Faculty Bibliography

We developed a detailed model of macaque auditory thalamocortical circuits, including primary auditory cortex (A1), medial geniculate body (MGB), and thalamic reticular nucleus, utilizing the NEURON simulator and NetPyNE tool. The A1 model simulates a cortical column with over 12,000 neurons and 25 million synapses, incorporating data on cell-type-specific neuron densities, morphology, and connectivity across six cortical layers. It is reciprocally connected to the MGB thalamus, which includes interneurons and core and matrix-layer-specific projections to A1. The model simulates multiscale measures, including physiological firing rates, local field potentials (LFPs), current source densities (CSDs), and electroencephalography (EEG) signals. Laminar CSD patterns, during spontaneous activity and in response to broadband noise stimulus trains, mirror experimental findings. Physiological oscillations emerge spontaneously across frequency bands comparable to those recorded in vivo. We elucidate population-specific contributions to observed oscillation events and relate them to firing and presynaptic input patterns. The model offers a quantitative theoretical framework to integrate and interpret experimental data and predict its underlying cellular and circuit mechanisms.

Understanding cortical function requires studying multiple scales: molecular, cellular, circuit, and behavioral. We develop a multiscale, biophysically detailed model of mouse primary motor cortex (M1) with over 10,000 neurons and 30 million synapses. Neuron types, densities, spatial distributions, morphologies, biophysics, connectivity, and dendritic synapse locations are constrained by experimental data. The model includes long-range inputs from seven thalamic and cortical regions and noradrenergic inputs. Connectivity depends on cell class and cortical depth at sublaminar resolution. The model accurately predicts in vivo layer- and cell-type-specific responses (firing rates and LFP) associated with behavioral states (quiet wakefulness and movement) and experimental manipulations (noradrenaline receptor blockade and thalamus inactivation). We generate mechanistic hypotheses underlying the observed activity and analyzed low-dimensional population latent dynamics. This quantitative theoretical framework can be used to integrate and interpret M1 experimental data and sheds light on the cell-type-specific multiscale dynamics associated with several experimental conditions and behaviors.

Recent models of spiking neuronal networks have been trained to perform behaviors in static environments using a variety of learning rules, with varying degrees of biological realism. Most of these models have not been tested in dynamic visual environments where models must make predictions on future states and adjust their behavior accordingly. The models using these learning rules are often treated as black boxes, with little analysis on circuit architectures and learning mechanisms supporting optimal performance. Here we developed visual/motor spiking neuronal network models and trained them to play a virtual racket-ball game using several reinforcement learning algorithms inspired by the dopaminergic reward system. We systematically investigated how different architectures and circuit-motifs (feed-forward, recurrent, feedback) contributed to learning and performance. We also developed a new biologically-inspired learning rule that significantly enhanced performance, while reducing training time. Our models included visual areas encoding game inputs and relaying the information to motor areas, which used this information to learn to move the racket to hit the ball. Neurons in the early visual area relayed information encoding object location and motion direction across the network. Neuronal association areas encoded spatial relationships between objects in the visual scene. Motor populations received inputs from visual and association areas representing the dorsal pathway. Two populations of motor neurons generated commands to move the racket up or down. Model-generated actions updated the environment and triggered reward or punishment signals that adjusted synaptic weights so that the models could learn which actions led to reward. Here we demonstrate that our biologically-plausible learning rules were effective in training spiking neuronal network models to solve problems in dynamic environments. We used our models to dissect the circuit architectures and learning rules most effective for learning. Our model shows that learning mechanisms involving different neural circuits produce similar performance in sensory-motor tasks. In biological networks, all learning mechanisms may complement one another, accelerating the learning capabilities of animals. Furthermore, this also highlights the resilience and redundancy in biological systems.

