A hierarchy of processing complexity and timescales for natural sounds in human auditory cortex.

Efficient behavior is supported by humans' ability to rapidly recognize acoustically distinct sounds as members of a common category. Within auditory cortex, there are critical unanswered questions regarding the organization and dynamics of sound categorization. Here, we performed intracerebral recordings in the context of epilepsy surgery as 20 patient-participants listened to natural sounds. We built encoding models to predict neural responses using features of these sounds extracted from different layers within a sound-categorization deep neural network (DNN). This approach yielded highly accurate models of neural responses throughout auditory cortex. The complexity of a cortical site's representation (measured by the depth of the DNN layer that produced the best model) was closely related to its anatomical location, with shallow, middle, and deep layers of the DNN associated with core (primary auditory cortex), lateral belt, and parabelt regions, respectively. Smoothly varying gradients of representational complexity also existed within these regions, with complexity increasing along a posteromedial-to-anterolateral direction in core and lateral belt, and along posterior-to-anterior and dorsal-to-ventral dimensions in parabelt. When we estimated the time window over which each recording site integrates information, we found shorter integration windows in core relative to lateral belt and parabelt. Lastly, we found a relationship between the length of the integration window and the complexity of information processing within core (but not lateral belt or parabelt). These findings suggest hierarchies of timescales and processing complexity, and their interrelationship, represent a functional organizational principle of the auditory stream that underlies our perception of complex, abstract auditory information.

Intracranial Mapping of Response Latencies and Task Effects for Spoken Syllable Processing in the Human Brain.

Prior lesion, noninvasive-imaging, and intracranial-electroencephalography (iEEG) studies have documented hierarchical, parallel, and distributed characteristics of human speech processing. Yet, there have not been direct, intracranial observations of the latency with which regions outside the temporal lobe respond to speech, or how these responses are impacted by task demands. We leveraged human intracranial recordings via stereo-EEG to measure responses from diverse forebrain sites during (i) passive listening to /bi/ and /pi/ syllables, and (ii) active listening requiring /bi/-versus-/pi/ categorization. We find that neural response latency increases from a few tens of ms in Heschl's gyrus (HG) to several tens of ms in superior temporal gyrus (STG), superior temporal sulcus (STS), and early parietal areas, and hundreds of ms in later parietal areas, insula, frontal cortex, hippocampus, and amygdala. These data also suggest parallel flow of speech information dorsally and ventrally, from HG to parietal areas and from HG to STG and STS, respectively. Latency data also reveal areas in parietal cortex, frontal cortex, hippocampus, and amygdala that are not responsive to the stimuli during passive listening but are responsive during categorization. Furthermore, multiple regions-spanning auditory, parietal, frontal, and insular cortices, and hippocampus and amygdala-show greater neural response amplitudes during active versus passive listening (a task-related effect). Overall, these results are consistent with hierarchical processing of speech at a macro level and parallel streams of information flow in temporal and parietal regions. These data also reveal regions where the speech code is stimulus-faithful and those that encode task-relevant representations.

Electrocorticography reveals the dynamics of famous voice responses in human fusiform gyrus.

Voice and face processing occur through convergent neural systems that facilitate speaker recognition. Neuroimaging studies suggest that familiar voice processing engages early visual cortex, including the bilateral fusiform gyrus (FG) on the basal temporal lobe. However, what role the FG plays in voice processing and whether it is driven by bottom-up or top-down mechanisms is unresolved. In this study we directly examined neural responses to famous voices and faces in human FG with direct cortical surface recordings (electrocorticography) in epilepsy surgery patients. We tested the hypothesis that neural populations in human FG respond to famous voices and investigated the temporal properties of voice responses in FG. Recordings were acquired from five adult participants during a person identification task using visual and auditory stimuli from famous speakers (U.S. Presidents Barack Obama, George W. Bush, and Bill Clinton). Patients were presented with images of presidents or clips of their voices and asked to identify the portrait/speaker. Our results demonstrate that a subset of face-responsive sites in and near FG also exhibit voice responses that are both lower in magnitude and delayed (300-600 ms) compared with visual responses. The dynamics of voice processing revealed by direct cortical recordings suggests a top-down feedback-mediated response to famous voices in FG that may facilitate speaker identification.NEW & NOTEWORTHY Interactions between auditory and visual cortices play an important role in person identification, but the dynamics of these interactions remain poorly understood. We performed direct brain recordings of fusiform face cortex in human epilepsy patients performing a famous voice naming task, revealing the dynamics of famous voice processing in human fusiform face cortex. The findings support a model of top-down interactions from auditory to visual cortex to facilitate famous voice recognition.

