Identification of Stable Spike-Timing-Dependent Plasticity from Spiking Activity with Generalized Multilinear Modeling

Efficient inference of synaptic plasticity rule with Gaussian process regression

Finding the form of synaptic plasticity is critical to understanding its functions underlying learning and memory. We investigated an efficient method to infer synaptic plasticity rules in various experimental settings. We considered biologically plausible models fitting a wide range of in-vitro studies and examined the recovery of their firing-rate dependence from sparse and noisy data. Among the methods assuming low-rankness or smoothness of plasticity rules, Gaussian process regression (GPR), a nonparametric Bayesian approach, performs the best. Under the conditions measuring changes in synaptic weights directly or measuring changes in neural activities as indirect observables of synaptic plasticity, which leads to different inference problems, GPR performs well. Also, GPR could simultaneously recover multiple plasticity rules and robustly perform under various plasticity rules and noise levels. Such flexibility and efficiency, particularly at the low sampling regime, make GPR suitable for recent experimental developments and inferring a broader class of plasticity models.

Low-dimensional encoding of decisions in parietal cortex reflects long-term training history

Neurons in parietal cortex exhibit task-related activity during decision-making tasks. However, it remains unclear how long-term training to perform different tasks over months or even years shapes neural computations and representations. We examine lateral intraparietal area (LIP) responses during a visual motion delayed-match-to-category task. We consider two pairs of male macaque monkeys with different training histories: one trained only on the categorization task, and another first trained to perform fine motion-direction discrimination (i.e., pretrained). We introduce a novel analytical approach-generalized multilinear models-to quantify low-dimensional, task-relevant components in population activity. During the categorization task, we found stronger cosine-like motion-direction tuning in the pretrained monkeys than in the category-only monkeys, and that the pretrained monkeys' performance depended more heavily on fine discrimination between sample and test stimuli. These results suggest that sensory representations in LIP depend on the sequence of tasks that the animals have learned, underscoring the importance of considering training history in studies with complex behavioral tasks.

A Computational Model of Working Memory Based on Spike-Timing-Dependent Plasticity

Working memory is closely involved in various cognitive activities, but its neural mechanism is still under exploration. The mainstream view has long been that persistent activity is the neural basis of working memory, but recent experiments have observed that activity-silent memory can also be correctly recalled. The underlying mechanism of activity-silent memory is considered to be an alternative scheme that rejects the theory of persistent activity. We propose a working memory model based on spike-timing-dependent plasticity (STDP). Different from models based on spike-rate coding, our model adopts temporal patterns of action potentials to represent information, so it can flexibly encode new memory representation. The model can work in both persistent and silent states, i.e., it is compatible with both of these seemingly conflicting neural mechanisms. We conducted a simulation experiment, and the results are similar to the real experimental results, which suggests that our model is plausible in biology.

Modeling the Short-Term Dynamics of in Vivo Excitatory Spike Transmission

Information transmission in neural networks is influenced by both short-term synaptic plasticity (STP) as well as nonsynaptic factors, such as after-hyperpolarization currents and changes in excitability. Although these effects have been widely characterized in vitro using intracellular recordings, how they interact in vivo is unclear. Here, we develop a statistical model of the short-term dynamics of spike transmission that aims to disentangle the contributions of synaptic and nonsynaptic effects based only on observed presynaptic and postsynaptic spiking. The model includes a dynamic functional connection with short-term plasticity as well as effects due to the recent history of postsynaptic spiking and slow changes in postsynaptic excitability. Using paired spike recordings, we find that the model accurately describes the short-term dynamics of in vivo spike transmission at a diverse set of identified and putative excitatory synapses, including a pair of connected neurons within thalamus in mouse, a thalamocortical connection in a female rabbit, and an auditory brainstem synapse in a female gerbil. We illustrate the utility of this modeling approach by showing how the spike transmission patterns captured by the model may be sufficient to account for stimulus-dependent differences in spike transmission in the auditory brainstem (endbulb of Held). Finally, we apply this model to large-scale multielectrode recordings to illustrate how such an approach has the potential to reveal cell type-specific differences in spike transmission in vivo Although STP parameters estimated from ongoing presynaptic and postsynaptic spiking are highly uncertain, our results are partially consistent with previous intracellular observations in these synapses.SIGNIFICANCE STATEMENT Although synaptic dynamics have been extensively studied and modeled using intracellular recordings of postsynaptic currents and potentials, inferring synaptic effects from extracellular spiking is challenging. Whether or not a synaptic current contributes to postsynaptic spiking depends not only on the amplitude of the current, but also on many other factors, including the activity of other, typically unobserved, synapses, the overall excitability of the postsynaptic neuron, and how recently the postsynaptic neuron has spiked. Here, we developed a model that, using only observations of presynaptic and postsynaptic spiking, aims to describe the dynamics of in vivo spike transmission by modeling both short-term synaptic plasticity (STP) and nonsynaptic effects. This approach may provide a novel description of fast, structured changes in spike transmission.

