Galvanic vs. pulsatile effects on decision-making networks: reshaping the neural activation landscape.

Objective. Primarily due to safety concerns, biphasic pulsatile stimulation (PS) is the present standard for electrical excitation of neural tissue with a diverse set of applications. While pulses have been shown to be effective to achieve functional outcomes, they have well-known deficits. Due to recent technical advances, galvanic stimulation (GS), delivery of current for extended periods of time (>1 s), has re-emerged as an alternative to PS.Approach. In this paper, we use a winner-take-all decision-making cortical network model to investigate differences between pulsatile and GS in the context of a perceptual decision-making task.Main results. Based on previous work, we hypothesized that GS would produce more spatiotemporally distributed, network-sensitive neural responses, while PS would produce highly synchronized activation of a limited group of neurons. Our results in-silico support these hypotheses for low-amplitude GS but deviate when galvanic amplitudes are large enough to directly activate or block nearby neurons.Significance. We conclude that with careful parametrization, GS could overcome some limitations of PS to deliver more naturalistic firing patterns in the group of targeted neurons.

Difference in Network Effects of Pulsatile and Galvanic Stimulation.

Biphasic pulsatile stimulation is the present standard for neural prosthetic use, and it is used to understand connectivity and functionality of the brain in brain mapping studies. While pulses have been shown to drive behavioral changes, such as biasing decision making, they have deficits. For example, cochlear implants restore hearing but lack the ability to restore pitch perception. Recent work shows that pulses produce artificial synchrony in networks of neurons and non-linear changes in firing rate with pulse amplitude. Studies also show galvanic stimulation, delivery of current for extended periods of time, produces more naturalistic behavioral responses than pulses. In this paper, we use a winner-take-all decision-making network model to investigate differences between pulsatile and galvanic stimulation at the single neuron and network level while accurately modeling the effects of pulses on neurons for the first time. Results show pulses bias spike timing and make neurons more resistive to natural network inputs than galvanic stimulation at an equivalent current amplitude. Clinical Relevance- This establishes that pulsatile stimulation may disrupt natural spike timing and network-level interactions while certain parameterizations of galvanic stimulation avoid these effects and can drive network firing more naturally.

A Machine Learning-based Neural Implant Front End for Inducing Naturalistic Firing.

Despite being able to restore speech perception with 99% success rate, cochlear implants cannot successfully restore pitch perception or music appreciation. Studies suggest that if auditory neurons were activated with fine timing closer to that of natural responses pitch would be restored. Predicting the timing of cochlear responses requires detailed biophysical models of sound transmission, inner hair cell responses, and outer hair cell responses. Performing these calculations is computationally costly for real time cochlear implant stimulation. Instead, implants typically modulate pulse amplitude of fixed pulse rate stimulation with the band-limited envelopes of incoming sound. This method is known to produce unrealistic responses, even to simple step inputs. Here we investigate using a machine learning algorithm to optimize the prediction of the desired firing patterns of the auditory afferents in response to sinusoidal and step modulation of pure tones. We conclude that a trained network that consists of 25 GRU nodes can reproduce fine timing with 4.4 percent error on a test set of sines and steps. This trained network can also transfer learn and capture features of natural sounds that are not captured by standard CI algorithms. Additionally, for 0.5 second test inputs, the ML algorithm completed the sound to spike rate conversion in 300x less time than the phenomenological model. This calculation occurs at a real-time compatible rate of 1 ms for 1 second of spike timing prediction on an i9 microprocessor. This suggests that this is a feasible approach to pursue for real-time CI implementation.

Direct current effects on afferent and hair cell to elicit natural firing patterns.

In contrast to the conventional pulsatile neuromodulation that excites neurons, galvanic or direct current stimulation can excite, inhibit, or sensitize neurons. The vestibular system presents an excellent system for studying galvanic neural interface due to the spontaneously firing afferent activity that needs to be either suppressed or excited to convey head motion sensation. We determine the cellular mechanisms underlying the beneficial properties of galvanic vestibular stimulation (GVS) by creating a computational model of the vestibular end organ that elicits all experimentally observed response characteristics to GVS simultaneously. When GVS was modeled to affect the axon alone, the complete experimental data could not be replicated. We found that if GVS affects hair cell vesicle release and axonal excitability simultaneously, our modeling results matched all experimental observations. We conclude that contrary to the conventional belief that GVS affects only axons, the hair cells are likely also affected by this stimulation paradigm.

Predicting Response of Spontaneously Firing Afferents to Prosthetic Pulsatile Stimulation.

