Strong inhibitory signaling underlies stable temporal dynamics and working memory in spiking neural networks

Timescales of learning in prefrontal cortex

The lateral prefrontal cortex (PFC) in humans and other primates is critical for immediate, goal-directed behaviour and working memory, which are classically considered distinct from the cognitive and neural circuits that support long-term learning and memory. Over the past few years, a reconsideration of this textbook perspective has emerged, in that different timescales of memory-guided behaviour are in constant interaction during the pursuit of immediate goals. Here, we will first detail how neural activity related to the shortest timescales of goal-directed behaviour (which requires maintenance of current states and goals in working memory) is sculpted by long-term knowledge and learning - that is, how the past informs present behaviour. Then, we will outline how learning across different timescales (from seconds to years) drives plasticity in the primate lateral PFC, from single neuron firing rates to mesoscale neuroimaging activity patterns. Finally, we will review how, over days and months of learning, dense local and long-range connectivity patterns in PFC facilitate longer-lasting changes in population activity by changing synaptic weights and recruiting additional neural resources to inform future behaviour. Our Review sheds light on how the machinery of plasticity in PFC circuits facilitates the integration of learned experiences across time to best guide adaptive behaviour.

Slow ramping emerges from spontaneous fluctuations in spiking neural networks

The capacity to initiate actions endogenously is critical for goal-directed behavior. Spontaneous voluntary actions are typically preceded by slow-ramping activity in medial frontal cortex that begins around two seconds before movement, which may reflect spontaneous fluctuations that influence action timing. However, the mechanisms by which these slow ramping signals emerge from single-neuron and network dynamics remain poorly understood. Here, we developed a spiking neural-network model that produces spontaneous slow ramping activity in single neurons and population activity with onsets ~2 s before threshold crossings. A key prediction of our model is that neurons that ramp together have correlated firing patterns before ramping onset. We confirmed this model-derived hypothesis in a dataset of human single neuron recordings from medial frontal cortex. Our results suggest that slow ramping signals reflect bounded spontaneous fluctuations that emerge from quasi-winner-take-all dynamics in clustered networks that are temporally stabilized by slow-acting synapses.

Biological computations: Limitations of attractor-based formalisms and the need for transients

Living systems, from single cells to higher vertebrates, receive a continuous stream of non-stationary inputs that they sense, for e.g. via cell surface receptors or sensory organs. By integrating these time-varying, multi-sensory, and often noisy information with memory using complex molecular or neuronal networks, they generate a variety of responses beyond simple stimulus-response association, including avoidance behavior, life-long-learning or social interactions. In a broad sense, these processes can be understood as a type of biological computation. Taking as a basis generic features of biological computations, such as real-time responsiveness or robustness and flexibility of the computation, we highlight the limitations of the current attractor-based framework for understanding computations in biological systems. We argue that frameworks based on transient dynamics away from attractors are better suited for the description of computations performed by neuronal and signaling networks. In particular, we discuss how quasi-stable transient dynamics from ghost states that emerge at criticality have a promising potential for developing an integrated framework of computations, that can help us understand how living system actively process information and learn from their continuously changing environment.

Top-down modulation in canonical cortical circuits with short-term plasticity

Cortical dynamics and computations are strongly influenced by diverse GABAergic interneurons, including those expressing parvalbumin (PV), somatostatin (SST), and vasoactive intestinal peptide (VIP). Together with excitatory (E) neurons, they form a canonical microcircuit and exhibit counterintuitive nonlinear phenomena. One instance of such phenomena is response reversal, whereby SST neurons show opposite responses to top-down modulation via VIP depending on the presence of bottom-up sensory input, indicating that the network may function in different regimes under different stimulation conditions. Combining analytical and computational approaches, we demonstrate that model networks with multiple interneuron subtypes and experimentally identified short-term plasticity mechanisms can implement response reversal. Surprisingly, despite not directly affecting SST and VIP activity, PV-to-E short-term depression has a decisive impact on SST response reversal. We show how response reversal relates to inhibition stabilization and the paradoxical effect in the presence of several short-term plasticity mechanisms demonstrating that response reversal coincides with a change in the indispensability of SST for network stabilization. In summary, our work suggests a role of short-term plasticity mechanisms in generating nonlinear phenomena in networks with multiple interneuron subtypes and makes several experimentally testable predictions.

Highly biomimetic spiking neuron using SiGe heterojunction bipolar transistors for energy-efficient neuromorphic systems

We demonstrate a highly biomimetic spiking neuron capable of fast and energy-efficient neuronal oscillation dynamics. Our simple neuron circuit is constructed using silicon-germanium heterojunction based bipolar transistors (HBTs) with nanowire structure. The HBT has a hysteresis window with steep switching characteristics and high current margin in the low voltage range, which enables a high spiking frequency (~ 245 kHz) with low energy consumption (≤ 1.37 pJ/spike). Also, gated structure achieves a stable balance in the activity of the neural system by incorporating both excitatory and inhibitory signal. Furthermore, inhibition of multiple strengths can be realized by adjusting the integration time according to the amplitude of the inhibitory signal. In addition, the spiking frequency can be tuned by mutually controlling the hysteresis window in the HBTs. These results ensure the sparse activity and homeostasis of neural networks.

Slow ramping emerges from spontaneous fluctuations in spiking neural networks

1. We reveal a mechanism for slow-ramping signals before spontaneous voluntary movements. 2. Slow synapses stabilize spontaneous fluctuations in spiking neural network. 3. We validate model predictions in human frontal cortical single-neuron recordings. 4. The model recreates the readiness potential in an EEG proxy signal. 5. Neurons that ramp together had correlated activity before ramping onset. The capacity to initiate actions endogenously is critical for goal-directed behavior. Spontaneous voluntary actions are typically preceded by slow-ramping activity in medial frontal cortex that begins around two seconds before movement, which may reflect spontaneous fluctuations that influence action timing. However, the mechanisms by which these slow ramping signals emerge from single-neuron and network dynamics remain poorly understood. Here, we developed a spiking neural-network model that produces spontaneous slow ramping activity in single neurons and population activity with onsets ∼2 seconds before threshold crossings. A key prediction of our model is that neurons that ramp together have correlated firing patterns before ramping onset. We confirmed this model-derived hypothesis in a dataset of human single neuron recordings from medial frontal cortex. Our results suggest that slow ramping signals reflect bounded spontaneous fluctuations that emerge from quasi-winner-take-all dynamics in clustered networks that are temporally stabilized by slow-acting synapses.

Intracortical recordings reveal vision-to-action cortical gradients driving human exogenous attention

Exogenous attention, the process that makes external salient stimuli pop-out of a visual scene, is essential for survival. How attention-capturing events modulate human brain processing remains unclear. Here we show how the psychological construct of exogenous attention gradually emerges over large-scale gradients in the human cortex, by analyzing activity from 1,403 intracortical contacts implanted in 28 individuals, while they performed an exogenous attention task. The timing, location and task-relevance of attentional events defined a spatiotemporal gradient of three neural clusters, which mapped onto cortical gradients and presented a hierarchy of timescales. Visual attributes modulated neural activity at one end of the gradient, while at the other end it reflected the upcoming response timing, with attentional effects occurring at the intersection of visual and response signals. These findings challenge multi-step models of attention, and suggest that frontoparietal networks, which process sequential stimuli as separate events sharing the same location, drive exogenous attention phenomena such as inhibition of return.

Synaptic wiring motifs in posterior parietal cortex support decision-making

The posterior parietal cortex exhibits choice-selective activity during perceptual decision-making tasks1-10. However, it is not known how this selective activity arises from the underlying synaptic connectivity. Here we combined virtual-reality behaviour, two-photon calcium imaging, high-throughput electron microscopy and circuit modelling to analyse how synaptic connectivity between neurons in the posterior parietal cortex relates to their selective activity. We found that excitatory pyramidal neurons preferentially target inhibitory interneurons with the same selectivity. In turn, inhibitory interneurons preferentially target pyramidal neurons with opposite selectivity, forming an opponent inhibition motif. This motif was present even between neurons with activity peaks in different task epochs. We developed neural-circuit models of the computations performed by these motifs, and found that opponent inhibition between neural populations with opposite selectivity amplifies selective inputs, thereby improving the encoding of trial-type information. The models also predict that opponent inhibition between neurons with activity peaks in different task epochs contributes to creating choice-specific sequential activity. These results provide evidence for how synaptic connectivity in cortical circuits supports a learned decision-making task.

