Inhibition has long been thought to stabilize the activity of cortical networks at low rates, and to shape significantly their response to sensory inputs. In this talk, I will describe three recent collaborative projects that shed light on these issues. (1) I will show how optogenetic excitation of inhibition neurons is consistent with cortex being inhibition stabilized even in the absence of sensory inputs, and how this data can constrain the coupling strengths of E-I cortical network models. (2) Recent analysis of the effects of optogenetic excitation of pyramidal cells in V1 of mice and monkeys shows that in some cases this optogenetic input reshuffles the firing rates of neurons of the network, leaving the distribution of rates unaffected. I will show how this surprising effect can be reproduced in sufficiently strongly coupled E-I networks. (3) Another puzzle has been to understand the respective roles of different inhibitory subtypes in network stabilization. Recent data reveal a novel, state dependent, paradoxical effect of weakening AMPAR mediated synaptic currents onto SST cells. Mathematical analysis of a network model with multiple inhibitory cell types shows that this effect tells us in which conditions SST cells are required for network stabilization.

Higher cortical areas carry a wide range of sensory, cognitive, and motor signals supporting complex goal-directed behavior. These signals mix in heterogeneous responses of single neurons, making it difficult to untangle underlying mechanisms. I will present two approaches for revealing interpretable circuit mechanisms from heterogeneous neural responses during cognitive tasks. First, I will show a flexible nonparametric framework for simultaneously inferring population dynamics on single trials and tuning functions of individual neurons to the latent population state. When applied to recordings from the premotor cortex during decision-making, our approach revealed that populations of neurons encoded the same dynamic variable predicting choices, and heterogeneous firing rates resulted from the diverse tuning of single neurons to this decision variable. The inferred dynamics indicated an attractor mechanism for decision computation. Second, I will show an approach for inferring an interpretable network model of a cognitive task—the latent circuit—from neural response data. We developed a theory to causally validate latent circuit mechanisms via patterned perturbations of activity and connectivity in the high-dimensional network. This work opens new possibilities for deriving testable mechanistic hypotheses from complex neural response data.

Neural activity is often described in terms of population-level factors extracted from the responses of many neurons. Factors provide a lower-dimensional description with the aim of shedding light on network computations. Yet, mechanistically, computations are performed not by continuously valued factors but by interactions among neurons that spike discretely and variably. Models provide a means of bridging these levels of description. We developed a general method for training model networks of spiking neurons by leveraging factors extracted from either data or firing-rate-based networks. In addition to providing a useful model-building framework, this formalism illustrates how reliable and continuously valued factors can arise from seemingly stochastic spiking. Our framework establishes procedures for embedding this property in network models with different levels of realism. The relationship between spikes and factors in such networks provides a foundation for interpreting (and subtly redefining) commonly used quantities such as firing rates.

A key issue in understanding circuit operations is the extent to which neuronal spiking reflects local computation or responses to upstream inputs. Several studies have lesioned or silenced inputs to area CA1 of the hippocampus - either area CA3 or the entorhinal cortex and examined the effect on CA1 pyramidal cells. However, the types of the reported physiological impairments vary widely, primarily because simultaneous manipulations of these redundant inputs have never been performed. In this study, I combined optogenetic silencing of unilateral and bilateral mEC, of the local CA1 region, and performed bilateral pharmacogenetic silencing of CA3. I combined this with high spatial resolution extracellular recordings along the CA1-dentate axis. Silencing the medial entorhinal largely abolished extracellular theta and gamma currents in CA1, without affecting firing rates. In contrast, CA3 and local CA1 silencing strongly decreased firing of CA1 neurons without affecting theta currents. Each perturbation reconfigured the CA1 spatial map. Yet, the ability of the CA1 circuit to support place field activity persisted, maintaining the same fraction of spatially tuned place fields. In contrast to these results, unilateral mEC manipulations that were ineffective in impacting place cells during awake behavior were found to alter sharp-wave ripple sequences activated during sleep. Thus, intrinsic excitatory-inhibitory circuits within CA1 can generate neuronal assemblies in the absence of external inputs, although external synaptic inputs are critical to reconfigure (remap) neuronal assemblies in a brain-state dependent manner.

