Cortical responses to auditory novelty across task conditions: An intracranial electrophysiology study

Abstract

Elucidating changes in sensory processing across attentional and arousal states is a major focus in neuroscience. The local/global deviant (LGD) stimulus paradigm engages auditory predictive coding over short (local deviance, LD) and long (global deviance, GD) time scales, and has been used to assay disruption of auditory predictive coding upon loss of consciousness. Our previous work (Nourski et al., 2018, J Neurosci 38:8441-52) examined effects of general anesthesia on short- and long-term novelty detection. GD effects were suppressed at subhypnotic doses of propofol, suggesting that they may be more related to task engagement than consciousness per se. The present study addressed this hypothesis by comparing cortical responses to auditory novelty during passive versus active listening conditions in awake listeners. Subjects were seven adult neurosurgical patients undergoing chronic invasive monitoring for medically intractable epilepsy. LGD stimuli were sequences of four identical vowels followed by a fifth identical or different vowel. In the passive condition, the stimuli were presented to subjects as they watched a silent TV program and were instructed to attend to its content. In the active condition, stimuli were presented in the absence of a TV program, and subjects were instructed to press a button in response to GD target stimuli. Intracranial recordings were made from multiple brain regions, including core and non-core auditory, auditory-related, prefrontal and sensorimotor cortex. Metrics of task performance included hit rate, sensitivity index, and reaction times. Cortical activity was measured as averaged auditory evoked potentials (AEPs) and event-related band power in high gamma (70-150 Hz) and alpha (8-14 Hz) frequency bands. The vowel stimuli and LD elicited robust AEPs in all studied brain areas in both passive and active conditions. High gamma responses to stimulus onset and LD were localized predominantly to the auditory cortex in the superior temporal plane and had a comparable prevalence and spatial extent between the two conditions. In contrast, GD effects (AEPs, high gamma and alpha suppression) were greatly enhanced during the active condition in all studied brain areas. The prevalence of high gamma GD effects was positively correlated with individual subjects' task performance. The data demonstrate distinct task engagement-related effects on responses to auditory novelty across the auditory cortical processing hierarchy. The results motivate a closer examination of effective connectivity underlying attentional modulation of cortical sensory responses, and serve as a foundation for examining changes in sensory processing associated with general anesthesia, sleep and disorders of consciousness.

Keywords: Alpha suppression; Auditory evoked potential; Electrocorticography; High gamma; Human auditory cortex.

Figure 1:

LGD experimental paradigm. a: Waveforms of the two vowel sounds /a/ and /i/ used to construct the experimental stimuli. b: Schematic of the four experimental stimuli. c: Stimulus sequences. d: Comparisons between trials to characterize local and global deviance effects. Modified from Strauss et al. (2015).

Figure 2:

GD target detection task performance. Summary of data from seven subjects. a: Hit rates (% correctly detected target stimuli). b: Sensitivity (d’). c: Reaction times (RTs). Violin plots depict RT distributions and median values. Dashed line corresponds to the grand median RT across all subjects and hit trials (0.527 s).

Figure 3.

Responses to vowels and auditory novelty. Exemplary data from subject L400. a: Electrode coverage. Lateral view of the left cerebral hemisphere, top-down view of the left superior temporal plane (STP) and the bottom-up view of the ventral brain surface are shown on top, middle and bottom, respectively. Recording sites are color-coded according to ROI (see Methods). Exemplary sites in core auditory cortex (posteromedial portion of Heschl’s gyrus, HGPM), non-core auditory cortex (anterolateral portion of Heschl’s gyrus, HGAL) and prefrontal cortex (middle frontal gyrus, MFG) are labeled A, B and C respectively. b: Exemplary responses to the experimental stimuli and LGD effects recorded from three cortical sites (A, B and C, as shown in panel a). Averaged evoked potential (AEP) waveforms (active block: red; passive block: blue) are plotted underneath time-frequency event-related band power (ERBP) plots of responses recorded during active and passive block. Columns, from left to right: responses to vowels averaged across all trials (stimulus schematic shown on top); responses to LS trials; responses to LD trials; difference between LS and LD responses (LD effect); responses to GS trials; responses to GD trials; difference between GS and GD responses (GD effect). Distribution of reaction times (RTs) and median RT is shown above GD and GS-GD panels. Note that the X-axis in the leftmost column begins at the onset of the 1st vowel; in all other columns the x-axis begins at the onset of the 5th vowel.