In Vivo Characterization of Intracortical Probes with Focused Ion Beam-Etched Nanopatterned Topographies

(1) Background: Intracortical microelectrodes (IMEs) are an important part of interfacing with the central nervous system (CNS) and recording neural signals. However, recording electrodes have shown a characteristic steady decline in recording performance owing to chronic neuroinflammation. The topography of implanted devices has been explored to mimic the nanoscale three-dimensional architecture of the extracellular matrix. Our previous work used histology to study the implant sites of non-recording probes and showed that a nanoscale topography at the probe surface mitigated the neuroinflammatory response compared to probes with smooth surfaces. Here, we hypothesized that the improvement in the neuroinflammatory response for probes with nanoscale surface topography would extend to improved recording performance. (2) Methods: A novel design modification was implemented on planar silicon-based neural probes by etching nanopatterned grooves (with a 500 nm pitch) into the probe shank. To assess the hypothesis, two groups of rats were implanted with either nanopatterned (n = 6) or smooth control (n = 6) probes, and their recording performance was evaluated over 4 weeks. Postmortem gene expression analysis was performed to compare the neuroinflammatory response from the two groups. (3) Results: Nanopatterned probes demonstrated an increased impedance and noise floor compared to controls. However, the recording performances of the nanopatterned and smooth probes were similar, with active electrode yields for control probes and nanopatterned probes being approximately 50% and 45%, respectively, by 4 weeks post-implantation. Gene expression analysis showed one gene, Sirt1, differentially expressed out of 152 in the panel. (4) Conclusions: this study provides a foundation for investigating novel nanoscale topographies on neural probes.

Neuromuscular Electrical Stimulation Does Not Influence Spinal Excitability in Multiple Sclerosis Patients

(1) Background: Neuromuscular electrical stimulation (NMES) has beneficial effects on physical functions in Multiple sclerosis (MS) patients. However, the neurophysiological mechanisms underlying these functional improvements are still unclear. This study aims at comparing acute responses in spinal excitability, as measured by soleus Hoffmann reflex (H-reflex), between MS patients and healthy individuals, under three experimental conditions involving the ankle planta flexor muscles: (1) passive NMES (pNMES); (2) NMES superimposed onto isometric voluntary contraction (NMES+); and (3) isometric voluntary contraction (ISO). (2) Methods: In total, 20 MS patients (MS) and 20 healthy individuals as the control group (CG) took part in a single experimental session. Under each condition, participants performed 15 repetitions of 6 s at 20% of maximal voluntary isometric contraction, with 6 s of recovery between repetitions. Before and after each condition, H-reflex amplitudes were recorded. (3) Results: In MS, H-reflex amplitude did not change under any experimental condition (ISO: p = 0.506; pNMES: p = 0.068; NMES+: p = 0.126). In CG, H-reflex amplitude significantly increased under NMES+ (p = 0.01), decreased under pNMES (p < 0.000) and was unaltered under ISO (p = 0.829). (4) Conclusions: The different H-reflex responses between MS and CG might reflect a reduced ability of MS patients in modulating spinal excitability.

Training-Dependent Gradients of Timescales of Neural Dynamics in the Primate Prefrontal Cortex and Their Contributions to Working Memory

Cortical neurons exhibit multiple timescales related to dynamics of spontaneous fluctuations (intrinsic timescales) and response to task events (seasonal timescales) in addition to selectivity to task-relevant signals. These timescales increase systematically across the cortical hierarchy, for example, from parietal to prefrontal and cingulate cortex, pointing to their role in cortical computations. It is currently unknown whether these timescales are inherent properties of neurons and/or depend on training in a specific task and if the latter, how their modulations contribute to task performance. To address these questions, we analyzed single-cell recordings within five subregions of the prefrontal cortex (PFC) of male macaques before and after training on a working-memory task. We found fine-grained but opposite gradients of intrinsic and seasonal timescales that mainly appeared after training. Intrinsic timescales decreased whereas seasonal timescales increased from posterior to anterior subregions within both dorsal and ventral PFC. Moreover, training was accompanied by increases in proportions of neurons that exhibited intrinsic and seasonal timescales. These effects were comparable to the emergence of response selectivity due to training. Finally, task selectivity accompanied opposite neural dynamics such that neurons with task-relevant selectivity exhibited longer intrinsic and shorter seasonal timescales. Notably, neurons with longer intrinsic and shorter seasonal timescales exhibited superior population-level coding, but these advantages extended to the delay period mainly after training. Together, our results provide evidence for plastic, fine-grained gradients of timescales within PFC that can influence both single-cell and population coding, pointing to the importance of these timescales in understanding cognition.

The neuronal implementation of representational geometry in primate prefrontal cortex

Modern neuroscience has seen the rise of a population-doctrine that represents cognitive variables using geometrical structures in activity space. Representational geometry does not, however, account for how individual neurons implement these representations. Leveraging the principle of sparse coding, we present a framework to dissect representational geometry into biologically interpretable components that retain links to single neurons. Applied to extracellular recordings from the primate prefrontal cortex in a working memory task with interference, the identified components revealed disentangled and sequential memory representations including the recovery of memory content after distraction, signals hidden to conventional analyses. Each component was contributed by small subpopulations of neurons with distinct spiking properties and response dynamics. Modeling showed that such sparse implementations are supported by recurrently connected circuits as in prefrontal cortex. The perspective of neuronal implementation links representational geometries to their cellular constituents, providing mechanistic insights into how neural systems encode and process information.

Decoding working memory information from neurons with and without persistent activity in the primate prefrontal cortex

Persistent activity of neurons in the prefrontal cortex has been thought to represent the information maintained in working memory, though alternative models have challenged this idea. Theories that depend on the dynamic representation of information posit that stimulus information may be maintained by the activity pattern of neurons whose firing rate is not significantly elevated above their baseline during the delay period of working memory tasks. We thus tested the ability of neurons that do and do not generate persistent activity in the prefrontal cortex of monkeys to represent spatial and object information in working memory. Neurons that generated persistent activity represented more information about the stimuli in both spatial and object working memory tasks. The amount of information that could be decoded from neural activity depended on the choice of decoder and parameters used but neurons with persistent activity outperformed non-persistent neurons consistently. Averaged across all neurons and stimuli, the firing rate did not appear clearly elevated above baseline during the maintenance of neural activity particularly for object working memory; however, this grand average masked neurons that generated persistent activity selective for their preferred stimuli, which carried the majority of stimulus information. These results reveal that prefrontal neurons that generate persistent activity maintain information more reliably during working memory.NEW & NOTEWORTHY Competing theories suggest that neurons that generate persistent activity or do not are primarily responsible for the maintenance of information, particularly regarding object working memory. Although the two models have been debated on theoretical terms, direct comparison of empirical results has been lacking. Analysis of neural activity in a large database of prefrontal recordings revealed that neurons that generate persistent activity were primarily responsible for the maintenance of both spatial and object working memory.