In understanding circuit operations, a key issue is the extent to which neuronal spiking reflects local computation or responses to upstream inputs. Because pyramidal cells in CA1 do not have local recurrent projections, it is currently assumed that firing in CA1 is inherited from its inputs – thus, entorhinal inputs provide communication with the rest of the neocortex and the outside world, whereas CA3 inputs provide internal and past memory representations. Several studies have attempted to prove this hypothesis, by lesioning or silencing either area CA3 or the entorhinal cortex and examining the effect of firing on CA1 pyramidal cells. Despite the intense and careful work in this research area, the magnitudes and types of the reported physiological impairments vary widely across experiments. At least part of the existing variability and conflicts is due to the different behavioral paradigms, designs and evaluation methods used by different investigators. Simultaneous manipulations in the same animal or even separate manipulations of the different inputs to the hippocampal circuits in the same experiment are rare. To address these issues, I used optogenetic silencing of unilateral and bilateral mEC, of the local CA1 region, and performed bilateral pharmacogenetic silencing of the entire CA3 region. I combined this with high spatial resolution recording of local field potentials (LFP) in the CA1-dentate axis and simultaneously collected firing pattern data from thousands of single neurons. Each experimental animal had up to two of these manipulations being performed simultaneously. Silencing the medial entorhinal (mEC) largely abolished extracellular theta and gamma currents in CA1, without affecting firing rates. In contrast, CA3 and local CA1 silencing strongly decreased firing of CA1 neurons without affecting theta currents. Each perturbation reconfigured the CA1 spatial map. Yet, the ability of the CA1 circuit to support place field activity persisted, maintaining the same fraction of spatially tuned place fields, and reliable assembly expression as in the intact mouse. Thus, the CA1 network can maintain autonomous computation to support coordinated place cell assemblies without reliance on its inputs, yet these inputs can effectively reconfigure and assist in maintaining stability of the CA1 map.

Optogenetics is a powerful tool that allows experimentalists to perturb neural circuits. What can we learn about a network from observing its response to perturbations? I will first describe the results of optogenetic activation of inhibitory neurons in mice cortex, and show that the results are consistent with inhibition stabilization. I will then move to experiments in which excitatory neurons are activated optogenetically, with or without visual inputs, in mice and monkeys. In some conditions, these experiments show a surprising result that the distribution of firing rates is not significantly changed by stimulation, even though firing rates of individual neurons are strongly modified. I will show in which conditions a network model of excitatory and inhibitory neurons can reproduce this feature.

Neurons in the suprachiasmatic nucleus (SCN, the brain’s master circadian clock) display a 24 hour cycle in the their rate of action potential discharge whereby firing rates are high during the light phase and lower during the dark phase. Although it is generally agreed that this cycle of activity is a key mediator of the clock’s neural and humoral output, surprisingly little is known about how changes in clock electrical activity can mediate scheduled physiological changes at different times of day. Using opto- and chemogenetic approaches in mice we have shown that the onset of electrical activity in vasopressin releasing SCN neurons near Zeitgeber time 22 (ZT22) activates glutamatergic thirst-promoting neurons in the OVLT (organum vasculosum lamina terminalis) to promote water intake prior to sleep. This effect is mediated by activity-dependent release of vasopressin from the axon terminals of SCN neurons which acts as a neurotransmitter on OVLT neurons. More recently we found that the clock receives excitatory input from a different subset of sodium sensing neurons in the OVLT. Activation of these neurons by a systemic salt load delivered at ZT19 stimulated the electrical activity of SCN neurons which are normally silent at this time. Remarkably, this effect induced an acute reduction in non-shivering thermogenesis and body temperature, which is an adaptive response to the salt load. These findings provide information regarding the mechanisms by which the SCN promotes scheduled physiological rhythms and indicates that the clock’s output circuitry can also be recruited to mediate an unscheduled homeostatic response.

