Search

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.

Neuroscientists are actively pursuing high-precision maps, or graphs consisting of networks of neurons and connecting synapses in mammalian and non-mammalian brains. Such graphs, when coupled with physiological and behavioral data, are likely to facilitate greater understanding of how circuits in these networks give rise to complex information processing capabilities. Given that the automated or semi-automated methods required to achieve the acquisition of these graphs are still evolving, we developed a metric for measuring the performance of such methods by comparing their output with those generated by human annotators (“ground truth” data). Whereas classic metrics for comparing annotated neural tissue reconstructions generally do so at the voxel level, the metric proposed here measures the “integrity” of neurons based on the degree to which a collection of synaptic terminals belonging to a single neuron of the reconstruction can be matched to those of a single neuron in the ground truth data. The metric is largely insensitive to small errors in segmentation and more directly measures accuracy of the generated brain graph. It is our hope that use of the metric will facilitate the broader community's efforts to improve upon existing methods for acquiring brain graphs. Herein we describe the metric in detail, provide demonstrative examples of the intuitive scores it generates, and apply it to a synthesized neural network with simulated reconstruction errors. Demonstration code is available.

Dimensionality is a serious challenge in human neuroimaging. 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. To deal with limited stimulus sets, zero-shot encoding and decoding models have been introduced to classify classes outside the training set. However, these models have been found to be particularly susceptible to the curse of dimensionality and the risk of over-fitting. While no systematic study of feature selection for these models has been reported, a relatively straightforward feature stability approach has been successfully applied across models and imaging modalities. To characterize the tradeoffs in feature selection for encoding and decoding models, we compared correlationbased stability to several other feature selection techniques on comparable data sets from two distinct imaging modalities: functional Magnetic Resonance Imaging and Electrocorticography. Semantic attribute encoding and decoding models were trained and tested with five distinct feature selection methods: correlation-based stability, a wrapper method, Fisher’s Method, Mutual Information-based stability, and a novel attribute/feature correlation technique. Results demonstrate that correlation-based stability performs well across imaging modalities, but other methods result in qualitatively different patterns of selected features while achieving similar levels of performance, and performance can be maintained with fewer features. Choices about feature selection can have a substantial impact when closed-set accuracy, open-set robustness, and runtime are competing design objectives.

Auditory neuropathy spectrum disorder (ANSD) describes a prevalent hearing disorder that is characterized by normal cochlear outer hair cell (OHC) function and abnormal auditory nerve function (1). Minimum diagnostic criteria include normal otoacoustic emissions (OAEs) and missing or aberrant auditory brainstem responses (ABRs). Changes in the ABR are assumed to arise from a dys-synchronization of sound-driven activity in the auditory nerve. The loss of temporal processing can impair speech, sound localization, and listening in noise. The impact of ANSD on speech has encouraged neonatal screening programs for early diagnosis of the disorder. These programs have revealed a higher incidence of ANSD among infants who have been treated in neonatal intensive care units (NICUs)(2,3). Although the health concerns that necessitate early hospitalization are varied, a constant factor in NICU treatments is the noise exposure that is generated by equipment, procedures, and staff (4,5). Several surveys have reported NICU noise levels consistently above the recommendations of the American Academy of Pediatrics (AAP) for both sustained (Leq) and maximum transient exposures (Lmax) (6). At present, significant ambiguities remain on what constitutes an acceptable level of noise exposure in human infants. While it is clear that a single loud exposure at a young age can accelerate hearing loss and promote spiral ganglion degeneration (7), the effects of persistent moderate exposures are difficult to estimate because environmental noise exerts pervasive effects on the developing auditory system. The deficits often manifest as perceptual disorders that are independent of overt hearing loss (8,9). Physiological studies of the cochlea and auditory nerve have provided detailed functional analyses of the medial olivocochlear system (MOCS). In normal individuals, the release of acetylcholine by the MOCS hyperpolarizes the membrane potentials of OHCs (10). The resulting decrease in cochlear mechanical sensitivity protects the ear from damaging levels of noise (11) and enhances auditory processing in distracting levels of noise (12,13). Recent clinical evidence suggests that ANSD is more prevalent among MOCS-compromised children (14). One explanation for this increased risk is that synaptic development of the central auditory pathways requires regulated sound-driven activity in the ascending inputs of the auditory nerve (15–17). When young individuals are exposed to environmental noise, MOC feedback may preserve normal synaptic development by regulating potentially disruptive levels of noise-driven activity. To investigate this novel developmental role for the MOCS, the feedback system was silenced in laboratory mice by a genetic modification of cholinergic signaling (18). The mice were reared in background noise, and then screened for ANSD using established diagnostic criteria. The audiological implications of the observed processing deficits were further delineated with a behavioral assay of temporal acuity. The refinement of this animal model allowed us to relate common clinical observations to direct electrophysiological recordings from the post-synaptic targets of the auditory nerve in the ventral cochlear nucleus. These recordings revealed significant functional impairments of the synaptic specializations that dictate the quality of temporal coding in the auditory brainstem.