Electrophysiological oscillations in the brain have been shown to occur as multi-cycle events, with onset and offset dependent on behavioral and cognitive state. To provide a baseline for state-related and task-related events, we quantified oscillation features in resting-state recordings. We developed an open-source wavelet-based tool to detect and characterize such oscillation events (OEvents) and exemplify the use of this tool in both simulations and two invasively-recorded electrophysiology datasets: one from human, and one from non-human primate auditory system. After removing incidentally occurring event related potentials, we used OEvents to quantify oscillation features. We identified about 2 million oscillation events, classified within traditional frequency bands: delta, theta, alpha, beta, low gamma, gamma, and high gamma. Oscillation events of 1-44 cycles could be identified in at least one frequency band 90% of the time in human and non-human primate recordings. Individual oscillation events were characterized by non-constant frequency and amplitude. This result necessarily contrasts with prior studies which assumed frequency constancy, but is consistent with evidence from event-associated oscillations. We measured oscillation event duration, frequency span, and waveform shape. Oscillations tended to exhibit multiple cycles per event, verifiable by comparing filtered to unfiltered waveforms. In addition to the clear intra-event rhythmicity, there was also evidence of inter-event rhythmicity within bands, demonstrated by finding that coefficient of variation of interval distributions and Fano Factor measures differed significantly from a Poisson distribution assumption. Overall, our study provides an easy-to-use tool to study oscillation events at the single-trial level or in ongoing recordings, and demonstrates that rhythmic, multi-cycle oscillation events dominate auditory cortical dynamics.Significance StatementTo provide a baseline for auditory system cortical dynamics, we quantified neuronal oscillation event features in resting-state recordings of the auditory system. We found that even at rest, event-like oscillations are the dominant operational mode of the auditory cortex in both humans and non-human primates. Our results highlight the importance of the auditory system's rhythmic neuronal fluctuations in setting the context on top of which auditory processing necessary for behavior and cognition occurs. In addition, we demonstrate the importance of studying basic features of oscillation events in ongoing and single-trial recordings to understand their role in cognition and the mechanisms generating them.

Artificial neural networks (ANNs) have been successfully trained to perform a wide range of sensory-motor behaviors. In contrast, the performance of spiking neuronal network (SNN) models trained to perform similar behaviors remains relatively suboptimal. In this work, we aimed to push the field of SNNs forward by exploring the potential of different learning mechanisms to achieve optimal performance. We trained SNNs to solve the CartPole reinforcement learning (RL) control problem using two learning mechanisms operating at different timescales: (1) spike-timing-dependent reinforcement learning (STDP-RL) and (2) evolutionary strategy (EVOL). Though the role of STDP-RL in biological systems is well established, several other mechanisms, though not fully understood, work in concert during learning in vivo. Recreating accurate models that capture the interaction of STDP-RL with these diverse learning mechanisms is extremely difficult. EVOL is an alternative method and has been successfully used in many studies to fit model neural responsiveness to electrophysiological recordings and, in some cases, for classification problems. One advantage of EVOL is that it may not need to capture all interacting components of synaptic plasticity and thus provides a better alternative to STDP-RL. Here, we compared the performance of each algorithm after training, which revealed EVOL as a powerful method for training SNNs to perform sensory-motor behaviors. Our modeling opens up new capabilities for SNNs in RL and could serve as a testbed for neurobiologists aiming to understand multi-timescale learning mechanisms and dynamics in neuronal circuits.

Motor cortical activity in the beta frequency range is one of the strongest and most studied movement-related neural signals. At the single trial level, beta band activity is often characterized by transient, high amplitude, bursting events rather than slowly modulating oscillations. The timing of these bursting events is tightly linked to behavior, suggesting a more dynamic functional role for beta activity than previously believed. However, the neural mechanisms underlying beta bursts in sensorimotor circuits are poorly understood. To address this, we here leverage and extend recent developments in high precision MEG for temporally resolved laminar analysis of burst activity, combined with a neocortical circuit model that simulates the biophysical generators of the electrical currents which drive beta bursts. This approach pinpoints the generation of beta bursts in human motor cortex to distinct excitatory synaptic inputs to deep and superficial cortical layers, which drive current flow in opposite directions. These laminar dynamics of beta bursts in motor cortex align with prior invasive animal recordings within the somatosensory cortex, and suggest a conserved mechanism for somatosensory and motor cortical beta bursts. More generally, we demonstrate the ability for uncovering the laminar dynamics of event-related neural signals in human non-invasive recordings. This provides important constraints to theories about the functional role of burst activity for movement control in health and disease, and crucial links between macro-scale phenomena measured in humans and micro-circuit activity recorded from animal models.