Training-dependent gradients of timescales of neural dynamics in the primate prefrontal cortex and their contributions to working memory

Cortical neurons exhibit multiple timescales related to dynamics of spontaneous fluctuations (intrinsic timescales) and response to task events (seasonal timescales) in addition to selectivity to task-relevant signals. These timescales increase systematically across the cortical hierarchy, e.g., from parietal to prefrontal and cingulate cortex, pointing to their role in cortical computations. It is currently unknown whether these timescales depend on training in a specific task and/or are an inherent property of neurons, and whether more fine-grained hierarchies of timescales exist within specific cortical regions. To address these questions, we analyzed single-cell recordings within five subregions of the prefrontal cortex (PFC) of male macaques before and after training on a working-memory task. We found fine-grained but opposite gradients of intrinsic and seasonal timescales that mainly appeared after training. Intrinsic timescales decreased whereas seasonal timescales increased from posterior to anterior subregions within both dorsal and ventral PFC. Moreover, training was accompanied by increases in proportions of neurons that exhibited intrinsic and seasonal timescales. These effects were comparable to the emergence of response selectivity due to training. Finally, task selectivity accompanied opposite neural dynamics such that neurons with task-relevant selectivity exhibited longer intrinsic and shorter seasonal timescales. Notably, neurons with longer intrinsic and shorter seasonal timescales exhibited superior population-level coding, but these advantages extended to the delay period mainly after training. Together, our results provide evidence for plastic, fine-grained gradients of timescales within PFC that can influence both single-cell and population coding, pointing to the importance of these timescales in understanding cognition.

Significance statement

Recent studies have demonstrated that neural responses exhibit dynamics with different timescales that follow a certain order or hierarchy across cortical areas. While the hierarchy of timescales is consistent across different tasks, it is unknown if these timescales emerge only after training or if they represent inherent properties of neurons. To answer this question, we estimated multiple timescales in neural response across five subregions of the monkeys' lateral prefrontal cortex before and after training on a working-memory task. Our results provide evidence for fine-grained gradients related to certain neural dynamics. Moreover, we show that these timescales depend on and can be modulated by training in a cognitive task, and contribute to encoding of task-relevant information at single-cell and population levels.

Predictive Mouse Ultrasonic Vocalization Sequences: Uncovering Behavioral Significance, Auditory Cortex Neuronal Preferences, and Social-Experience-Driven Plasticity

Mouse ultrasonic vocalizations (USVs) contain predictable sequential structures like bird songs and speech. Neural representation of USVs in the mouse primary auditory cortex (Au1) and its plasticity with experience has been largely studied with single-syllables or dyads, without using the predictability in USV sequences. Studies using playback of USV sequences have used randomly selected sequences from numerous possibilities. The current study uses mutual information to obtain context-specific natural sequences (NSeqs) of USV syllables capturing the observed predictability in male USVs in different contexts of social interaction with females. Behavioral and physiological significance of NSeqs over random sequences (RSeqs) lacking predictability were examined. Female mice, never having the social experience of being exposed to males, showed higher selectivity for NSeqs behaviorally and at cellular levels probed by expression of immediate early gene c-fos in Au1. The Au1 supragranular single units also showed higher selectivity to NSeqs over RSeqs. Social-experience-driven plasticity in encoding NSeqs and RSeqs in adult females was probed by examining neural selectivities to the same sequences before and after the above social experience. Single units showed enhanced selectivity for NSeqs over RSeqs after the social experience. Further, using two-photon Ca2+ imaging, we observed social experience-dependent changes in the selectivity of sequences of excitatory and somatostatin-positive inhibitory neurons but not parvalbumin-positive inhibitory neurons of Au1. Using optogenetics, somatostatin-positive neurons were identified as a possible mediator of the observed social-experience-driven plasticity. Our study uncovers the importance of predictive sequences and introduces mouse USVs as a promising model to study context-dependent speech like communications.SIGNIFICANCE STATEMENT Humans need to detect patterns in the sensory world. For instance, speech is meaningful sequences of acoustic tokens easily differentiated from random ordered tokens. The structure derives from the predictability of the tokens. Similarly, mouse vocalization sequences have predictability and undergo context-dependent modulation. Our work investigated whether mice differentiate such informative predictable sequences (NSeqs) of communicative significance from RSeqs at the behavioral, molecular, and neuronal levels. Following a social experience in which NSeqs occur as a crucial component, mouse auditory cortical neurons become more sensitive to differences between NSeqs and RSeqs, although preference for individual tokens is unchanged. Thus, speech-like communication and its dysfunction may be studied in circuit, cellular, and molecular levels in mice.

Neuronal avalanches: Sandpiles of self-organized criticality or critical dynamics of brain waves?

Analytical expressions for scaling of brain wave spectra derived from the general nonlinear wave Hamiltonian form show excellent agreement with experimental "neuronal avalanche" data. The theory of the weakly evanescent nonlinear brain wave dynamics [Phys. Rev. Research 2, 023061 (2020); J. Cognitive Neurosci. 32, 2178 (2020)] reveals the underlying collective processes hidden behind the phenomenological statistical description of the neuronal avalanches and connects together the whole range of brain activity states, from oscillatory wave-like modes, to neuronal avalanches, to incoherent spiking, showing that the neuronal avalanches are just the manifestation of the different nonlinear side of wave processes abundant in cortical tissue. In a more broad way these results show that a system of wave modes interacting through all possible combinations of the third order nonlinear terms described by a general wave Hamiltonian necessarily produces anharmonic wave modes with temporal and spatial scaling properties that follow scale free power laws. To the best of our knowledge this has never been reported in the physical literature and may be applicable to many physical systems that involve wave processes and not just to neuronal avalanches.

Decoding working memory information from persistent and activity-silent neurons in the primate prefrontal cortex

Persistent activity of neurons in the prefrontal cortex has been thought to represent the information maintained in working memory, though alternative models have recently challenged this idea. Activity-silent theories posit that stimulus information may be maintained by the activity pattern of neurons that do not produce firing rate significantly elevated about their baseline during the delay period of working memory tasks. We thus tested the ability of neurons that do and do not generate persistent activity in the prefrontal cortex of monkeys to represent spatial and object information in working memory. Neurons that generated persistent activity represented more information about the stimuli in both spatial and object working memory tasks. The amount of information that could be decoded from neural activity depended on the choice of decoder and parameters used but neurons with persistent activity outperformed non-persistent neurons consistently. Although averaged across all neurons and stimuli, firing rate did not appear clearly elevated above baseline during the maintenance of neural activity particularly for object working memory, this grant average masked neurons that generated persistent activity selective for their preferred stimuli, which carried the majority of information about the stimulus identity. These results reveal that prefrontal neurons with generate persistent activity constitute the primary mechanism of working memory maintenance in the cortex.

NEW AND NOTEWORTHY

Competing theories suggest that neurons that generate persistent activity or do not are primarily responsible for the maintenance of information, particularly regarding object working memory. While the two models have been debated on theoretical terms, direct comparison of empirical results have been lacking. Analysis of neural activity in a large database of prefrontal recordings revealed that neurons that generate persistent activity were primarily responsible for the maintenance of both spatial and object working memory.

Fronto-parietal networks shape human conscious report through attention gain and reorienting

How do attention and consciousness interact in the human brain? Rival theories of consciousness disagree on the role of fronto-parietal attentional networks in conscious perception. We recorded neural activity from 727 intracerebral contacts in 13 epileptic patients, while they detected near-threshold targets preceded by attentional cues. Clustering revealed three neural patterns: first, attention-enhanced conscious report accompanied sustained right-hemisphere fronto-temporal activity in networks connected by the superior longitudinal fasciculus (SLF) II-III, and late accumulation of activity (>300 ms post-target) in bilateral dorso-prefrontal and right-hemisphere orbitofrontal cortex (SLF I-III). Second, attentional reorienting affected conscious report through early, sustained activity in a right-hemisphere network (SLF III). Third, conscious report accompanied left-hemisphere dorsolateral-prefrontal activity. Task modeling with recurrent neural networks revealed multiple clusters matching the identified brain clusters, elucidating the causal relationship between clusters in conscious perception of near-threshold targets. Thus, distinct, hemisphere-asymmetric fronto-parietal networks support attentional gain and reorienting in shaping human conscious experience.

Spiking Recurrent Neural Networks Represent Task-Relevant Neural Sequences in Rule-Dependent Computation

Background

Prefrontal cortical neurons play essential roles in performing rule-dependent tasks and working memory-based decision making.