Figure 4.

Vowel responses in passive and active conditions. a: Percentages of responsive sites within each ROI across all 7 subjects in the passive and active condition are plotted on abscissa and ordinate, respectively. None of the differences in percentage reached statistical significance threshold (p = 0.05, Fisher exact test, FDR-corrected). b: Distribution of sites that exhibited significant responses in the passive (blue), active (red) or both conditions (light gray), shown for AEP and high gamma ERBP (left and right panel, respectively). Summary of data from 7 subjects, plotted in MNI coordinate space and projected onto FreeSurfer average template brain. Top-down views of the left superior temporal plane are plotted underneath side views of the right lateral hemispheric convexity, aligned with respect to the y_MNI coordinate. c: Time course of vowel responses. Number of sites within each ROI, exhibiting significant vowel responses are plotted as a function of time after the stimulus onset for AEP and high gamma in left and right panel, respectively. Values above and below the horizontal axis represent site counts for the active and passive condition, respectively.

Figure 5.

LD effects in passive and active conditions. See caption of Figure 4 for details.

Figure 6.

GD effects in passive and active conditions. a: Percentages of sites with significant GD effects within each ROI across all 7 subjects in the passive and active condition are plotted on abscissa and ordinate, respectively. Crosses and filled circles represent data points associated with non-significant and significant differences, respectively (threshold p = 0.05, Fisher exact test, FDR-corrected) between passive and active conditions. Note the different scales for AEP and high gamma. b: Distribution of sites that exhibited significant GD effects in the passive (blue), active (red) or both conditions (light gray), shown for AEP and high gamma ERBP (left and right panel, respectively). Summary of data from 7 subjects, plotted in MNI coordinate space and projected onto FreeSurfer average template brain. Top-down views of the left superior temporal plane are plotted underneath side views of the right lateral hemispheric convexity, aligned with respect to the y_MNI coordinate. c: Time course of GD effects. Number of sites within each ROI, exhibiting significant GD effects are plotted as a function of time after the stimulus onset for AEP and high gamma in left and right panel, respectively. Values above and below the horizontal axis represent site counts for the active and passive condition, respectively. Distribution of reaction times (RTs) and median RT is shown as violin plots above GD time course plots for the active condition.

Figure 7.

Alpha suppression LGD effects in passive and active conditions. a: Percentages of sites within each ROI across all 7 subjects that exhibited significant alpha LD and GD effects (left and right panel, respectively). Percentages of sites in the passive and active conditions are plotted on abscissa and ordinate, respectively. Crosses and filled circles represent data points associated with non-significant and significant differences, respectively (threshold p = 0.05, Fisher exact test, FDR-corrected) between passive and active conditions. b: Distribution of sites that exhibited significant alpha LD and GD effects (left and right panel, respectively) in the passive (blue), active (red) or both conditions (light gray). Summary of data from 7 subjects, plotted in MNI coordinate space and projected onto FreeSurfer average template brain. Top-down views of the left superior temporal plane are plotted underneath side views of the right lateral hemispheric convexity, aligned with respect to the y_MNI coordinate. c: Time course of alpha LD and GD effects. Number of sites within each ROI, exhibiting significant LD and GD alpha suppression effects are plotted as a function of time after the fifth vowel onset in left and right panel, respectively. Values above and below the horizontal axis represent site counts for the active and passive condition, respectively. Distribution of reaction times (RTs) and median RT are shown above the GD time course plot for the active condition.