Pyramidal neurons in neocortex have complex input-output relationships that depend on their morphologies, ion channel distributions, and the nature of their inputs, but which cannot be replicated by simple integrate-and-fire models. The impedance properties of their dendritic arbors, such as resonance and phase shift, shape neuronal responses to synaptic inputs and provide intraneuronal functional maps reflecting their intrinsic dynamics and excitability. Experimental studies of dendritic impedance have shown that neocortical pyramidal tract neurons exhibit distance-dependent changes in resonance and impedance phase with respect to the soma. We therefore investigated how well several biophysically-detailed multi-compartment models of neocortical layer 5 pyramidal tract neurons reproduce the location-dependent impedance profiles observed experimentally. Each model tested here exhibited location-dependent impedance profiles, but most captured either the observed impedance amplitude or phase, not both. The only model that captured features from both incorporates HCN channels and a shunting current, like that produced by Twik-related acid-sensitive K⁺ (TASK) channels. TASK-like channel density in this model was proportional to local HCN channel density. We found that while this shunting current alone is insufficient to produce resonance or realistic phase response, it modulates all features of dendritic impedance, including resonance frequencies, resonance strength, synchronous frequencies, and total inductive phase. We also explored how the interaction of HCN channel current (I_h) and a TASK-like shunting current shape synaptic potentials and produce degeneracy in dendritic impedance profiles, wherein different combinations of I_h and shunting current can produce the same impedance profile.

Treatment of schizophrenia has had limited success in treating core cognitive symptoms. The evidence of multi-gene involvement suggests that multi-target therapy may be needed. Meanwhile, the complexity of schizophrenia pathophysiology and psychopathology, coupled with the species-specificity of much of the symptomatology, places limits on analysis via animal models, in vitro assays, and patient assessment. Multiscale computer modeling complements these traditional modes of study. Using a hippocampal CA3 computer model with 1200 neurons, we examined the effects of alterations in NMDAR, HCN (I_h current), and GABA_AR on information flow (measured with normalized transfer entropy), and in gamma activity in local field potential (LFP). We found that altering NMDARs, GABA_AR, I_h, individually or in combination, modified information flow in an inverted-U shape manner, with information flow reduced at low and high levels of these parameters. Theta-gamma phase-amplitude coupling also had an inverted-U shape relationship with NMDAR augmentation. The strong information flow was associated with an intermediate level of synchrony, seen as an intermediate level of gamma activity in the LFP, and an intermediate level of pyramidal cell excitability. Our results are consistent with the idea that overly low or high gamma power is associated with pathological information flow and information processing. These data suggest the need for careful titration of schizophrenia pharmacotherapy to avoid extremes that alter information flow in different ways. These results also identify gamma power as a potential biomarker for monitoring pathology and multi-target pharmacotherapy.

BACKGROUND:Mismatch negativity (MMN) is an extensively validated biomarker of cognitive function across both normative and clinical populations and has previously been localized to supratemporal auditory cortex. MMN is thought to represent a comparison of the features of the present stimulus versus a mnemonic template formed by the prior stimuli. METHODS:We used concurrent thalamic and primary auditory cortical (A1) laminar recordings in 7 macaques to evaluate the relative contributions of core (lemniscal) and matrix (nonlemniscal) thalamic afferents to MMN generation. RESULTS:We demonstrated that deviance-related activity is observed mainly in matrix regions of auditory thalamus, MMN generators are most prominent in layer 1 of cortex as opposed to sensory responses that activate layer 4 first and sequentially all cortical layers, and MMN is elicited independent of the frequency tuning of A1 neuronal ensembles. Consistent with prior reports, MMN-related thalamocortical activity was strongly inhibited by ketamine. CONCLUSIONS:Taken together, our results demonstrate distinct matrix versus core thalamocortical circuitry underlying the generation of a higher-order brain response (MMN) versus sensory responses.

INTRODUCTION:(CB) might be a susceptibility factor for AD. The subiculum is affected early in AD, for unknown reasons. METHODS:-buffering capacity in subicular neurons. CB expression levels in wild-type and AD mice were also analyzed. RESULTS:extrusion pumps rather than by buffers. DISCUSSION:homeostasis in AD has an age dependency that comprises multiple mechanisms, including compensatory processes.