Methods

Motivated by PFG recordings of task-performing mice, we developed an excitatory-inhibitory spiking recurrent neural network (SRNN) to perform a rule-dependent two-alternative forced choice (2AFC) task. We imposed several important biological constraints onto the SRNN, and adapted spike frequency adaptation (SFA) and SuperSpike gradient methods to train the SRNN efficiently.

Results

The trained SRNN produced emergent rule-specific tunings in single-unit representations, showing rule-dependent population dynamics that resembled experimentally observed data. Under varying test conditions, we manipulated the SRNN parameters or configuration in computer simulations, and we investigated the impacts of rule-coding error, delay duration, recurrent weight connectivity and sparsity, and excitation/inhibition (E/I) balance on both task performance and neural representations.

Conclusions

Overall, our modeling study provides a computational framework to understand neuronal representations at a fine timescale during working memory and cognitive control, and provides new experimentally testable hypotheses in future experiments.

Multiplexing working memory and time in the trajectories of neural networks

Working memory (WM) and timing are generally considered distinct cognitive functions, but similar neural signatures have been implicated in both. To explore the hypothesis that WM and timing may rely on shared neural mechanisms, we used psychophysical tasks that contained either task-irrelevant timing or WM components. In both cases, the task-irrelevant component influenced performance. We then developed recurrent neural network (RNN) simulations that revealed that cue-specific neural sequences, which multiplexed WM and time, emerged as the dominant regime that captured the behavioural findings. During training, RNN dynamics transitioned from low-dimensional ramps to high-dimensional neural sequences, and depending on task requirements, steady-state or ramping activity was also observed. Analysis of RNN structure revealed that neural sequences relied primarily on inhibitory connections, and could survive the deletion of all excitatory-to-excitatory connections. Our results indicate that in some instances WM is encoded in time-varying neural activity because of the importance of predicting when WM will be used.

Modularity Facilitates Classification Performance of Spiking Neural Networks for Decoding Cortical Spike Trains

After the introduction of recurrence, an important property of the biological brain, spiking neural networks (SNNs) have achieved unprecedented classification performance. But they still cannot outperform many artificial neural networks. Modularity is another crucial feature of the biological brain. It remains unclear if modularity can also improve the performance of SNNs. To investigate this idea, we proposed the modular SNN, and compared its performance with a uniform SNN without modularity by employing them to classify cortical spike trains. For the first time, a significant improvement was found in our modular SNN. Further, we probed into the factors influencing the performance of the modular SNN and found: (a). The modular SNN outperformed the uniform SNN more significantly when the number of neurons in the networks increased; (b). The performance of the modular SNNs increased as the number of modules dropped. These preliminary but novel findings suggest that modularity may help develop better artificial intelligence and brain-machine interfaces. Also, the modular SNN may serve as a model for the study of neuronal spike synchrony.

Spiking neural network with working memory can integrate and rectify spatiotemporal features

In the real world, information is often correlated with each other in the time domain. Whether it can effectively make a decision according to the global information is the key indicator of information processing ability. Due to the discrete characteristics of spike trains and unique temporal dynamics, spiking neural networks (SNNs) show great potential in applications in ultra-low-power platforms and various temporal-related real-life tasks. However, the current SNNs can only focus on the information a short time before the current moment, its sensitivity in the time domain is limited. This problem affects the processing ability of SNN in different kinds of data, including static data and time-variant data, and reduces the application scenarios and scalability of SNN. In this work, we analyze the impact of such information loss and then integrate SNN with working memory inspired by recent neuroscience research. Specifically, we propose Spiking Neural Networks with Working Memory (SNNWM) to handle input spike trains segment by segment. On the one hand, this model can effectively increase SNN's ability to obtain global information. On the other hand, it can effectively reduce the information redundancy between adjacent time steps. Then, we provide simple methods to implement the proposed network architecture from the perspectives of biological plausibility and neuromorphic hardware friendly. Finally, we test the proposed method on static and sequential data sets, and the experimental results show that the proposed model can better process the whole spike train, and achieve state-of-the-art results in short time steps. This work investigates the contribution of introducing biologically inspired mechanisms, e.g., working memory, and multiple delayed synapses to SNNs, and provides a new perspective to design future SNNs.

Distributing task-related neural activity across a cortical network through task-independent connections

Task-related neural activity is widespread across populations of neurons during goal-directed behaviors. However, little is known about the synaptic reorganization and circuit mechanisms that lead to broad activity changes. Here we trained a subset of neurons in a spiking network with strong synaptic interactions to reproduce the activity of neurons in the motor cortex during a decision-making task. Task-related activity, resembling the neural data, emerged across the network, even in the untrained neurons. Analysis of trained networks showed that strong untrained synapses, which were independent of the task and determined the dynamical state of the network, mediated the spread of task-related activity. Optogenetic perturbations suggest that the motor cortex is strongly-coupled, supporting the applicability of the mechanism to cortical networks. Our results reveal a cortical mechanism that facilitates distributed representations of task-variables by spreading the activity from a subset of plastic neurons to the entire network through task-independent strong synapses.

Intrinsic timescales in the visual cortex change with selective attention and reflect spatial connectivity

Intrinsic timescales characterize dynamics of endogenous fluctuations in neural activity. Variation of intrinsic timescales across the neocortex reflects functional specialization of cortical areas, but less is known about how intrinsic timescales change during cognitive tasks. We measured intrinsic timescales of local spiking activity within columns of area V4 in male monkeys performing spatial attention tasks. The ongoing spiking activity unfolded across at least two distinct timescales, fast and slow. The slow timescale increased when monkeys attended to the receptive fields location and correlated with reaction times. By evaluating predictions of several network models, we found that spatiotemporal correlations in V4 activity were best explained by the model in which multiple timescales arise from recurrent interactions shaped by spatially arranged connectivity, and attentional modulation of timescales results from an increase in the efficacy of recurrent interactions. Our results suggest that multiple timescales may arise from the spatial connectivity in the visual cortex and flexibly change with the cognitive state due to dynamic effective interactions between neurons.

Critically synchronized brain waves form an effective, robust and flexible basis for human memory and learning

The effectiveness, robustness, and flexibility of memory and learning constitute the very essence of human natural intelligence, cognition, and consciousness. However, currently accepted views on these subjects have, to date, been put forth without any basis on a true physical theory of how the brain communicates internally via its electrical signals. This lack of a solid theoretical framework has implications not only for our understanding of how the brain works, but also for wide range of computational models developed from the standard orthodox view of brain neuronal organization and brain network derived functioning based on the Hodgkin-Huxley ad-hoc circuit analogies that have produced a multitude of Artificial, Recurrent, Convolution, Spiking, etc., Neural Networks (ARCSe NNs) that have in turn led to the standard algorithms that form the basis of artificial intelligence (AI) and machine learning (ML) methods. Our hypothesis, based upon our recently developed physical model of weakly evanescent brain wave propagation (WETCOW) is that, contrary to the current orthodox model that brain neurons just integrate and fire under accompaniment of slow leaking, they can instead perform much more sophisticated tasks of efficient coherent synchronization/desynchronization guided by the collective influence of propagating nonlinear near critical brain waves, the waves that currently assumed to be nothing but inconsequential subthreshold noise. In this paper we highlight the learning and memory capabilities of our WETCOW framework and then apply it to the specific application of AI/ML and Neural Networks. We demonstrate that the learning inspired by these critically synchronized brain waves is shallow, yet its timing and accuracy outperforms deep ARCSe counterparts on standard test datasets. These results have implications for both our understanding of brain function and for the wide range of AI/ML applications.