Activity dependent synaptic plasticity is considered to be a primary mechanism underlying learning and memory. Yet it is unclear whether plasticity rules such as STDP measured in vitro apply in vivo. Network models with STDP predict that activity patterns (e.g., place-cell spatial selectivity) should change much faster than observed experimentally. We address this gap by investigating a nonlinear calcium-based plasticity rule fit to experiments done in physiological conditions. In this model, LTP and LTD result from intracellular calcium transients arising almost exclusively from synchronous coactivation of pre- and postsynaptic neurons. We analytically approximate the full distribution of nonlinear calcium transients as a function of pre- and postsynaptic firing rates, and temporal correlations. This analysis directly relates activity statistics that can be measured in vivo to the changes in synaptic efficacy they cause. Our results highlight that both high-firing rates and temporal correlations can lead to significant changes to synaptic efficacy. Using a mean-field theory, we show that the nonlinear plasticity rule, without any fine-tuning, gives a stable, unimodal synaptic weight distribution characterized by many strong synapses which remain stable over long periods of time, consistent with electrophysiological and behavioral studies. Moreover, our theory explains how memories encoded by strong synapses can be preferentially stabilized by the plasticity rule. We confirmed our analytical results in a spiking recurrent network. Interestingly, although most synapses are weak and undergo rapid turnover, the fraction of strong synapses are sufficient for supporting realistic spiking dynamics and serve to maintain the network’s cluster structure. Our results provide a mechanistic understanding of how stable memories may emerge on the behavioral level from an STDP rule measured in physiological conditions. Furthermore, the plasticity rule we investigate is mathematically equivalent to other learning rules which rely on the statistics of coincidences, so we expect that our formalism will be useful to study other learning processes beyond the calcium-based plasticity rule.

How psychedelic drugs change the activity of cortical neuronal populations is not well understood. It is also not clear which changes are specific to transition into the psychedelic brain state and which are shared with other brain state transitions. Here, we used Neuropixels probes to record from large populations of neurons in prefrontal cortex of mice given the psychedelic drug TCB-2. The primary effect of drug ingestion was stretching of the distribution of log firing rates of the recorded population. This phenomenon was previously observed across transitions between sleep and wakefulness, which prompted us to examine how common it is. We found that modulation of the width of the log-rate distribution of a neuronal population occurred in multiple areas of the cortex and in the hippocampus even in awake drug-free mice, driven by intrinsic fluctuations in their arousal level. Arousal, however, did not explain the stretching of the log-rate distribution by TCB-2. In both psychedelic and intrinsically occurring brain state transitions, the stretching or squeezing of the log-rate distribution of an entire neuronal population were the result of a more close overlap between log-rate distributions of the upregulated and downregulated subpopulations in one brain state compared to the other brain state. Often, we also observed that the log-rate distribution of the downregulated subpopulation was stretched, whereas the log-rate distribution of the upregulated subpopulation was squeezed. In both subpopulations, the stretching and squeezing were a signature of a greater relative impact of the brain state transition on the rates of the slow-firing neurons. These findings reveal a generic pattern of reorganisation of neuronal firing rates by different kinds of brain state transitions.

Restoring hand function is a top priority for individuals with tetraplegia. This challenge motivates considerable research on brain-computer interfaces (BCIs), which bypass damaged neural pathways to control paralyzed or prosthetic limbs. Here, we demonstrate the BCI control of a prosthetic hand using intracortical recordings from the posterior parietal cortex (PPC). As part of an ongoing clinical trial, two participants with cervical spinal cord injury were each implanted with a 96-channel array in the left PPC. Across four sessions each, we recorded neural activity while they attempted to press individual fingers of the contralateral (right) hand. Single neurons modulated selectively for different finger movements. Offline, we accurately classified finger movements from neural firing rates using linear discriminant analysis (LDA) with cross-validation (accuracy = 90%; chance = 17%). Finally, the participants used the neural classifier online to control all five fingers of a BCI hand. Online control accuracy (86%; chance = 17%) exceeded previous state-of-the-art finger BCIs. Furthermore, offline, we could classify both flexion and extension of the right fingers, as well as flexion of all ten fingers. Our results indicate that neural recordings from PPC can be used to control prosthetic fingers, which may help contribute to a hand restoration strategy for people with tetraplegia.

