Implications of the Dependence of Neuronal Activity on Neural Network States for the Design of Brain-Machine Interfaces

Calibrating Bayesian Decoders of Neural Spiking Activity

Accurately decoding external variables from observations of neural activity is a major challenge in systems neuroscience. Bayesian decoders, which provide probabilistic estimates, are some of the most widely used. Here we show how, in many common settings, the probabilistic predictions made by traditional Bayesian decoders are overconfident. That is, the estimates for the decoded stimulus or movement variables are more certain than they should be. We then show how Bayesian decoding with latent variables, taking account of low-dimensional shared variability in the observations, can improve calibration, although additional correction for overconfidence is still needed. Using data from males, we examine (1) decoding the direction of grating stimuli from spike recordings in the primary visual cortex in monkeys, (2) decoding movement direction from recordings in the primary motor cortex in monkeys, (3) decoding natural images from multiregion recordings in mice, and (4) decoding position from hippocampal recordings in rats. For each setting, we characterize the overconfidence, and we describe a possible method to correct miscalibration post hoc. Properly calibrated Bayesian decoders may alter theoretical results on probabilistic population coding and lead to brain-machine interfaces that more accurately reflect confidence levels when identifying external variables.

Classification of Whisker Deflections From Evoked Responses in the Somatosensory Barrel Cortex With Spiking Neural Networks

Spike-based neuromorphic hardware has great potential for low-energy brain-machine interfaces, leading to a novel paradigm for neuroprosthetics where spiking neurons in silicon read out and control activity of brain circuits. Neuromorphic processors can receive rich information about brain activity from both spikes and local field potentials (LFPs) recorded by implanted neural probes. However, it was unclear whether spiking neural networks (SNNs) implemented on such devices can effectively process that information. Here, we demonstrate that SNNs can be trained to classify whisker deflections of different amplitudes from evoked responses in a single barrel of the rat somatosensory cortex. We show that the classification performance is comparable or even superior to state-of-the-art machine learning approaches. We find that SNNs are rather insensitive to recorded signal type: both multi-unit spiking activity and LFPs yield similar results, where LFPs from cortical layers III and IV seem better suited than those of deep layers. In addition, no hand-crafted features need to be extracted from the data—multi-unit activity can directly be fed into these networks and a simple event-encoding of LFPs is sufficient for good performance. Furthermore, we find that the performance of SNNs is insensitive to the network state—their performance is similar during UP and DOWN states.

Rapid and transient enhancement of thalamic information transmission induced by vagus nerve stimulation

OBJECTIVE

Vagus nerve stimulation (VNS) has been FDA-approved as a long-term, therapeutic treatment for multiple disorders, including pharmacoresistant epilepsy and depression. Here we elucidate the short-term effects of VNS on sensory processing.

APPROACH

We employed an information theoretic approach to examine the effects of VNS on thalamocortical transmission of sensory-related information along the somatosensory pathway.

MAIN RESULTS

We found that VNS enhanced the selectivity of the response of thalamic neurons to specific kinetic features in the stimuli, resulting in a significant increase in the efficiency and rate of stimulus-related information conveyed by thalamic spikes. VNS-induced improvements in thalamic sensory processing coincided with a decrease in thalamic burst firing. Importantly, we found VNS-induced enhancement of sensory processing had a rapid onset and offset, completely disappearing one minute after cessation of VNS. The timescales of these effects indicate against an underlying mechanism involving long-term neuroplasticity. We found several patterns of VNS (tonic, standard duty-cycle, and fast duty-cycle) all induced similar improvements in sensory processing. Under closer inspection we noticed that due to the fast timescale of VNS effects on sensory processing, standard duty-cycle VNS induced a fluctuating sensory processing state which may be sub-optimal for perceptual behavior. Fast duty-cycle VNS and continuous, tonic VNS induced quantitatively similar improvements in thalamic information transmission as standard duty-cycle VNS without inducing a fluctuating thalamic state. Further, we found the strength of VNS-induced improvements in sensory processing increased monotonically with amplitude and frequency of VNS.

SIGNIFICANCE

These results demonstrate, for the first time, the feasibility of utilizing specific patterns of VNS to rapidly improve sensory processing and confirm fast duty-cycle and tonic patterns as optimal for this purpose, while showing standard duty-cycle VNS causes non-optimal fluctuations in thalamic state.