Critical brain wave dynamics of neuronal avalanches

Analytical expressions for scaling of brain wave spectra derived from the general non-linear wave Hamiltonian form show excellent agreement with experimental "neuronal avalanche" data. The theory of the weakly evanescent non-linear brain wave dynamics reveals the underlying collective processes hidden behind the phenomenological statistical description of the neuronal avalanches and connects together the whole range of brain activity states, from oscillatory wave-like modes, to neuronal avalanches, to incoherent spiking, showing that the neuronal avalanches are just the manifestation of the different non-linear side of wave processes abundant in cortical tissue. In a more broad way these results show that a system of wave modes interacting through all possible combinations of the third order non-linear terms described by a general wave Hamiltonian necessarily produces anharmonic wave modes with temporal and spatial scaling properties that follow scale free power laws. To the best of our knowledge this has never been reported in the physical literature and may be applicable to many physical systems that involve wave processes and not just to neuronal avalanches.

Choice selective inhibition drives stability and competition in decision circuits

During perceptual decision-making, the firing rates of cortical neurons reflect upcoming choices. Recent work showed that excitatory and inhibitory neurons are equally selective for choice. However, the functional consequences of inhibitory choice selectivity in decision-making circuits are unknown. We developed a circuit model of decision-making which accounts for the specificity of inputs to and outputs from inhibitory neurons. We found that selective inhibition expands the space of circuits supporting decision-making, allowing for weaker or stronger recurrent excitation when connected in a competitive or feedback motif. The specificity of inhibitory outputs sets the trade-off between speed and accuracy of decisions by either stabilizing or destabilizing the saddle-point dynamics underlying decisions in the circuit. Recurrent neural networks trained to make decisions display the same dependence on inhibitory specificity and the strength of recurrent excitation. Our results reveal two concurrent roles for selective inhibition in decision-making circuits: stabilizing strongly connected excitatory populations and maximizing competition between oppositely selective populations.

Cognitive performance in multiple sclerosis: what is the role of the gamma-aminobutyric acid system?

Cognitive impairment occurs in 40-65% of persons with multiple sclerosis and may be related to alterations in glutamatergic and GABAergic neurotransmission. Therefore, the aim of this study was to determine how glutamatergic and GABAergic changes relate to cognitive functioning in multiple sclerosis in vivo. Sixty persons with multiple sclerosis (mean age 45.5 ± 9.6 years, 48 females, 51 relapsing-remitting multiple sclerosis) and 22 age-matched healthy controls (45.6 ± 22.0 years, 17 females) underwent neuropsychological testing and MRI. Persons with multiple sclerosis were classified as cognitively impaired when scoring at least 1.5 standard deviations below normative scores on ≥30% of tests. Glutamate and GABA concentrations were determined in the right hippocampus and bilateral thalamus using magnetic resonance spectroscopy. GABA-receptor density was assessed using quantitative [¹¹C]flumazenil positron emission tomography in a subset of participants. Positron emission tomography outcome measures were the influx rate constant (a measure predominantly reflecting perfusion) and volume of distribution, which is a measure of GABA-receptor density. Twenty persons with multiple sclerosis (33%) fulfilled the criteria for cognitive impairment. No differences were observed in glutamate or GABA concentrations between persons with multiple sclerosis and healthy controls, or between cognitively preserved, impaired and healthy control groups. Twenty-two persons with multiple sclerosis (12 cognitively preserved and 10 impaired) and 10 healthy controls successfully underwent [¹¹C]flumazenil positron emission tomography. Persons with multiple sclerosis showed a lower influx rate constant in the thalamus, indicating lower perfusion. For the volume of distribution, persons with multiple sclerosis showed higher values than controls in deep grey matter, reflecting increased GABA-receptor density. When comparing cognitively impaired and preserved patients to controls, the preserved group showed a significantly higher volume of distribution in cortical and deep grey matter and hippocampus. Positive correlations were observed between both positron emission tomography measures and information processing speed in the multiple sclerosis group only. Whereas concentrations of glutamate and GABA did not differ between multiple sclerosis and control nor between cognitively impaired, preserved and control groups, increased GABA-receptor density was observed in preserved persons with multiple sclerosis that was not seen in cognitively impaired patients. In addition, GABA-receptor density correlated to cognition, in particular with information processing speed. This could indicate that GABA-receptor density is upregulated in the cognitively preserved phase of multiple sclerosis as a means to regulate neurotransmission and potentially preserve cognitive functioning.

Excitatory-inhibitory recurrent dynamics produce robust visual grids and stable attractors

Spatially modulated grid cells have been recently found in the rat secondary visual cortex (V2) during active navigation. However, the computational mechanism and functional significance of V2 grid cells remain unknown. To address the knowledge gap, we train a biologically inspired excitatory-inhibitory recurrent neural network to perform a two-dimensional spatial navigation task with multisensory input. We find grid-like responses in both excitatory and inhibitory RNN units, which are robust with respect to spatial cues, dimensionality of visual input, and activation function. Population responses reveal a low-dimensional, torus-like manifold and attractor. We find a link between functional grid clusters with similar receptive fields and structured excitatory-to-excitatory connections. Additionally, multistable torus-like attractors emerged with increasing sparsity in inter- and intra-subnetwork connectivity. Finally, irregular grid patterns are found in recurrent neural network (RNN) units during a visual sequence recognition task. Together, our results suggest common computational mechanisms of V2 grid cells for spatial and non-spatial tasks.

Regulation of circuit organization and function through inhibitory synaptic plasticity

Diverse inhibitory neurons in the mammalian brain shape circuit connectivity and dynamics through mechanisms of synaptic plasticity. Inhibitory plasticity can establish excitation/inhibition (E/I) balance, control neuronal firing, and affect local calcium concentration, hence regulating neuronal activity at the network, single neuron, and dendritic level. Computational models can synthesize multiple experimental results and provide insight into how inhibitory plasticity controls circuit dynamics and sculpts connectivity by identifying phenomenological learning rules amenable to mathematical analysis. We highlight recent studies on the role of inhibitory plasticity in modulating excitatory plasticity, forming structured networks underlying memory formation and recall, and implementing adaptive phenomena and novelty detection. We conclude with experimental and modeling progress on the role of interneuron-specific plasticity in circuit computation and context-dependent learning.

Effects of exercise interventions on executive function in old adults with mild cognitive impairment: A systematic review and meta-analysis of randomized controlled trials

AIMS

To assess the effect of exercise interventions on subdomains of executive function (EF) in older adults with mild cognitive impairment (MCI).

METHODS

Nine electronic databases were comprehensively searched from their inception to February 2021. Randomized controlled trials examining the effect of exercise training on EF in MCI were included.

RESULTS

Twenty-four eligible articles involving 2278 participants were identified. The results showed that exercise interventions had positive benefits on working memory, switching and inhibition in MCI. Subgroup analysis based on exercise prescriptions revealed that both aerobatic exercise and mind-body exercise had similar positive effect size on working memory. However, only mind-body exercise had significant effect on switching. Exercise training with moderate frequency (3-4 times/week) had larger effect size than low frequency (1-2 times/week) and only moderate frequency had positive benefits on switching. Both short (4-12 weeks), medium (13-24 weeks) and long (more than 24 weeks) exercise duration significantly ameliorate working memory and switching, however with short duration having slight larger effect sizes than medium and long.

CONCLUSION

Exercise significantly improves three subdomains of EF in MCI, especially mind-body exercise. Exercise training sticking to at least 4 weeks with 3-4 times a week tends to have larger effect size.

Extended liquid state machines for speech recognition

A liquid state machine (LSM) is a biologically plausible model of a cortical microcircuit. It exists of a random, sparse reservoir of recurrently connected spiking neurons with fixed synapses and a trainable readout layer. The LSM exhibits low training complexity and enables backpropagation-free learning in a powerful, yet simple computing paradigm. In this work, the liquid state machine is enhanced by a set of bio-inspired extensions to create the extended liquid state machine (ELSM), which is evaluated on a set of speech data sets. Firstly, we ensure excitatory/inhibitory (E/I) balance to enable the LSM to operate in edge-of-chaos regime. Secondly, spike-frequency adaptation (SFA) is introduced in the LSM to improve the memory capabilities. Lastly, neuronal heterogeneity, by means of a differentiation in time constants, is introduced to extract a richer dynamical LSM response. By including E/I balance, SFA, and neuronal heterogeneity, we show that the ELSM consistently improves upon the LSM while retaining the benefits of the straightforward LSM structure and training procedure. The proposed extensions led up to an 5.2% increase in accuracy while decreasing the number of spikes in the ELSM up to 20.2% on benchmark speech data sets. On some benchmarks, the ELSM can even attain similar performances as the current state-of-the-art in spiking neural networks. Furthermore, we illustrate that the ELSM input-liquid and recurrent synaptic weights can be reduced to 4-bit resolution without any significant loss in classification performance. We thus show that the ELSM is a powerful, biologically plausible and hardware-friendly spiking neural network model that can attain near state-of-the-art accuracy on speech recognition benchmarks for spiking neural networks.

