Scalable Variational Inference for Low-Rank Spatiotemporal Receptive Fields.

An important problem in systems neuroscience is to characterize how a neuron integrates sensory inputs across space and time. The linear receptive field provides a mathematical characterization of this weighting function and is commonly used to quantify neural response properties and classify cell types. However, estimating receptive fields is difficult in settings with limited data and correlated or high-dimensional stimuli. To overcome these difficulties, we propose a hierarchical model designed to flexibly parameterize low-rank receptive fields. The model includes gaussian process priors over spatial and temporal components of the receptive field, encouraging smoothness in space and time. We also propose a new temporal prior, temporal relevance determination, which imposes a variable degree of smoothness as a function of time lag. We derive a scalable algorithm for variational Bayesian inference for both spatial and temporal receptive field components and hyperparameters. The resulting estimator scales to high-dimensional settings in which full-rank maximum likelihood or a posteriori estimates are intractable. We evaluate our approach on neural data from rat retina and primate cortex and show that it substantially outperforms a variety of existing estimators. Our modeling approach will have useful extensions to a variety of other high-dimensional inference problems with smooth or low-rank structure.

Ignoring correlated activity causes a failure of retinal population codes.

From starlight to sunlight, adaptation alters retinal output, changing both the signal and noise among populations of retinal ganglion cells (RGCs). Here we determine how these light level-dependent changes impact decoding of retinal output, testing the importance of accounting for RGC noise correlations to optimally read out retinal activity. We find that at moonlight conditions, correlated noise is greater and assuming independent noise severely diminishes decoding performance. In fact, assuming independence among a local population of RGCs produces worse decoding than using a single RGC, demonstrating a failure of population codes when correlated noise is substantial and ignored. We generalize these results with a simple model to determine what conditions dictate this failure of population processing. This work elucidates the circumstances in which accounting for noise correlations is necessary to take advantage of population-level codes and shows that sensory adaptation can strongly impact decoding requirements on downstream brain areas.

Sustained Melanopsin Photoresponse Is Supported by Specific Roles of ß-Arrestin 1 and 2 in Deactivation and Regeneration of Photopigment.

Melanopsin-expressing intrinsically photosensitive retinal ganglion cells (ipRGCs) are indispensable for non-image-forming visual responses that sustain under prolonged illumination. For sustained signaling of ipRGCs, the melanopsin photopigment must continuously regenerate. The underlying mechanism is unknown. We discovered that a cluster of Ser/Thr sites within the C-terminal region of mammalian melanopsin is phosphorylated after a light pulse. This forms a binding site for ß-arrestin 1 (ßARR1) and ß-arrestin 2. ß-arrestin 2 primarily regulates the deactivation of melanopsin; accordingly, ßαrr2-/- mice exhibit prolonged ipRGC responses after cessation of a light pulse. ß-arrestin 1 primes melanopsin for regeneration. Therefore, ßαrr1-/- ipRGCs become desensitized after repeated or prolonged photostimulation. The lack of either ß-arrestin attenuates ipRGC response under prolonged illumination, suggesting that ß-arrestin 2-mediated deactivation and ß-arrestin 1-dependent regeneration of melanopsin function in sequence. In conclusion, we discovered a molecular mechanism by which ß-arrestins regulate different aspects of melanopsin photoresponses and allow ipRGC-sustained responses under prolonged illumination.

Rapid evaluation of the durability of cortical neural implants using accelerated aging with reactive oxygen species.

OBJECTIVE: A challenge for implementing high bandwidth cortical brain-machine interface devices in patients is the limited functional lifespan of implanted recording electrodes. Development of implant technology currently requires extensive non-clinical testing to demonstrate device performance. However, testing the durability of the implants in vivo is time-consuming and expensive. Validated in vitro methodologies may reduce the need for extensive testing in animal models. APPROACH: Here we describe an in vitro platform for rapid evaluation of implant stability. We designed a reactive accelerated aging (RAA) protocol that employs elevated temperature and reactive oxygen species (ROS) to create a harsh aging environment. Commercially available microelectrode arrays (MEAs) were placed in a solution of hydrogen peroxide at 87 °C for a period of 7 days. We monitored changes to the implants with scanning electron microscopy and broad spectrum electrochemical impedance spectroscopy (1 Hz-1 MHz) and correlated the physical changes with impedance data to identify markers associated with implant failure. MAIN RESULTS: RAA produced a diverse range of effects on the structural integrity and electrochemical properties of electrodes. Temperature and ROS appeared to have different effects on structural elements, with increased temperature causing insulation loss from the electrode microwires, and ROS concentration correlating with tungsten metal dissolution. All array types experienced impedance declines, consistent with published literature showing chronic (>30 days) declines in array impedance in vivo. Impedance change was greatest at frequencies <10 Hz, and smallest at frequencies 1 kHz and above. Though electrode performance is traditionally characterized by impedance at 1 kHz, our results indicate that an impedance change at 1 kHz is not a reliable predictive marker of implant degradation or failure. SIGNIFICANCE: ROS, which are known to be present in vivo, can create structural damage and change electrical properties of MEAs. Broad-spectrum electrical impedance spectroscopy demonstrates increased sensitivity to electrode damage compared with single-frequency measurements. RAA can be a useful tool to simulate worst-case in vivo damage resulting from chronic electrode implantation, simplifying the device development lifecycle.