Active High-Density Electrode Arrays: Technology and Applications in Neuronal Cell Cultures

Active high-density electrode arrays realized with complementary metal-oxide-semiconductor (CMOS) technology provide electrophysiological recordings from several thousands of closely spaced microelectrodes. This has drastically advanced the spatiotemporal recording resolution of conventional multielectrode arrays (MEAs). Thus, today's electrophysiology in neuronal cultures can exploit label-free electrical readouts from a large number of single neurons within the same network. This provides advanced capabilities to investigate the properties of self-assembling neuronal networks, to advance studies on neurotoxicity and neurodevelopmental alterations associated with human brain diseases, and to develop cell culture models for testing drug- or cell-based strategies for therapies.Here, after introducing the reader to this neurotechnology, we summarize the results of different recent studies demonstrating the potential of active high-density electrode arrays for experimental applications. We also discuss ongoing and possible future research directions that might allow for moving these platforms forward for screening applications.

State-aware detection of sensory stimuli in the cortex of the awake mouse

Cortical responses to sensory inputs vary across repeated presentations of identical stimuli, but how this trial-to-trial variability impacts detection of sensory inputs is not fully understood. Using multi-channel local field potential (LFP) recordings in primary somatosensory cortex (S1) of the awake mouse, we optimized a data-driven cortical state classifier to predict single-trial sensory-evoked responses, based on features of the spontaneous, ongoing LFP recorded across cortical layers. Our findings show that, by utilizing an ongoing prediction of the sensory response generated by this state classifier, an ideal observer improves overall detection accuracy and generates robust detection of sensory inputs across various states of ongoing cortical activity in the awake brain, which could have implications for variability in the performance of detection tasks across brain states.

Establishing the link between neural activity and behavior is a central goal of neuroscience. One context in which to examine this link is in a sensory detection task, in which an animal is trained to report the presence of a barely perceptible sensory stimulus. In such tasks, both sensory responses in the brain and behavioral responses are highly variable. A simple hypothesis, originating in signal detection theory, is that perceived inputs generate neural activity that cross some threshold for detection. According to this hypothesis, sensory response variability would predict behavioral variability, but previous studies have not born out this prediction. Further complicating the picture, sensory response variability is partially dependent on the ongoing state of cortical activity, and we wondered whether this could resolve the mismatch between response variability and behavioral variability. Here, we use a computational approach to study an adaptive observer that utilizes an ongoing prediction of sensory responsiveness to detect sensory inputs. This observer has higher overall accuracy than the standard ideal observer. Moreover, because of the adaptation, the observer breaks the direct link between neural and behavioral variability, which could resolve discrepancies arising in past studies. We suggest new experiments to test our theory.

Implantable Neural Probes for Brain-Machine Interfaces - Current Developments and Future Prospects

A Brain-Machine interface (BMI) allows for direct communication between the brain and machines. Neural probes for recording neural signals are among the essential components of a BMI system. In this report, we review research regarding implantable neural probes and their applications to BMIs. We first discuss conventional neural probes such as the tetrode, Utah array, Michigan probe, and electroencephalography (ECoG), following which we cover advancements in next-generation neural probes. These next-generation probes are associated with improvements in electrical properties, mechanical durability, biocompatibility, and offer a high degree of freedom in practical settings. Specifically, we focus on three key topics: (1) novel implantable neural probes that decrease the level of invasiveness without sacrificing performance, (2) multi-modal neural probes that measure both electrical and optical signals, (3) and neural probes developed using advanced materials. Because safety and precision are critical for practical applications of BMI systems, future studies should aim to enhance these properties when developing next-generation neural probes.