This report introduces a system for the objective physiological classification of single-unit activity in the anteroventral cochlear nucleus (AVCN) of anesthetized CBA/129 and CBA/CaJ mice. As in previous studies, the decision criteria are based on the temporal properties of responses to short tone bursts that are visualized in the form of peri-stimulus time histograms (PSTHs). Individual unit types are defined by the statistical distribution of quantifiable metrics that relate to the onset latency, regularity, and adaptation of sound-driven discharge rates. Variations of these properties reflect the unique synaptic organizations and intrinsic membrane properties that dictate the selective tuning of sound coding in the AVCN. When these metrics are applied to the mouse AVCN, responses to best frequency (BF) tones reproduce the major PSTH patterns that have been previously demonstrated in other mammalian species. The consistency of response types in two genetically diverse strains of laboratory mice suggests that the present classification system is appropriate for additional strains with normal peripheral function. The general agreement of present findings to established classifications validates laboratory mice as an adequate model for general principles of mammalian sound coding. Nevertheless, important differences are noted for the reliability of specialized endbulb transmission within the AVCN, suggesting less secure temporal coding in this high-frequency species.

Previous investigations have shown that a subset of inferior colliculus neurons, which have been designated type O units, respond selectively to isolated features of the cat's head-related transfer functions (HRTFs: the directional transformation of a free-field sound as it propagates from the head to the eardrum). Based on those results, it was hypothesized that type O units would show enhanced spatial tuning in a virtual sound field that conveyed the full complement of HRTF-based localization cues. As anticipated, a number of neurons produced representations of virtual sound source locations that were spatially tuned, level tolerant, and effective under monaural conditions. Preferred locations were associated with spectral cues that complemented the highly individualized broadband inhibitory patterns of tuned neurons. That is, higher response magnitudes were achieved when spectral peaks coincided with excitatory influences at best frequency (BF: the most sensitive frequency) and spectral notches fell within flanking inhibitory regions. The directionally dependent modulation of narrowband ON-BF excitation by broadband OFF-BF inhibition was not a unique property of type O units.

Five narrowband frequency pairs in the 220-222 MHz land mobile band have been allocated for applications in intelligent transportation systems (ITS). Low rate data communications in the range of 4 to 30 kbps can be supported on these 4 kHz channels. This paper is the result of an ongoing project to design, construct, and field test a high performance wireless modem architecture for application in the ITS spectrum allocation. Here we concentrate on an architecture optimized for stationary point-to-point applications. Field testing of this modem architecture has demonstrated that bandwidth efficiencies in excess of 5 bps/Hz are achievable at bit error probabilities of around 10/sup -5/. The experimental system which has been built serves as an effective testbed for further advances in narrowband wireless communications and for ITS application development.

Five narrowband frequency pairs in the 220-222 MHz land mobile band have been allocated for applications in intelligent transportation systems (ITS). Low rate data communications in the range of 4 to 30 kbps can be supported on these 4 kHz channels. This paper describes an ongoing project to design, construct, and field test a high performance wireless modem architecture for application in the ITS spectrum allocation. In an effort to achieve high bandwidth efficiency, the modem architecture employs recent advances in transmitter diversity, pilot symbol assisted modulation, frame and frequency synchronization, and forward error control coding. Field testing of this modem architecture has demonstrated that bandwidth efficiencies in excess of 3 bps/Hz are achievable at bit error probabilities of around 10/sup -5/. The experimental system which has been built serves as an effective testbed for further advances in narrowband wireless communications and for ITS application development.

HMAX is a well-known computational model of visual recognition in cortex consisting of just two computational operations – a “template match” and non-linear pooling – alternating in a feedforward hierarchy in which receptive fields exhibit increasing specificity and invariance [1]. Interestingly, auditory recognition problems (such as speech recognition) share similar computational requirements, and recent work in auditory neuroscience suggests that auditory and visual cortex share similar anatomical and functional organization. Based on these similarities, we tested whether HMAX could support an auditory recognition task (specifically, word spotting). To test HMAX on word spotting, recorded speech samples from the TIMIT corpus [2] were first converted into time-frequency spectrograms using a computational model of the auditory periphery [3]. These spectrograms were then split into 750 ms frames and input to a standard HMAX model [4]. Based on observed similarities between the receptive fields in primary auditory cortex (spectro-temporal receptive fields, or STRFs) and primary visual cortex (typically modeled as oriented Gabor filters), we used S1 filters identical to those used in vision [4]. Similarly, S2 “patches” were randomly selected from C1 representations of speech sounds drawn from an independent speech corpus. One vs. all linear support vector machines (SVMs) were then trained to discriminate frames that contain a target word from those that did not. These SVMs were then tested on a novel set of test sentences using a sliding frame approach (750 ms frame size, 20 ms step size). For each frame in a sentence, the SVM produced a distance from the hyperplane, and a threshold value was applied to produce a binary classification whether or not the target word was present in the sentence. When tested on target words that appeared in a fixed context (i.e. SA sentences in TIMIT), performance was highly robust, with ROC areas consistently above 0.9. When tested on target words that appeared in variable contexts (i.e., SI sentences in TIMIT), performance was somewhat decreased with ROC areas around 0.8. This decrease in performance is likely due to the inclusion of “clutter” (i.e., target irrelevant features) within the frame, also commonly observed when HMAX is applied to visual object recognition tasks [1]. These results are novel in that they provide support for the hypothesis that the simple computational framework implemented in HMAX – consisting of a feedforward hierarchy of only two alternating computational operations – may generalize beyond vision to support auditory recognition as well. It is possible that such a representation could give rise to stable neural encodings that are invariant to behaviorally irrelevant characteristics as seen in higher order visual and auditory cortices [5,6]. While it is likely that this auditory version of the HMAX model would benefit from the use of more auditory-specific filters based on STRF models [7], the Gabor features used here are largely compatible with previous computational models based on STRFs up to the level of primary auditory cortex [8]. Additional benefit may also be gained by learning sparse representations from natural sounds, at both the S1 and S2 levels [9].