A Recurrent Neural Network Model Accounts for Both Timing and Working Memory Components of an Interval Discrimination Task

Interval discrimination is of fundamental importance to many forms of sensory processing, including speech and music. Standard interval discrimination tasks require comparing two intervals separated in time, and thus include both working memory (WM) and timing components. Models of interval discrimination invoke separate circuits for the timing and WM components. Here we examine if, in principle, the same recurrent neural network can implement both. Using human psychophysics, we first explored the role of the WM component by varying the interstimulus delay. Consistent with previous studies, discrimination was significantly worse for a 250 ms delay, compared to 750 and 1500 ms delays, suggesting that the first interval is stably stored in WM for longer delays. We next successfully trained a recurrent neural network (RNN) on the task, demonstrating that the same network can implement both the timing and WM components. Many units in the RNN were tuned to specific intervals during the sensory epoch, and others encoded the first interval during the delay period. Overall, the encoding strategy was consistent with the notion of mixed selectivity. Units generally encoded more interval information during the sensory epoch than in the delay period, reflecting categorical encoding of short versus long in WM, rather than encoding of the specific interval. Our results demonstrate that, in contrast to standard models of interval discrimination that invoke a separate memory module, the same network can, in principle, solve the timing, WM, and comparison components of an interval discrimination task.

Excitatory and inhibitory hippocampal neurons differ in their homeostatic adaptation to chronic M-channel modulation

A large body of studies has investigated bidirectional homeostatic plasticity both in vitro and in vivo using numerous pharmacological manipulations of activity or behavioral paradigms. However, these experiments rarely explored in the same cellular system the bidirectionality of the plasticity and simultaneously on excitatory and inhibitory neurons. M-channels are voltage-gated potassium channels that play a crucial role in regulating neuronal excitability and plasticity. In cultured hippocampal excitatory neurons, we previously showed that chronic exposure to the M-channel blocker XE991 leads to adaptative compensations, thereby triggering at different timescales intrinsic and synaptic homeostatic plasticity. This plastic adaptation barely occurs in hippocampal inhibitory neurons. In this study, we examined whether this homeostatic plasticity induced by M-channel inhibition was bidirectional by investigating the acute and chronic effects of the M-channel opener retigabine on hippocampal neuronal excitability. Acute retigabine exposure decreased excitability in both excitatory and inhibitory neurons. Chronic retigabine treatment triggered in excitatory neurons homeostatic adaptation of the threshold current and spontaneous firing rate at a time scale of 4–24 h. These plastic changes were accompanied by a substantial decrease in the M-current density and by a small, though significant, proximal relocation of Kv7.3-FGF14 segment along the axon initial segment. Thus, bidirectional homeostatic changes were observed in excitatory neurons though not symmetric in kinetics and mechanisms. Contrastingly, in inhibitory neurons, the compensatory changes in intrinsic excitability barely occurred after 48 h, while no homeostatic normalization of the spontaneous firing rate was observed. Our results indicate that excitatory and inhibitory hippocampal neurons differ in their adaptation to chronic alterations in neuronal excitability induced by M-channel bidirectional modulation.

Inhibitory synaptic loss drives network changes in multiple sclerosis: An ex vivo to in silico translational study

Background:

Synaptic and neuronal loss contribute to network dysfunction and disability in multiple sclerosis (MS). However, it is unknown whether excitatory or inhibitory synapses and neurons are more vulnerable and how their losses impact network functioning.

Objective:

To quantify excitatory and inhibitory synapses and neurons and to investigate how synaptic loss affects network functioning through computational modeling.

Methods:

Using immunofluorescent staining and confocal microscopy, densities of glutamatergic and GABAergic synapses and neurons were compared between post-mortem MS and non-neurological control cases. Then, a corticothalamic biophysical model was employed to study how MS-induced excitatory and inhibitory synaptic loss affect network functioning.

Results:

In layer VI of normal-appearing MS cortex, excitatory and inhibitory synaptic densities were significantly lower than controls (reductions up to 14.9%), but demyelinated cortex showed larger losses of inhibitory synapses (29%). In our computational model, reducing inhibitory synapses impacted the network most, leading to a disinhibitory increase in neuronal activity and connectivity.

Conclusion:

In MS, excitatory and inhibitory synaptic losses were observed, predominantly for inhibitory synapses in demyelinated cortex. Inhibitory synaptic loss affected network functioning most, leading to increased neuronal activity and connectivity. As network disinhibition relates to cognitive impairment, inhibitory synaptic loss seems particularly relevant in MS.

The sloppy relationship between neural circuit structure and function

Investigating and describing the relationships between the structure of a circuit and its function has a long tradition in neuroscience. Since neural circuits acquire their structure through sophisticated developmental programmes, and memories and experiences are maintained through synaptic modification, it is to be expected that structure is closely linked to function. Recent findings challenge this hypothesis from three different angles: Function does not strongly constrain circuit parameters, many parameters in neural circuits are irrelevant and contribute little to function, and circuit parameters are unstable and subject to constant random drift. At the same time however, recent work also showed that dynamics in neural circuit activity that is related to function are robust over time and across individuals. Here this apparent contradiction is addressed by considering the properties of neural manifolds that restrict circuit activity to functionally relevant subspaces, and it will be suggested that degenerate, anisotropic and unstable parameter spaces are a closely related to the structure and implementation of functionally relevant neural manifolds. Abstract figure legend What are the relationships between noisy and highly variable microscopic neural circuit variables on the one hand and the generation of behaviour on the other? Here it is proposed that an intermediate level of description exists where this relationship can be understood in terms of low-dimensional dynamics. Recordings of neural activity during unconstrained behaviour and the development of new machine learning methods will help to uncover these links. This article is protected by copyright. All rights reserved.

Connectomic comparison of mouse and human cortex

The human cerebral cortex houses 1,000 times more neurons than the cerebral cortex of a mouse, but the possible differences in synaptic circuits between these species are still poorly understood. We used 3-dimensional electron microscopy of mouse, macaque and human cortical samples to study their cell type composition and synaptic circuit architecture. The 2.5-fold increase in interneurons in humans compared to mouse was compensated by a change in axonal connection probabilities and therefore did not yield a commensurate increase in inhibitory-vs-excitatory synaptic input balance on human pyramidal cells. Rather, increased inhibition created an expanded interneuron-to-interneuron network, driven by an expansion of interneuron-targeting interneuron types and an increase in their synaptic selectivity for interneuron innervation. These constitute key neuronal network alterations in human cortex.

A hybrid biological neural network model for solving problems in cognitive planning

A variety of behaviors, like spatial navigation or bodily motion, can be formulated as graph traversal problems through cognitive maps. We present a neural network model which can solve such tasks and is compatible with a broad range of empirical findings about the mammalian neocortex and hippocampus. The neurons and synaptic connections in the model represent structures that can result from self-organization into a cognitive map via Hebbian learning, i.e. into a graph in which each neuron represents a point of some abstract task-relevant manifold and the recurrent connections encode a distance metric on the manifold. Graph traversal problems are solved by wave-like activation patterns which travel through the recurrent network and guide a localized peak of activity onto a path from some starting position to a target state.

