Neural responses in sensory cortex decrease with stimulus repetition, known as adaptation or repetition suppression (Li et al., 1993; Grill-Spector et al., 2006). In auditory cortex, adaptation is thought to enhance detection of novel sounds and facilitate listening in noisy environments (Ulanovsky et al., 2003; Taaseh et al., 2011). Adaptation has been studied in animal primary auditory cortex using pure tones (Condon and Weinberger, 1991; Ulanovsky et al., 2003, 2004; Wehr and Zador, 2005; Taaseh et al., 2011). Primary auditory cortex is tonotopic and corresponds to Heschl’s gyrus, the first cortical region to receive auditory input. Scalp EEG studies have confirmed rapid adaptation to tones in human cortical auditory areas (Costa-Faidella et al., 2011; Herrmann et al., 2013; Lanting et al., 2013). Although the high temporal resolution of scalp recordings is well suited for capturing the rapid changes in neural activity (ms), poor spatial resolution has precluded precise localization within auditory cortex.

Intracranial electrocorticography (ECoG) is performed in epilepsy patients who are surgical candidates. ECoG signals are recorded directly from cortex, providing high spatial-temporal resolution and robust sampling of high-gamma (HG) frequencies (>60 Hz) implicated in sensory perception (Bertrand and Tallon-Baudry, 2000). Event-related increases in HG power are largely non-phase locked and thought to reflect local neuronal population activity (Mukamel et al., 2005; Ray et al., 2008; Ray and Maunsell, 2011). ECoG electrodes are often implanted over perisylvian cortex, including auditory association areas on the lateral superior temporal gyrus. Auditory association cortex receives input from primary auditory cortex and is implicated in processing complex sounds as an intermediate stage along the hierarchical auditory processing pathway (Blumstein et al., 1977; Boatman et al., 1995; Rauschecker et al., 1995; Howard et al., 2000; Kaas and Hackett, 2000; Miglioretti and Boatman, 2005). Previous ECoG studies have shown robust HG responses to speech in auditory association cortex (Crone et al., 2001; Sinai et al., 2009; Chang et al., 2010; Steinschneider et al., 2011; Mesgarani et al., 2014). HG responses in auditory association cortex also show repetition-related adaptation (Eliades et al., 2014) and increased local network activity as measured by low-frequency (4–7 Hz; theta) coupling (Malinowska and Boatman-Reich, 2016).

It is not known, however, whether adaptation occurs locally in auditory association cortex or originates from primary auditory cortex. Resolving this issue is important for advancing our understanding of cortical adaptation and to identify potential post-operative deficits that could affect the every-day listening abilities of patients undergoing temporal lobe resections (Drew et al., 2003; Nagle et al., 2013). Determining the neural bases of adaptation also has implications for functional neuroimaging (fMRI) studies that use adaptation paradigms to investigate sub-regional functional selectively (Kar and Krekelberg, 2016).

It has been difficult to resolve this issue because ECoG electrodes are rarely implanted directly in primary auditory cortex (Howard et al., 1996, 2000). We had a rare opportunity to study a patient with electrodes implanted in primary and association auditory cortex for clinical purposes. By recording simultaneously from both regions it was possible to directly compare temporal-spatial patterns of adaptation. We hypothesized that adaptation to simple tones occurs in primary auditory cortex, whereas adaptation to complex speech and non-speech sounds originates in auditory association cortex.

Twelve HG adapting sites were identified: three posterior depth and nine grid sites. The three depth contacts (DP3-5) were contiguous and responsive to all stimuli, with pure tones eliciting the largest responses (Figures 1A–C). Imaging confirmed the locations of depth contacts in Heschl’s gyrus (Figure 1 and Supplementary Figure 1A). Corresponding stereotactic coordinates (MNI, Talairach) mapped to primary auditory cortex (Supplementary Table 1). Post hoc comparisons by tone frequency showed larger responses to 1200 Hz vs. 1000 Hz (p ≤ 0.0059, Wilcoxon rank sum) at two depth contacts (DP3-4); the third contact (DP5) showed no significant differences (p = 0.224). No HG responses were elicited at the other two posterior depth contacts (DP1-2) or at the anterior depths (DA) located in inferior parietal cortex and insula.

High-gamma adapting sites on the grid localized mainly to superior temporal gyrus (N = 7) in auditory association cortex, as confirmed by imaging and stereotactic mapping (Figure 1A and Supplementary Table 1). Two additional sites were identified over the Sylvian fissure, partially overlapping the posterior superior temporal gyrus. The largest responses localized to the lateral posterior superior temporal gyrus. All HG sites responded to speech; fewer to FM tones (N = 7) and pure tones (N = 4). Visual inspection of time-frequency and population HG plots revealed smaller responses to repeated stimuli at both depth and grid sites, consistent with adaptation (Figures 1B,C).