Neural responses in human superior temporal cortex support coding of voice representations.

The ability to recognize abstract features of voice during auditory perception is an intricate feat of human audition. For the listener, this occurs in near-automatic fashion to seamlessly extract complex cues from a highly variable auditory signal. Voice perception depends on specialized regions of auditory cortex, including superior temporal gyrus (STG) and superior temporal sulcus (STS). However, the nature of voice encoding at the cortical level remains poorly understood. We leverage intracerebral recordings across human auditory cortex during presentation of voice and nonvoice acoustic stimuli to examine voice encoding at the cortical level in 8 patient-participants undergoing epilepsy surgery evaluation. We show that voice selectivity increases along the auditory hierarchy from supratemporal plane (STP) to the STG and STS. Results show accurate decoding of vocalizations from human auditory cortical activity even in the complete absence of linguistic content. These findings show an early, less-selective temporal window of neural activity in the STG and STS followed by a sustained, strongly voice-selective window. Encoding models demonstrate divergence in the encoding of acoustic features along the auditory hierarchy, wherein STG/STS responses are best explained by voice category and acoustics, as opposed to acoustic features of voice stimuli alone. This is in contrast to neural activity recorded from STP, in which responses were accounted for by acoustic features. These findings support a model of voice perception that engages categorical encoding mechanisms within STG and STS to facilitate feature extraction.

Frequency-Following Responses to Speech Sounds Are Highly Conserved across Species and Contain Cortical Contributions.

Time-varying pitch is a vital cue for human speech perception. Neural processing of time-varying pitch has been extensively assayed using scalp-recorded frequency-following responses (FFRs), an electrophysiological signal thought to reflect integrated phase-locked neural ensemble activity from subcortical auditory areas. Emerging evidence increasingly points to a putative contribution of auditory cortical ensembles to the scalp-recorded FFRs. However, the properties of cortical FFRs and precise characterization of laminar sources are still unclear. Here we used direct human intracortical recordings as well as extracranial and intracranial recordings from macaques and guinea pigs to characterize the properties of cortical sources of FFRs to time-varying pitch patterns. We found robust FFRs in the auditory cortex across all species. We leveraged representational similarity analysis as a translational bridge to characterize similarities between the human and animal models. Laminar recordings in animal models showed FFRs emerging primarily from the thalamorecipient layers of the auditory cortex. FFRs arising from these cortical sources significantly contributed to the scalp-recorded FFRs via volume conduction. Our research paves the way for a wide array of studies to investigate the role of cortical FFRs in auditory perception and plasticity.

Feature Selection Methods for Zero-Shot Learning of Neural Activity.