Inhibitory control of frontal metastability sets the temporal signature of cognition

Cortical dynamics are organized over multiple anatomical and temporal scales. The mechanistic origin of the temporal organization and its contribution to cognition remain unknown. Here we demonstrate the cause of this organization by studying a specific temporal signature (time constant and latency) of neural activity. In monkey frontal areas, recorded during flexible decisions, temporal signatures display specific area-dependent ranges, as well as anatomical and cell-type distributions. Moreover, temporal signatures are functionally adapted to behaviorally relevant timescales. Fine-grained biophysical network models, constrained to account for experimentally observed temporal signatures, reveal that after-hyperpolarization potassium and inhibitory GABA-B conductances critically determine areas' specificity. They mechanistically account for temporal signatures by organizing activity into metastable states, with inhibition controlling state stability and transitions. As predicted by models, state durations non-linearly scale with temporal signatures in monkey, matching behavioral timescales. Thus, local inhibitory-controlled metastability constitutes the dynamical core specifying the temporal organization of cognitive functions in frontal areas.

Layer-specific pyramidal neuron properties underlie diverse anterior cingulate cortical motor and limbic networks

The laminar cellular and circuit mechanisms by which the anterior cingulate cortex (ACC) exerts flexible control of motor and affective information for goal-directed behavior have not been elucidated. Using multimodal tract-tracing, in vitro patch-clamp recording and computational approaches in rhesus monkeys (M. mulatta), we provide evidence that specialized motor and affective network dynamics can be conferred by layer-specific biophysical and structural properties of ACC pyramidal neurons targeting two key downstream structures -the dorsal premotor cortex (PMd) and the amygdala (AMY). AMY-targeting neurons exhibited significant laminar differences, with L5 more excitable (higher input resistance and action potential firing rates) than L3 neurons. Between-pathway differences were found within L5, with AMY-targeting neurons exhibiting greater excitability, apical dendritic complexity, spine densities, and diversity of inhibitory inputs than PMd-targeting neurons. Simulations using a pyramidal-interneuron network model predict that these layer- and pathway-specific single-cell differences contribute to distinct network oscillatory dynamics. L5 AMY-targeting networks are more tuned to slow oscillations well-suited for affective and contextual processing timescales, while PMd-targeting networks showed strong beta/gamma synchrony implicated in rapid sensorimotor processing. These findings are fundamental to our broad understanding of how layer-specific cellular and circuit properties can drive diverse laminar activity found in flexible behavior.

SAM: A Unified Self-Adaptive Multicompartmental Spiking Neuron Model for Learning With Working Memory

Working memory is a fundamental feature of biological brains for perception, cognition, and learning. In addition, learning with working memory, which has been show in conventional artificial intelligence systems through recurrent neural networks, is instrumental to advanced cognitive intelligence. However, it is hard to endow a simple neuron model with working memory, and to understand the biological mechanisms that have resulted in such a powerful ability at the neuronal level. This article presents a novel self-adaptive multicompartment spiking neuron model, referred to as SAM, for spike-based learning with working memory. SAM integrates four major biological principles including sparse coding, dendritic non-linearity, intrinsic self-adaptive dynamics, and spike-driven learning. We first describe SAM's design and explore the impacts of critical parameters on its biological dynamics. We then use SAM to build spiking networks to accomplish several different tasks including supervised learning of the MNIST dataset using sequential spatiotemporal encoding, noisy spike pattern classification, sparse coding during pattern classification, spatiotemporal feature detection, meta-learning with working memory applied to a navigation task and the MNIST classification task, and working memory for spatiotemporal learning. Our experimental results highlight the energy efficiency and robustness of SAM in these wide range of challenging tasks. The effects of SAM model variations on its working memory are also explored, hoping to offer insight into the biological mechanisms underlying working memory in the brain. The SAM model is the first attempt to integrate the capabilities of spike-driven learning and working memory in a unified single neuron with multiple timescale dynamics. The competitive performance of SAM could potentially contribute to the development of efficient adaptive neuromorphic computing systems for various applications from robotics to edge computing.

Encoding time in neural dynamic regimes with distinct computational tradeoffs

Converging evidence suggests the brain encodes time in dynamic patterns of neural activity, including neural sequences, ramping activity, and complex dynamics. Most temporal tasks, however, require more than just encoding time, and can have distinct computational requirements including the need to exhibit temporal scaling, generalize to novel contexts, or robustness to noise. It is not known how neural circuits can encode time and satisfy distinct computational requirements, nor is it known whether similar patterns of neural activity at the population level can exhibit dramatically different computational or generalization properties. To begin to answer these questions, we trained RNNs on two timing tasks based on behavioral studies. The tasks had different input structures but required producing identically timed output patterns. Using a novel framework we quantified whether RNNs encoded two intervals using either of three different timing strategies: scaling, absolute, or stimulus-specific dynamics. We found that similar neural dynamic patterns at the level of single intervals, could exhibit fundamentally different properties, including, generalization, the connectivity structure of the trained networks, and the contribution of excitatory and inhibitory neurons. Critically, depending on the task structure RNNs were better suited for generalization or robustness to noise. Further analysis revealed different connection patterns underlying the different regimes. Our results predict that apparently similar neural dynamic patterns at the population level (e.g., neural sequences) can exhibit fundamentally different computational properties in regards to their ability to generalize to novel stimuli and their robustness to noise—and that these differences are associated with differences in network connectivity and distinct contributions of excitatory and inhibitory neurons. We also predict that the task structure used in different experimental studies accounts for some of the experimentally observed variability in how networks encode time.

The ability to tell time and anticipate when external events will occur are among the most fundamental computations the brain performs. Converging evidence suggests the brain encodes time through changing patterns of neural activity. Different temporal tasks, however, have distinct computational requirements, such as the need to flexibly scale temporal patterns or generalize to novel inputs. To understand how networks can encode time and satisfy different computational requirements we trained recurrent neural networks (RNNs) on two timing tasks that have previously been used in behavioral studies. Both tasks required producing identically timed output patterns. Using a novel framework to quantify how networks encode different intervals, we found that similar patterns of neural activity—neural sequences—were associated with fundamentally different underlying mechanisms, including the connectivity patterns of the RNNs. Critically, depending on the task the RNNs were trained on, they were better suited for generalization or robustness to noise. Our results predict that similar patterns of neural activity can be produced by distinct RNN configurations, which in turn have fundamentally different computational tradeoffs. Our results also predict that differences in task structure account for some of the experimentally observed variability in how networks encode time.

A flexible Bayesian framework for unbiased estimation of timescales

Timescales characterize the pace of change for many dynamic processes in nature. Timescales are usually estimated by fitting the exponential decay of data autocorrelation in the time or frequency domain. Here we show that this standard procedure often fails to recover the correct timescales due to a statistical bias arising from the finite sample size. We develop an alternative approach which estimates timescales by fitting the sample autocorrelation or power spectrum with a generative model based on a mixture of Ornstein-Uhlenbeck (OU) processes using adaptive approximate Bayesian computations (aABC). Our method accounts for finite sample size and noise in data and returns a posterior distribution of timescales that quantifies the estimation uncertainty and can be used for model selection. We demonstrate the accuracy of our method on synthetic data and illustrate its application to recordings from primate cortex. We provide a customizable Python package implementing our framework with different generative models suitable for diverse applications.

Differential mechanisms underlie trace and delay conditioning in Drosophila

Two forms of associative learning-delay conditioning and trace conditioning-have been widely investigated in humans and higher-order mammals¹. In delay conditioning, an unconditioned stimulus (for example, an electric shock) is introduced in the final moments of a conditioned stimulus (for example, a tone), with both ending at the same time. In trace conditioning, a 'trace' interval separates the conditioned stimulus and the unconditioned stimulus. Trace conditioning therefore relies on maintaining a neural representation of the conditioned stimulus after its termination (hence making distraction possible²), to learn the conditioned stimulus-unconditioned stimulus contingency³; this makes it more cognitively demanding than delay conditioning⁴. Here, by combining virtual-reality behaviour with neurogenetic manipulations and in vivo two-photon brain imaging, we show that visual trace conditioning and delay conditioning in Drosophila mobilize R2 and R4m ring neurons in the ellipsoid body. In trace conditioning, calcium transients during the trace interval show increased oscillations and slower declines over repeated training, and both of these effects are sensitive to distractions. Dopaminergic activity accompanies signal persistence in ring neurons, and this is decreased by distractions solely during trace conditioning. Finally, dopamine D1-like and D2-like receptor signalling in ring neurons have different roles in delay and trace conditioning; dopamine D1-like receptor 1 mediates both forms of conditioning, whereas the dopamine D2-like receptor is involved exclusively in sustaining ring neuron activity during the trace interval of trace conditioning. These observations are similar to those previously reported in mammals during arousal⁵, prefrontal activation⁶ and high-level cognitive learning^7,8.