Evoked (ERP) responses to frequent stimulus trials yielded reliable waveforms with robust N100 responses at all depth and grid sites where HG responses were observed in the same post-stimulus period. The largest N100 response at depths was for pure tones (DP4, -53.86 μV); the largest N100 at grid sites was for speech (RH28, -38.07 μV) (Supplementary Figure 2), consistent with the HG findings. To investigate ERP adaptation as a function of repetition, averaging by trial position was performed; however, the waveforms were too noisy to identify reliable responses likely reflecting the small number of trials in each position (<12 trials). Therefore, no further analysis of the ERPs was performed.

Simultaneous ECoG recordings from primary and association auditory cortex revealed repetition effects that were stimulus- and region-specific. For pure tones, we observed similar patterns of HG adaptation and PAC in both cortical regions, suggesting a single origin in primary auditory cortex. Depth contacts in Heschl’s gyrus showed greater adaptation for pure tones and frequency selectivity, consistent with tonotopic organization of human primary auditory cortex as first shown in a seminal depth recording study by Howard et al. (1996).

FM tones and speech sounds elicited HG responses in both primary and association auditory areas, consistent with prior studies (Brugge et al., 2009; Nourski et al., 2013), and showed distinct regional patterns of adaptation. We observed slower adaptation for FM tones in primary auditory cortex compared with auditory association cortex. Adaptation might be slower in primary auditory cortex if each component frequency is processed sequentially along the tonotopic gradient, whereas auditory association cortex also receives input from non-lemniscal thalamic pathways (Kaas and Hackett, 2000) and, therefore, may initiate adaptation earlier. Alternatively, slower adaptation to FM tones in primary auditory cortex could reflect feedback from auditory association cortex (Kumar et al., 2011). Because ECoG recordings are limited to cortical activity, it was not possible to differentiate between these two possibilities.

HG adaptation to speech was observed only in auditory association cortex, suggesting that the lateral temporal lobe plays a role in suppressing repetitive, complex sounds. This has potential clinical implications for patients undergoing right temporal lobe resections as they may have increased listening difficulties in real-world environments where repetitive background sounds (noise) are common (Drew et al., 2003; Nagle et al., 2013).

Significant increases in PAC were observed in auditory association cortex for FM tones and speech, but not pure tones. This finding is consistent with our single-trial results and suggests that tone adaptation occurs mainly in primary auditory cortex, whereas adaptation to complex sounds occurs in auditory association cortex. Repetition-related increases in inter-regional PAC were also observed for speech and FM tones but not for pure tones, suggesting multiple, stimulus-dependent sources of theta modulation in auditory cortex. Substituting alpha frequencies yielded no increases in PAC, indicating that phase-modulation of HG may also be theta specific.

Taken together, these results suggest that adaptation in human auditory cortex during passive listening has multiple stimulus-dependent origins, consistent with hierarchical models of cortical auditory processing (Boatman et al., 1995; Rauschecker et al., 1995; Kaas and Hackett, 2000). Although we did not observe PAC increases at frontal lobe sites, this does not preclude top-down modulation (feedback) of auditory cortex under other auditory conditions, such as attended listening, or using other connectivity methods. Results from a single subject with limited spatial sampling also preclude generalization and require verification with larger studies.

Our results suggest that adaptation in human auditory cortex has multiple regional and inter-regional origins. Neural mechanisms of adaptation appear sensitive to stimulus complexity, consistent with hierarchical cortical auditory models. These results have implications for neuroimaging studies using adaptation paradigms as stimulus factors may determine sources of sub-regional functional selectivity. Our results reveal a novel functional network for adaptation involving the non-dominant temporal lobe. These results underscore the need for accurate localization to avoid potential post-operative deficits, including previously unrecognized difficulty suppressing repetitive background sounds (noise) in everyday listening environments.

The procedures were approved by the Johns Hopkins Institutional Review Board. Written informed consent in accordance with Johns Hopkins Institutional Review Board requirements. The research had no impact on clinical care of the patient.

DB-R designed the study. FL performed the surgery. NC and DB-R collected the data. UM, DB-R, and MC analyzed the data. UM and DB-R wrote the manuscript. UM, NC, FL, and MC revised the manuscript. All authors approved the final version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.