Dimensionality poses a serious challenge when making predictions from human neuroimaging data. Across imaging modalities, large pools of potential neural features (e.g., responses from particular voxels, electrodes, and temporal windows) have to be related to typically limited sets of stimuli and samples. In recent years, zero-shot prediction models have been introduced for mapping between neural signals and semantic attributes, which allows for classification of stimulus classes not explicitly included in the training set. While choices about feature selection can have a substantial impact when closed-set accuracy, open-set robustness, and runtime are competing design objectives, no systematic study of feature selection for these models has been reported. Instead, a relatively straightforward feature stability approach has been adopted and successfully applied across models and imaging modalities. To characterize the tradeoffs in feature selection for zero-shot learning, we compared correlation-based stability to several other feature selection techniques on comparable data sets from two distinct imaging modalities: functional Magnetic Resonance Imaging and Electrocorticography. While most of the feature selection methods resulted in similar zero-shot prediction accuracies and spatial/spectral patterns of selected features, there was one exception; A novel feature/attribute correlation approach was able to achieve those accuracies with far fewer features, suggesting the potential for simpler prediction models that yield high zero-shot classification accuracy.

Pre and post hoc analysis of electronic health record implementation on emergency department metrics.

Longitudinal time-based emergency department (ED) performance measures were quantified 12 months before and 12 months after (March 2012-February 2014) implementation of a Meditech 6.0® electronic health record (EHR) at a single urban academic ED. Data assessed were length of stay from door to door, door to admission, door to bed, bed to provider, provider to disposition, and disposition to admission, as well as number of patients leaving against medical advice and number of patients leaving without being seen. Analysis of variance was used to compare levels before and after EHR implementation for each variable, with adjustments made for the number of admissions, transfers, and month. No difference was seen in monthly volume, admissions, or transfers. Implementation of an EHR resulted in a sustained increase in ED time metrics for mean length of stay and times from door to door, door to admission, door to bed, and provider to disposition. Decreased ED time metrics were seen in bed-to-provider and disposition-to-admit times. The number of patients who left against medical advice increased after implementation, but the number of patients who left without being seen was not significantly different. Thus, EHR implementation was associated with an increase in time with most performance metrics. Although general times trended back to near preimplementation baselines, most ED time metrics remained elevated beyond the study length of 12 months. Understanding the impact of EHR system implementation on the overall performance of an ED can help departments prepare for potential adverse effects of such systems on overall efficiency.

Semantic attributes are encoded in human electrocorticographic signals during visual object recognition.

Non-invasive neuroimaging studies have shown that semantic category and attribute information are encoded in neural population activity. Electrocorticography (ECoG) offers several advantages over non-invasive approaches, but the degree to which semantic attribute information is encoded in ECoG responses is not known. We recorded ECoG while patients named objects from 12 semantic categories and then trained high-dimensional encoding models to map semantic attributes to spectral-temporal features of the task-related neural responses. Using these semantic attribute encoding models, untrained objects were decoded with accuracies comparable to whole-brain functional Magnetic Resonance Imaging (fMRI), and we observed that high-gamma activity (70-110Hz) at basal occipitotemporal electrodes was associated with specific semantic dimensions (manmade-animate, canonically large-small, and places-tools). Individual patient results were in close agreement with reports from other imaging modalities on the time course and functional organization of semantic processing along the ventral visual pathway during object recognition. The semantic attribute encoding model approach is critical for decoding objects absent from a training set, as well as for studying complex semantic encodings without artificially restricting stimuli to a small number of semantic categories.

Local field potentials mitigate decline in motor decoding performance caused by loss of spiking units.

The technology underlying brain computer interfaces has recently undergone rapid development, though a variety of issues remain that are currently preventing it from becoming a viable clinical assistive tool. Though decoding of motor output has been shown to be particularly effective when using spikes, these decoders tend to degrade with the loss of subsets of these signals. One potential solution to this problem is to include features derived from LFP signals in the decoder to mitigate these negative effects. We explored this solution and found that the decline in decoding performance that accompanies spiking unit dropout was significantly reduced when LFP power features were included in the decoder. Additionally, high frequency LFP features in the 100-170 Hz band were more effective than low frequency LFP features in the 2-4 Hz band at protecting the decoder from a dropoff in performance. LFP power appears to be an effective signal to improve the robustness of spiking unit decoders. Future studies will explore online classification and performance improvements in chronic implants by the proposed method.