A recent approach to understanding the mammalian visual system is to show correspondence between the sequential stages of processing in the ventral stream with layers in a deep convolutional neural network (DCNN), providing evidence that visual information is processed hierarchically, with successive stages containing ever higher-level information. However, correspondence is usually defined as shared variance between brain region and model layer. We propose that task-relevant variance is a stricter test: If a DCNN layer corresponds to a brain region, then substituting the model’s activity with brain activity should successfully drive the model’s object recognition decision. Using this approach on three datasets (human fMRI and macaque neuron firing rates) we found that in contrast to the hierarchical view, all ventral stream regions corresponded best to later model layers. That is, all regions contain high-level information about object category. We hypothesised that this is due to recurrent connections propagating high-level visual information from later regions back to early regions, in contrast to the exclusively feed-forward connectivity of DCNNs. Using task-relevant correspondence with a late DCNN layer akin to a tracer, we used Granger causal modelling to show late-DCNN correspondence in IT drives correspondence in V4. Our analysis suggests, effectively, that no ventral stream region can be appropriately characterised as ‘early’ beyond 70ms after stimulus presentation, challenging hierarchical models. More broadly, we ask what it means for a model component and brain region to correspond: beyond quantifying shared variance, we must consider the functional role in the computation. We also demonstrate that using a DCNN to decode high-level conceptual information from ventral stream produces a general mapping from brain to model activation space, which generalises to novel classes held-out from training data. This suggests future possibilities for brain-machine interface with high-level conceptual information, beyond current designs that interface with the sensorimotor periphery.

Depth estimation is an important computer vision task, useful in particular for navigation in autonomous vehicles, or for object manipulation in robotics. Here we solved it using an end-to-end neuromorphic approach, combining two event-based cameras and a Spiking Neural Network (SNN) with a slightly modified U-Net-like encoder-decoder architecture, that we named StereoSpike. More specifically, we used the Multi Vehicle Stereo Event Camera Dataset (MVSEC). It provides a depth ground-truth, which was used to train StereoSpike in a supervised manner, using surrogate gradient descent. We propose a novel readout paradigm to obtain a dense analog prediction –the depth of each pixel– from the spikes of the decoder. We demonstrate that this architecture generalizes very well, even better than its non-spiking counterparts, leading to state-of-the-art test accuracy. To the best of our knowledge, it is the first time that such a large-scale regression problem is solved by a fully spiking network. Finally, we show that low firing rates (<10%) can be obtained via regularization, with a minimal cost in accuracy. This means that StereoSpike could be implemented efficiently on neuromorphic chips, opening the door for low power real time embedded systems.

The mammalian retina is considered an autonomous neuronal tissue, yet there is evidence that it receives inputs from the brain in the form of retinopetal axons. A sub-population of these axons was suggested to belong to histaminergic neurons located in the tuberomammillarynucleus (TMN) of the hypothalamus. Using viral injections to the TMN, we identified these retinopetal axons and found that although few in number, they extensively branch to cover a large portion of the retina. Using Ca2+ imaging and electrophysiology, we show that histamine application increases spontaneous firing rates and alters the light responses of a significant portion of retinal ganglion cells (RGCs). Direct activation of the histaminergic axons also induced significant changes in RGCs activity. Since activity in the TMN was shown to correlate with arousal state, our data suggest the retinal code may change with the animal's behavioral state through the release of histamine from TMN histaminergic neurons.

Ascending neuromodulatory projections from the locus coeruleus (LC) affect cortical neural networks via the release of norepinephrine (NE). However, the exact nature of these neuromodulatory effects on neural activity patterns in vivo is not well understood. Here we show that in awake monkeys, LC activation is associated with changes in coordinated activity patterns in the anterior cingulate cortex (ACC). These relationships, which are largely independent of changes in firing rates of individual ACC neurons, depend on the type of LC activation: ACC pairwise correlations tend to be reduced when tonic (baseline) LC activity increases but are enhanced when external events drive phasic LC responses. Both relationships covary with pupil changes that reflect LC activation and arousal. These results suggest that modulations of information processing that reflect changes in coordinated activity patterns in cortical networks can result partly from ongoing, context-dependent, arousal-related changes in activation of the LC-NE system.