Pulsatile electrical stimulation is used in neural prostheses such as the vestibular prosthesis. In a healthy vestibular system, head motion is encoded by changes in the firing rates of afferents around their spontaneous baseline rate. For people suffering from bilateral vestibular disorder (BVD), head motion no longer modulates firing rate. Vestibular prostheses use a gyroscope to detect head motion and stimulate neurons directly in a way that mimics natural modulation. Proper restoration of vestibular function relies on the ability of stimulation to evoke the same firing patterns as the healthy system. For this reason, it is necessary to understand what firing rates are produced for different stimulation parameters. Two stimulation parameters commonly controlled in pulsatile neuromodulation are pulse rate and pulse amplitude. Previous neural recording experiments in the vestibular nerve contradict widely held assumptions about the relationship between pulse rates and evoked spike activity, and the relationship between pulse amplitude and neural activity has not been explored. Here we use a well-established computational model of the vestibular afferent to simulate responses to different pulse rates and amplitudes. We confirm that our simulated neural results agree with the existing experimental data. Finally, we developed the "Action Potential Collision" (APC) equation that defines induced firing as a function of spontaneous firing rate, pulse rate, and pulse amplitude. We show that this relationship can successfully predict simulated vestibular activity by accounting for interactions between pulses and spontaneous firing.

Learning to control the brain through adaptive closed-loop patterned stimulation.

OBJECTIVE: Stimulation of neural activity is an important scientific and clinical tool, causally testing hypotheses and treating neurodegenerative and neuropsychiatric diseases. However, current stimulation approaches cannot flexibly control the pattern of activity in populations of neurons. To address this, we developed a model-free, adaptive, closed-loop stimulation (ACLS) system that learns to use multi-site electrical stimulation to control the pattern of activity of a population of neurons. APPROACH: The ACLS system combined multi-electrode electrophysiological recordings with multi-site electrical stimulation to simultaneously record the activity of a population of 5-15 multiunit neurons and deliver spatially-patterned electrical stimulation across 4-16 sites. Using a closed-loop learning system, ACLS iteratively updated the pattern of stimulation to reduce the difference between the observed neural response and a specific target pattern of firing rates in the recorded multiunits. MAIN RESULTS: In silico and in vivo experiments showed ACLS learns to produce specific patterns of neural activity (in ∼15 min) and was robust to noise and drift in neural responses. In visual cortex of awake mice, ACLS learned electrical stimulation patterns that produced responses similar to the natural response evoked by visual stimuli. Similar to how repetition of a visual stimulus causes an adaptation in the neural response, the response to electrical stimulation was adapted when it was preceded by the associated visual stimulus. SIGNIFICANCE: Our results show an ACLS system that can learn, in real-time, to generate specific patterns of neural activity. This work provides a framework for using model-free closed-loop learning to control neural activity.

Characterizing and predicting cortical evoked responses to direct electrical stimulation of the human brain.

BACKGROUND: Direct electrical stimulation of the human brain has been used to successfully treat several neurological disorders, but the precise effects of stimulation on neural activity are poorly understood. Characterizing the neural response to stimulation, however, could allow clinicians and researchers to more accurately predict neural responses, which could in turn lead to more effective stimulation for treatment and to fundamental knowledge regarding neural function. OBJECTIVE: Here we use a linear systems approach in order to characterize the response to electrical stimulation across cortical locations and then to predict the responses to novel inputs. METHODS: We use intracranial electrodes to directly stimulate the human brain with single pulses of stimulation using amplitudes drawn from a random distribution. Based on the evoked responses, we generate a simple model capturing the characteristic response to stimulation at each cortical site. RESULTS: We find that the variable dynamics of the evoked response across cortical locations can be captured using the same simple architecture, a linear time-invariant system that operates separately on positive and negative input pulses of stimulation. We demonstrate that characterizing the response to stimulation using this simple and tractable model of evoked responses enables us to predict the responses to subsequent stimulation with single pulses with novel amplitudes, and the compound response to stimulation with multiple pulses. CONCLUSION: Our data suggest that characterizing the response to stimulation in an approximately linear manner can provide a powerful and principled approach for predicting the response to direct electrical stimulation.

Fly wing vein patterns have spatial reproducibility of a single cell.

Developmental processes in multicellular organisms occur in fluctuating environments and are prone to noise, yet they produce complex patterns with astonishing reproducibility. We measure the left-right and inter-individual precision of bilaterally symmetric fly wings across the natural range of genetic and environmental conditions and find that wing vein patterns are specified with identical spatial precision and are reproducible to within a single-cell width. The early fly embryo operates at a similar degree of reproducibility, suggesting that the overall spatial precision of morphogenesis in Drosophila performs at the single-cell level. Could development be operating at the physical limit of what a biological system can achieve?