Intracortical Microstimulation Modulates Cortical Induced Responses

Recent advances in cortical prosthetics relied on intracortical microstimulation (ICMS) to activate the cortical neural network and convey information to the brain. Here we show that activity elicited by low-current ICMS modulates induced cortical responses to a sensory stimulus in the primary auditory cortex (A1). A1 processes sensory stimuli in a stereotyped manner, encompassing two types of activity: evoked activity (phase-locked to the stimulus) and induced activity (non-phase-locked to the stimulus). Time-frequency analyses of extracellular potentials recorded from all layers and the surface of the auditory cortex of anesthetized guinea pigs of both sexes showed that ICMS during the processing of a transient acoustic stimulus differentially affected the evoked and induced response. Specifically, ICMS enhanced the long-latency-induced component, mimicking physiological gain increasing top-down feedback processes. Furthermore, the phase of the local field potential at the time of stimulation was predictive of the response amplitude for acoustic stimulation, ICMS, as well as combined acoustic and electric stimulation. Together, this was interpreted as a sign that the response to electrical stimulation was integrated into the ongoing cortical processes in contrast to substituting them. Consequently, ICMS modulated the cortical response to a sensory stimulus. We propose such targeted modulation of cortical activity (as opposed to a stimulation that substitutes the ongoing processes) as an alternative approach for cortical prostheses.SIGNIFICANCE STATEMENT Intracortical microstimulation (ICMS) is commonly used to activate a specific subset of cortical neurons, without taking into account the ongoing activity at the time of stimulation. Here, we found that a low-current ICMS pulse modulated the way the auditory cortex processed a peripheral stimulus, by supra-additively combining the response to the ICMS with the cortical processing of the peripheral stimulus. This artificial modulation mimicked natural modulations of response magnitude such as attention or expectation. In contrast to what was implied in earlier studies, this shows that the response to electrical stimulation is not substituting ongoing cortical activity but is integrated into the natural processes.

State-dependent representation of stimulus-evoked activity in high-density recordings of neural cultures

Neuronal responses to external stimuli vary from trial to trial partly because they depend on continuous spontaneous variations of the state of neural circuits, reflected in variations of ongoing activity prior to stimulus presentation. Understanding how post-stimulus responses relate to the pre-stimulus spontaneous activity is thus important to understand how state dependence affects information processing and neural coding, and how state variations can be discounted to better decode single-trial neural responses. Here we exploited high-resolution CMOS electrode arrays to record simultaneously from thousands of electrodes in in-vitro cultures stimulated at specific sites. We used information-theoretic analyses to study how ongoing activity affects the information that neuronal responses carry about the location of the stimuli. We found that responses exhibited state dependence on the time between the last spontaneous burst and the stimulus presentation and that the dependence could be described with a linear model. Importantly, we found that a small number of selected neurons carry most of the stimulus information and contribute to the state-dependent information gain. This suggests that a major value of large-scale recording is that it individuates the small subset of neurons that carry most information and that benefit the most from knowledge of its state dependence.

Decoding English Alphabet Letters Using EEG Phase Information

Increasing evidence indicates that the phase pattern and power of the low frequency oscillations of brain electroencephalograms (EEG) contain significant information during the human cognition of sensory signals such as auditory and visual stimuli. Here, we investigate whether and how the letters of the alphabet can be directly decoded from EEG phase and power data. In addition, we investigate how different band oscillations contribute to the classification and determine the critical time periods. An English letter recognition task was assigned, and statistical analyses were conducted to decode the EEG signal corresponding to each letter visualized on a computer screen. We applied support vector machine (SVM) with gradient descent method to learn the potential features for classification. It was observed that the EEG phase signals have a higher decoding accuracy than the oscillation power information. Low-frequency theta and alpha oscillations have phase information with higher accuracy than do other bands. The decoding performance was best when the analysis period began from 180 to 380 ms after stimulus presentation, especially in the lateral occipital and posterior temporal scalp regions (PO7 and PO8). These results may provide a new approach for brain-computer interface techniques (BCI) and may deepen our understanding of EEG oscillations in cognition.

State-Dependent Decoding Algorithms Improve the Performance of a Bidirectional BMI in Anesthetized Rats

Brain-machine interfaces (BMIs) promise to improve the quality of life of patients suffering from sensory and motor disabilities by creating a direct communication channel between the brain and the external world. Yet, their performance is currently limited by the relatively small amount of information that can be decoded from neural activity recorded form the brain. We have recently proposed that such decoding performance may be improved when using state-dependent decoding algorithms that predict and discount the large component of the trial-to-trial variability of neural activity which is due to the dependence of neural responses on the network's current internal state. Here we tested this idea by using a bidirectional BMI to investigate the gain in performance arising from using a state-dependent decoding algorithm. This BMI, implemented in anesthetized rats, controlled the movement of a dynamical system using neural activity decoded from motor cortex and fed back to the brain the dynamical system's position by electrically microstimulating somatosensory cortex. We found that using state-dependent algorithms that tracked the dynamics of ongoing activity led to an increase in the amount of information extracted form neural activity by 22%, with a consequently increase in all of the indices measuring the BMI's performance in controlling the dynamical system. This suggests that state-dependent decoding algorithms may be used to enhance BMIs at moderate computational cost.