disease

Abstract

Speakers respond automatically and rapidly to compensate for brief perturbations of pitch in their auditory feedback. The specific adjustments in vocal output require integration of brain regions involved in speech-motor-control in order to detect the sensory-feedback error and implement the motor correction. Cortical regions involved in the pitch reflex phenomenon are highly vulnerable targets of network disruption in Alzheimer's disease (AD). We examined the pitch reflex in AD patients (n = 19) compared to an age-matched control group (n = 16). We measured the degree of behavioral compensation (peak compensation) and the extent of the adaptive response (pitch-response persistence). Healthy-controls reached a peak compensation of 18.7 ± 0.8 cents, and demonstrated a sustained compensation at 8.9 ± 0.69 cents. AD patients, in contrast, demonstrated a significantly elevated peak compensation (22.4 ± 1.2 cents, p < 0.05), and a reduced sustained response (pitch-response persistence, 4.5 ± 0.88 cents, p < 0.001). The degree of increased peak compensation predicted executive dysfunction, while the degree of impaired pitch-response persistence predicted memory dysfunction, in AD patients. The current study demonstrates pitch reflex as a sensitive behavioral index of impaired prefrontal modulation of sensorimotor integration, and compromised plasticity mechanisms of memory, in AD.

Keywords: Alzheimer's disease; Executive dysfunction; Network disruption; Pitch perturbation; Prefrontal modulation; Sensorimotor integration.

Copyright © 2017 Elsevier Inc. All rights reserved.

Figures

Figure 1. Cortical circuits of speech motor control

Anatomical locations of candidate cortical areas are depicted on a schematic brain diagram. The arrows indicate auditory feedback control pathways where feedback predictions (green arrows) are compared with incoming feedback to generate feedback corrections (red arrows), whose key processing nodes (premotor cortex and posterior superior temporal/inferior parietal cortex) are modulated by prefrontal cortex (blue lines).

Figure 2. Apparatus and behavior

(A) Diagram of the pitch-perturbation apparatus. A digital signal processing method was used to shift the pitch of participants’ vocalizations (orange line) and delivered this auditory feedback (purple line) to participants’ earphones. (B) A schematic of the mechanisms involved in pitch-reflex. Auditory feedback of the participant's voice (orange) is briefly perturbed in pitch and heard by the participant (purple). This perturbed auditory feedback is conveyed to auditory areas in the central nervous system, where it is mismatched with the motor-derived predictions. This mismatch gives rise to a feedback prediction error which then modulates the ongoing speech output to compensate for the perturbation. (C) An example of pitch-perturbation response from a single control participant. The purple line denotes the mean audio feedback of the auditory input across all down perturbation trials. Perturbations start at 0ms on the x-axis and last for 400ms. The orange line denotes the vocal output (mean response across all the down perturbation trials) demonstrating the compensatory response entailing an increase in pitch. (D) Histogram of peak compensatory responses as a percentage of pitch shift for the down perturbation trials for the same response shown in C.

Figure 3. AD patients showed an altered behavioral response to auditory feedback perturbation

(A) Vocal response to down (top panel) and up (bottom panel) perturbations of 100 cents, for controls and patients. (B) Single combined vocal response to pitch-perturbations including both up and down perturbations. Combined responses were generated by flipping the deviations from the mean time-course in response to up perturbations and adding these flipped trials to the data from down perturbations. The combined data were generated for each participant separately. Dark lines indicate the mean responses of each group and the shaded areas indicate standard error across the trials per each group. The time axis is time locked to perturbation onset (0ms). Grey shaded area indicates the duration of perturbation. The dotted lines in B indicate the two segments of the behavioral response—pitch-compensatory-response and pitch-response-persistence.

Figure 4. Increased peak-compensation predicted poor executive function in AD patients

Pearson partial correlations between the residuals of peak-compensation after regressing on pitch-response-persistence, age, sex and MMSE, and the composite scores of: (A) Executive; (B) Fluency (generation tasks); (C) Memory; (D) Language. Each composite score was estimated as the average of z scores derived for each of its component tasks. The component tasks included: executive – cognitive control (Stroop) and set-shifting (modified Trail Making); fluency (generation tasks) – lexical fluency (D words), category fluency (animals) and design fluency; memory – short delay verbal recall (CVLT 30 second recall) and delayed verbal recall (CVLT 10 minute recall); language – naming (BNT), repetition, and PPVT. Abbreviations: AD=Alzheimer's disease; BNT=Boston Naming Test; CVLT=California Verbal Learning Test (short form); MMSE=Mini-Mental State examination; PPVT=Peabody Picture Vocabulary Test.

Figure 5. Decreased pitch-response-persistence predicted better memory function in AD patients

Pearson partial correlations between the residuals of pitch-response-persistence after regressing on peak-compensation, age, sex and MMSE, and the composite scores of: (A) Executive; (B) Fluency; (C) Memory; (D) Language. Each composite score was estimated as the average of z scores derived for each of its component tasks (same as for Figure 4). Abbreviations: AD=Alzheimer's disease; MMSE=Mini-Mental State examination.