The typical goal of homeostatic mechanisms is to ensure a system operates at or in the vicinity of a stable set point, where a particular measure is relatively constant and stable. Neural firing rate homeostasis is unusual in that a set point of fixed firing rate is at odds with the goal of a neuron to convey information, or produce timed motor responses, which require temporal variations in firing rate. Therefore, for a neuron, a range of firing rates is required for optimal function, which could, for example, be set by a dual system that controls both mean and variance of firing rate. We explore, both via simulations and analysis, how two experimentally measured mechanisms for firing rate homeostasis can cooperate to improve information processing and avoid the pitfall of pulling in different directions when their set points do not appear to match.

Odor memories are exceptionally robust and essential for the survival of many species. In rodents, the olfactory cortex shows features of an autoassociative memory network and plays a key role in the retrieval of olfactory memories (Meissner-Bernard et al., 2019). Interestingly, the telencephalic area Dp, the zebrafish homolog of olfactory cortex, transiently enters a state of precise balance during the presentation of an odor (Rupprecht and Friedrich, 2018). This state is characterized by large synaptic conductances (relative to the resting conductance) and by co-tuning of excitation and inhibition in odor space and in time at the level of individual neurons. Our aim is to understand how this precise synaptic balance affects memory function. For this purpose, we build a simplified, yet biologically plausible spiking neural network model of Dp using experimental observations as constraints: besides precise balance, key features of Dp dynamics include low firing rates, odor-specific population activity and a dominance of recurrent inputs from Dp neurons relative to afferent inputs from neurons in the olfactory bulb. To achieve co-tuning of excitation and inhibition, we introduce structured connectivity by increasing connection probabilities and/or strength among ensembles of excitatory and inhibitory neurons. These ensembles are therefore structural memories of activity patterns representing specific odors. They form functional inhibitory-stabilized subnetworks, as identified by the “paradoxical effect” signature (Tsodyks et al., 1997): inhibition of inhibitory “memory” neurons leads to an increase of their activity. We investigate the benefits of co-tuning for olfactory and memory processing, by comparing inhibitory-stabilized networks with and without co-tuning. We find that co-tuned excitation and inhibition improves robustness to noise, pattern completion and pattern separation. In other words, retrieval of stored information from partial or degraded sensory inputs is enhanced, which is relevant in light of the instability of the olfactory environment. Furthermore, in co-tuned networks, odor-evoked activation of stored patterns does not persist after removal of the stimulus and may therefore subserve fast pattern classification. These findings provide valuable insights into the computations performed by the olfactory cortex, and into general effects of balanced state dynamics in associative memory networks.

Neural circuit functions are stabilized by homeostatic mechanisms at long timescales in response to changes in experience and learning. However, we still do not know which specific physiological variables are being stabilized, nor which cellular or neural-network components comprise the homeostatic machinery. At this point, most evidence suggests that the distribution of firing rates amongst neurons in a brain circuit is the key variable that is maintained around a circuit-specific set-point value in a process called firing rate homeostasis. Here, I will discuss our recent findings that implicate mitochondria as a central player in mediating firing rate homeostasis and its impairments. While mitochondria are known to regulate neuronal variables such as synaptic vesicle release or intracellular calcium concentration, we searched for the mitochondrial signaling pathways that are essential for homeostatic regulation of firing rates. We utilize basic concepts of control theory to build a framework for classifying possible components of the homeostatic machinery in neural networks. This framework may facilitate the identification of new homeostatic pathways whose malfunctions drive instability of neural circuits in distinct brain disorders.