Designing Patient-Specific Optimal Neurostimulation Patterns for Seizure Suppression

Neurostimulation is a promising therapy for abating epileptic seizures. However, it is extremely difficult to identify optimal stimulation patterns experimentally. In this study, human recordings are used to develop a functional 24 neuron network statistical model of hippocampal connectivity and dynamics. Spontaneous seizure-like activity is induced in silico in this reconstructed neuronal network. The network is then used as a testbed to design and validate a wide range of neurostimulation patterns. Commonly used periodic trains were not able to permanently abate seizures at any frequency. A simulated annealing global optimization algorithm was then used to identify an optimal stimulation pattern, which successfully abated 92% of seizures. Finally, in a fully responsive, or closed-loop, neurostimulation paradigm, the optimal stimulation successfully prevented the network from entering the seizure state. We propose that the framework presented here for algorithmically identifying patient-specific neurostimulation patterns can greatly increase the efficacy of neurostimulation devices for seizures.

Acute in vivo testing of a conformal polymer microelectrode array for multi-region hippocampal recordings

OBJECTIVE

The success of a cortical prosthetic device relies upon its ability to attain resolvable spikes from many neurons in particular neural networks over long periods of time. Traditionally, lifetimes of neural recordings are greatly limited by the body's immune response against the foreign implant which causes neuronal death and glial scarring. This immune reaction is posited to be exacerbated by micromotion between the implant, which is often rigid, and the surrounding, soft brain tissue, and attenuates the quality of recordings over time.

APPROACH

In an attempt to minimize the foreign body response to a penetrating neural array that records from multiple brain regions, Parylene C, a flexible, biocompatible polymer was used as the substrate material for a functional, proof-of-concept neural array with a reduced elastic modulus. This probe array was designed and fabricated to have 64 electrodes positioned to match the anatomy of the rat hippocampus and allow for simultaneous recordings between two cell-body layers of interest. A dissolvable brace was used for deep-brain penetration of the flexible array.

MAIN RESULTS

Arrays were electrochemically characterized at the benchtop, and a novel insertion technique that restricts acute insertion injury enabled accurate target placement of four, bare, flexible arrays to greater than 4 mm deep into the rat brain. Arrays were tested acutely and in vivo recordings taken intra-operatively reveal spikes in both targeted regions of the hippocampus with spike amplitudes and noise levels similar to those recorded with microwires. Histological staining of a sham array implanted for one month reveals limited astrocytic scarring and neuronal death around the implant.

SIGNIFICANCE

This work represents one of the first examples of a penetrating polymer probe array that records from individual neurons in structures that lie deep within the brain.

Using computational theory to constrain statistical models of neural data

Computational neuroscience is, to first order, dominated by two approaches: the 'bottom-up' approach, which searches for statistical patterns in large-scale neural recordings, and the 'top-down' approach, which begins with a theory of computation and considers plausible neural implementations. While this division is not clear-cut, we argue that these approaches should be much more intimately linked. From a Bayesian perspective, computational theories provide constrained prior distributions on neural data-albeit highly sophisticated ones. By connecting theory to observation via a probabilistic model, we provide the link necessary to test, evaluate, and revise our theories in a data-driven and statistically rigorous fashion. This review highlights examples of this theory-driven pipeline for neural data analysis in recent literature and illustrates it with a worked example based on the temporal difference learning model of dopamine.