Neural Mechanisms of Working Memory Accuracy Revealed by Recurrent Neural Networks

Understanding the neural mechanisms of working memory has been a long-standing Neuroscience goal. Bump attractor models have been used to simulate persistent activity generated in the prefrontal cortex during working memory tasks and to study the relationship between activity and behavior. How realistic the assumptions of these models are has been a matter of debate. Here, we relied on an alternative strategy to gain insights into the computational principles behind the generation of persistent activity and on whether current models capture some universal computational principles. We trained Recurrent Neural Networks (RNNs) to perform spatial working memory tasks and examined what aspects of RNN activity accounted for working memory performance. Furthermore, we compared activity in fully trained networks and immature networks, achieving only imperfect performance. We thus examined the relationship between the trial-to-trial variability of responses simulated by the network and different aspects of unit activity as a way of identifying the critical parameters of memory maintenance. Properties that spontaneously emerged in the artificial network strongly resembled persistent activity of prefrontal neurons. Most importantly, these included drift of network activity during the course of a trial that was causal to the behavior of the network. As a consequence, delay period firing rate and behavior were positively correlated, in strong analogy to experimental results from the prefrontal cortex. These findings reveal that delay period activity is computationally efficient in maintaining working memory, as evidenced by unbiased optimization of parameters in artificial neural networks, oblivious to the properties of prefrontal neurons.

Muscarinic Acetylcholine Receptor Localization on Distinct Excitatory and Inhibitory Neurons Within the ACC and LPFC of the Rhesus Monkey

Acetylcholine (ACh) can act on pre- and post-synaptic muscarinic receptors (mAChR) in the cortex to influence a myriad of cognitive processes. Two functionally-distinct regions of the prefrontal cortex-the lateral prefrontal cortex (LPFC) and the anterior cingulate cortex (ACC)-are differentially innervated by ascending cholinergic pathways yet, the nature and organization of prefrontal-cholinergic circuitry in primates are not well understood. Using multi-channel immunohistochemical labeling and high-resolution microscopy, we found regional and laminar differences in the subcellular localization and the densities of excitatory and inhibitory subpopulations expressing m1 and m2 muscarinic receptors, the two predominant cortical mAChR subtypes, in the supragranular layers of LPFC and ACC in rhesus monkeys (Macaca mulatta). The subset of m1+/m2+ expressing SMI-32+ pyramidal neurons labeled in layer 3 (L3) was denser in LPFC than in ACC, while m1+/m2+ SMI-32+ neurons co-expressing the calcium-binding protein, calbindin (CB) was greater in ACC. Further, we found between-area differences in laminar m1+ dendritic expression, and m2+ presynaptic localization on cortico-cortical (VGLUT1+) and sub-cortical inputs (VGLUT2+), suggesting differential cholinergic modulation of top-down vs. bottom-up inputs in the two areas. While almost all inhibitory interneurons-identified by their expression of parvalbumin (PV+), CB+, and calretinin (CR+)-expressed m1+, the localization of m2+ differed by subtype and area. The ACC exhibited a greater proportion of m2+ inhibitory neurons compared to the LPFC and had a greater density of presynaptic m2+ localized on inhibitory (VGAT+) inputs targeting proximal somatodendritic compartments and axon initial segments of L3 pyramidal neurons. These data suggest a greater capacity for m2+-mediated cholinergic suppression of inhibition in the ACC compared to the LPFC. The anatomical localization of muscarinic receptors on ACC and LPFC micro-circuits shown here contributes to our understanding of diverse cholinergic neuromodulation of functionally-distinct prefrontal areas involved in goal-directed behavior, and how these interactions maybe disrupted in neuropsychiatric and neurological conditions.

Traveling waves in the prefrontal cortex during working memory

Neural oscillations are evident across cortex but their spatial structure is not well- explored. Are oscillations stationary or do they form "traveling waves", i.e., spatially organized patterns whose peaks and troughs move sequentially across cortex? Here, we show that oscillations in the prefrontal cortex (PFC) organized as traveling waves in the theta (4-8Hz), alpha (8-12Hz) and beta (12-30Hz) bands. Some traveling waves were planar but most rotated. The waves were modulated during performance of a working memory task. During baseline conditions, waves flowed bidirectionally along a specific axis of orientation. Waves in different frequency bands could travel in different directions. During task performance, there was an increase in waves in one direction over the other, especially in the beta band.

Towards biologically constrained attractor models of schizophrenia

Alterations in neuromodulation or synaptic transmission in biophysical attractor network models, as proposed by the dominant dopaminergic and glutamatergic theories of schizophrenia, successfully mimic working memory (WM) deficits in people with schizophrenia (PSZ). Yet, multiple, often opposing alterations in memory circuits can lead to the same behavioral patterns in these network models. Here, we critically revise the computational and experimental literature that links NMDAR hypofunction to WM precision loss in PSZ. We show in network simulations that currently available experimental evidence cannot set apart competing biophysical accounts. Critical points to resolve are the effects of increases vs. decreases in E/I ratio (e.g. through NMDAR blockade) on firing rate tuning and shared noise modulations and possible concomitant deficits in short-term plasticity. We argue that these concerted experimental and computational efforts will lead to a better understanding of the neurobiology underlying cognitive deficits in PSZ.

Emergence of prefrontal neuron maturation properties by training recurrent neural networks in cognitive tasks

Working memory and response inhibition are functions that mature relatively late in life, after adolescence, paralleling the maturation of the prefrontal cortex. The link between behavioral and neural maturation is not obvious, however, making it challenging to understand how neural activity underlies the maturation of cognitive function. To gain insights into the nature of observed changes in prefrontal activity between adolescence and adulthood, we investigated the progressive changes in unit activity of recurrent neural networks as they were trained to perform working memory and response inhibition tasks. These included increased delay period activity during working memory tasks and increased activation in antisaccade tasks. These findings reveal universal properties underlying the neuronal computations behind cognitive tasks and explicate the nature of changes that occur as the result of developmental maturation.

•

Properties of RNN networks during training offer insights in prefrontal maturation

•

Fully trained networks exhibit higher levels of activity in working memory tasks

•

Trained networks also exhibit higher activation in antisaccade tasks

•

Partially trained RNNs can generate accurate predictions of immature PFC activity

Natural and Artificial Intelligence: A brief introduction to the interplay between AI and neuroscience research

Neuroscience and artificial intelligence (AI) share a long history of collaboration. Advances in neuroscience, alongside huge leaps in computer processing power over the last few decades, have given rise to a new generation of in silico neural networks inspired by the architecture of the brain. These AI systems are now capable of many of the advanced perceptual and cognitive abilities of biological systems, including object recognition and decision making. Moreover, AI is now increasingly being employed as a tool for neuroscience research and is transforming our understanding of brain functions. In particular, deep learning has been used to model how convolutional layers and recurrent connections in the brain's cerebral cortex control important functions, including visual processing, memory, and motor control. Excitingly, the use of neuroscience-inspired AI also holds great promise for understanding how changes in brain networks result in psychopathologies, and could even be utilized in treatment regimes. Here we discuss recent advancements in four areas in which the relationship between neuroscience and AI has led to major advancements in the field; (1) AI models of working memory, (2) AI visual processing, (3) AI analysis of big neuroscience datasets, and (4) computational psychiatry.