Neural codes, such as temporal codes (precisely timed spikes) and rate codes (instantaneous spike firing rates), are believed to be used in encoding sensory information into spike trains of cortical neurons. Temporal and rate codes co-exist in the spike train and such multiplexed neural code-carrying spike trains have been shown to be spatially synchronized in multiple neurons across different cortical layers during sensory information processing. Inhibition is suggested to promote such synchronization, but it is unclear whether distinct subtypes of interneurons make different contributions in the synchronization of multiplexed neural codes. To test this, in vivo single-unit recordings from barrel cortex were combined with optogenetic manipulations to determine the contributions of parvalbumin (PV)- and somatostatin (SST)-positive interneurons to synchronization of precisely timed spike sequences. We found that PV interneurons preferentially promote the synchronization of spike times when instantaneous firing rates are low (<12 Hz), whereas SST interneurons preferentially promote the synchronization of spike times when instantaneous firing rates are high (>12 Hz). Furthermore, using a computational model, we demonstrate that these effects can be explained by PV and SST interneurons having preferential contribution to feedforward and feedback inhibition, respectively. Overall, these results show that PV and SST interneurons have distinct frequency (rate code)-selective roles in dynamically gating the synchronization of spike times (temporal code) through preferentially recruiting feedforward and feedback inhibitory circuit motifs. The inhibitory neural circuit mechanisms we uncovered here his may have critical roles in regulating neural code-based somatosensory information processing in the neocortex.

Neural integrators are circuits that are able to code analog information such as spatial location or amplitude. Storing amplitude requires the network to have a large number of attractors. In classic models with recurrent excitation, such networks require very careful tuning to behave as integrators and are not robust to small mistuning of the recurrent weights. In this talk, I introduce a circuit with recurrent connectivity that is subjected to a slow subthreshold oscillation (such as the theta rhythm in the hippocampus). I show that such a network can robustly maintain many discrete attracting states. Furthermore, the firing rates of the neurons in these attracting states are much closer to those seen in recordings of animals. I show the mechanism for this can be explained by the instability regions of the Mathieu equation. I then extend the model in various ways and, for example, show that in a spatially distributed network, it is possible to code location and amplitude simultaneously. I show that the resulting mean field equations are equivalent to a certain discontinuous differential equation.

Neocortical networks must generate and maintain stable activity patterns despite perturbations induced by learning and experience- dependent plasticity. There is abundant theoretical and experimental evidence that network stability is achieved through homeostatic plasticity mechanisms that adjust synaptic and neuronal properties to stabilize some measure of average activity, and this process has been extensively studied in primary visual cortex (V1), where chronic visual deprivation induces an initial drop in activity and ensemble average firing rates (FRs), but over time activity is restored to baseline despite continued deprivation. Here I discuss recent work from the lab in which we followed this FR homeostasis in individual V1 neurons in freely behaving animals during a prolonged visual deprivation/eye-reopening paradigm. We find that - when FRs are perturbed by manipulating sensory experience - over time they return precisely to a cell-autonomous set-point. Finally, we find that homeostatic plasticity is perturbed in a mouse model of Autism spectrum disorder, and this results in a breakdown of FRH within V1. These data suggest that loss of homeostatic plasticity is one primary cause of excitation/inhibition imbalances in ASD models. Together these studies illuminate the role of stabilizing plasticity mechanisms in the ability of neocortical circuits to recover robust function following challenges to their excitability.

In the olfactory system, cholinergic modulation has been associated with contrast modulation and changes in receptive fields in the olfactory bulb, as well the learning of odor associations in the olfactory cortex. Computational modeling and behavioral studies suggest that cholinergic modulation could improve sensory processing and learning while preventing pro-active interference when task demands are high. However, how sensory inputs and/or learning regulate incoming modulation has not yet been elucidated. We here use a computational model of the olfactory bulb, piriform cortex (PC) and horizontal limb of the diagonal band of Broca (HDB) to explore how olfactory learning could regulate cholinergic inputs to the system in a closed feedback loop. In our model, the novelty of an odor is reflected in firing rates and sparseness of cortical neurons in response to that odor and these firing rates can directly regulate learning in the system by modifying cholinergic inputs to the system.