Combining ultra-high-field fMRI adaptation and computational modelling to estimate neuronal frequency selectivity in human auditory cortex

Frequency selectivity is a ubiquitous property of auditory neurons. Measuring it in human auditory cortex may be crucial for understanding common auditory deficits, but current non-invasive neuroimaging techniques can only measure the aggregate response of large populations of cells, thereby overestimating tuning width. Here we attempted to estimate neuronal frequency tuning in human auditory cortex using a combination of fMRI-adaptation paradigm at 7T and computational modelling. We measured the BOLD response in the auditory cortex of eleven participants to a high frequency (3.8 kHz) probe presented alone or preceded by adaptors at different frequencies (0.5 to 3.8 kHz). From these data, we derived both the response tuning curves (the BOLD response to adaptors alone as a function of adaptor frequency) and adaptation tuning curves (the degree of response suppression to the probe as a function of adaptor frequency, assumed to reflect neuronal tuning) in primary and secondary auditory cortical areas, delineated in each participant. Results suggested the existence of both frequency-independent and frequency-specific adaptation components, with the latter being more frequency-tuned than response tuning curves. Using a computational model of neuronal adaptation and BOLD non-linearity in topographically-organized cortex, we demonstrate both that the frequency-specific adaptation component overestimates the underlying neuronal frequency tuning and that frequency-specific and frequency-independent adaptation component cannot easily be disentangled from the adaptation tuning curve. By fitting our model directly to the response and adaptation tuning curves, we derive a range of plausible values for neuronal frequency tuning. Our results suggest that fMRI adaptation is suitable for measuring neuronal frequency tuning properties in human auditory cortex, provided population effects and the non-linearity of BOLD response are taken into account.

Increasing neural network robustness improves match to macaque V1 eigenspectrum, spatial frequency preference and predictivity

Task-optimized convolutional neural networks (CNNs) show striking similarities to the ventral visual stream. However, human-imperceptible image perturbations can cause a CNN to make incorrect predictions. Here we provide insight into this brittleness by investigating the representations of models that are either robust or not robust to image perturbations. Theory suggests that the robustness of a system to these perturbations could be related to the power law exponent of the eigenspectrum of its set of neural responses, where power law exponents closer to and larger than one would indicate a system that is less susceptible to input perturbations. We show that neural responses in mouse and macaque primary visual cortex (V1) obey the predictions of this theory, where their eigenspectra have power law exponents of at least one. We also find that the eigenspectra of model representations decay slowly relative to those observed in neurophysiology and that robust models have eigenspectra that decay slightly faster and have higher power law exponents than those of non-robust models. The slow decay of the eigenspectra suggests that substantial variance in the model responses is related to the encoding of fine stimulus features. We therefore investigated the spatial frequency tuning of artificial neurons and found that a large proportion of them preferred high spatial frequencies and that robust models had preferred spatial frequency distributions more aligned with the measured spatial frequency distribution of macaque V1 cells. Furthermore, robust models were quantitatively better models of V1 than non-robust models. Our results are consistent with other findings that there is a misalignment between human and machine perception. They also suggest that it may be useful to penalize slow-decaying eigenspectra or to bias models to extract features of lower spatial frequencies during task-optimization in order to improve robustness and V1 neural response predictivity.

Optical cavity-referenced terahertz synthesizer with 15-digit short-term stability

Stable terahertz sources are required to advance high precision terahertz applications such as molecular spectroscopy, terahertz radars, and wireless communications. Here, we demonstrate a photonic scheme of terahertz synthesis using an optical comb in stabilization to an ultra-low expansion optical cavity offering a 15-digit accuracy. By heterodyne photomixing of comb lines, terahertz frequencies of 0.10 – 1.10 THz are synthesized with a 2-mHz linewidth and a fractional instability of 3.26×10-15 at 1.3-s integration. Compared to other state-of-the-art counterparts, our terahertz synthesizer offers a fine frequency tuning capability in steps of 100 MHz and an extremely low level of phase noise below -70 dBc/Hz even at 1 Hz offset. Such unprecedented performance is expected to drastically improve the signal-to-noise ratio of terahertz radars, the resolving power of terahertz molecular spectroscopy, and the transmission capacity of wireless communications.

Analysis of frequency response sensor of MEMS gyroscope in vacuum chamber

This article presents a study of the frequency response of a MEMS gyroscope in a vacuum chamber. On the basis of experimental studies by the method of laser Doppler vibrometry, the dependences of the amplitude of oscillations of the inertial mass in the vertical plane at various pressures are obtained. The bandwidth of the MEMS sensor was also measured.As a result of the experiments, the damping factors were determined to compose a more complete mathematical model and for more accurate finite element modeling in ANSYS, and refined parameters of the electrostatic drive and the amplitude of oscillations along the axis of the drive were obtained. These studies will be useful for determining the residual degree of vacuum in the case for further frequency tuning of the device.

Mammalian octopus cells are direction selective to frequency sweeps by synaptic sequence detection

Octopus cells are remarkable projection neurons of the mammalian cochlear nucleus, with extremely fast membranes and wide frequency tuning. They are considered prime examples of coincidence detectors but are poorly characterized in vivo. We discover that octopus cells are selective to frequency sweep direction, a feature that is absent in their auditory nerve inputs. In vivo intracellular recordings reveal that direction selectivity does not derive from cross-channel coincidence detection but hinges on the amplitudes and activation sequence of auditory nerve inputs tuned to clusters of hotspot frequencies. A simple biophysical model of octopus cells excited with real nerve spike trains recreates direction selectivity through interaction of intrinsic membrane conductances with activation sequence of clustered inputs. We conclude that octopus cells are sequence detectors, sensitive to temporal patterns across cochlear frequency channels. The detection of sequences rather than coincidences is a much simpler but powerful operation to extract temporal information.

On-Chip Voltage-Controlled Oscillator Based on a Center-Tapped Switched Inductor Using GaN-on-SiC HEMT Technology

This study presents a voltage-controlled oscillator (VCO) in a cross-coupled pair configuration using a multi-tapped switched inductor with two switch-loaded transformers in 0.5 µm GaN technology. Two switch-loaded transformers are placed at the inner and outer portions of the multi-tapped inductor. All the switches are turned off to obtain the lowest sub-band. The outer transformer with three pairs of switches is turned on alternately to provide three sub-band modes. A pair of switches at the inner transformer provide a high-frequency band. Two switch-loaded transformers are turned on to provide the highest sub-band. Six modes are selected to provide a wide tuning range. The frequency tuning range (FTR) of the VCO is 27.8% from 3.81 GHz to 8.04 GHz with a varactor voltage from 13 V to 22 V. At a 1 MHz frequency offset from the carrier frequency of 4.27 GHz, the peak phase noise is −119.17 dBc/Hz. At a power supply of 12 V, the output power of the carrier at 4.27 GHz is 20.9 dBm. The figure of merit is −186.93 dB because the VCO exhibits a high output power, low phase noise, and wide FTR. To the best of the author’s knowledge, the FTR in VCOs made of GaN-based high electron mobility transistors is the widest reported thus far.