Michal Zochowski

Introduction: We characterized properties of class of binary models where, as observed in biological networks, excitatory neurons are structurally and functionally separated from inhibitory units. We investigate the respective roles the two populations play in memory storage. Methods: The network is composed of separated excitatory and inhibitory layer. New patterns, represented as activation and inactivation of binary units in excitatory layer, are stored in the network through recruitment and training of inhibitory units that are grouped into individual modules and interact with excitatory layer. At the same time, the inhibitory modules compete for activation based on the signal magnitude they receive from the excitatory layer. Results: We show that inhibitory layer plays a critical role in memory storage and management, and that capacity of the network scales proportionally to number of inhibitory neurons. Further, we demonstrate that performance of the network is only gradually diminished when excitatory‐to‐excitatory (E‐E) connections are removed but critically depends on inhibitory‐to‐excitatory (I‐E) connections. We further show advantages of so designed coding scheme in terms of memory capacity, its expansion with progressive storage of new memories as well as network behavior for large memory loading. Discussion: These results are in line with new experimental work showing that inhibitory interneurons are playing critical role in memory storage and recall in the brain networks and may also address why generally excitatory networks exhibit sparser reciprocal connectivity as compared to connections to/from inhibitory units.

Hidden hearing loss (HHL) is an auditory neuropathy characterized by altered auditory nerve responses despite normal hearing thresholds. Recent experimental and computational studies suggest that permanent disruptions to heminode positions in spiral ganglion neuron (SGN) fibers can contribute to these deficits. However, the interaction between heminode disruption and noisy backgrounds ubiquitous in daily listening remains unexplored. This study investigates how background noise affects auditory processing with these peripheral disorders and how deficits propagate to downstream sound localization circuits in the superior olivary complex. We developed computational models of SGN fibers with mild and severe degrees of heminode disruption, subjected to sinusoidal tone stimuli in the presence of background noise with varying spectral characteristics. We analyzed the phase-locking of SGN fiber responses to the stimulus tone and modeled the subsequent effects on interaural time difference (ITD) sensitivity in the medial superior olive (MSO) using a binaural localization network. We found that near-tone-frequency noise disrupted SGN phase locking through cycle-to-cycle variability in spike phases, with effects consistent across tone frequencies. Mild heminode disruption produced frequency-dependent degradation in SGN phase locking, with effects observed only at higher frequencies tested (600–1000 Hz), without reducing overall firing rates. Critically, the effects of noise and heminode disruption were additive, with combined exposure leading to reduced ITD sensitivity and large temporal fluctuations in MSO responses. Severe heminode disruption, which additionally reduced firing rates at the SGN fibers and subsequent stages, produced profound localization deficits across all frequencies tested. Thus, our model results suggest that noisy environments exacerbate auditory deficits from peripheral disorders implicated in HHL and could potentially impair speech intelligibility through degradation in localization ability. This model may be useful for understanding the downstream impacts of SGN neuropathies.

We developed and characterized properties of new class of binary models where, as observed in biological networks, excitatory neurons are functionally separated from inhibitory units. New patterns, represented as activation and inactivation of binary units in excitatory layer, are stored in the network through recruitment and training of inhibitory units that are group into individual modules and interact with excitatory layer. We investigate the roles the two populations play in memory storage and show that inhibitory layer plays a critical role in memory storage and management and that capacity of this new type of network scales with number of inhibitory neurons. Further, we show that performance of the network is only gradually diminished when excitatory to excitatory connections are removed, but critically depends on inhibitory to excitatory connections. These results are in line with new experimental work showing that inhibitory interneurons are playing critical role in memory storage and recall in the brain networks and may also address why generally excitatory networks exhibit sparser reciprocal connectivity as compared to connections to/from inhibitory units. We further show advantages of so designed coding shame in terms of memory capacity, its expansion with progressive storage of new memories as well as network behavior for large memory loading.

Across vertebrate species, sleep consists of repeating cycles of NREM followed by REM. However, their respective functions, and their stereotypic cycling pattern are not well understood. Using a simplified biophysical network model, we investigate the potential role of cholinergic modulation, acting via the muscarinic receptors, on network dynamics and memory consolidation. We show that low and high cholinergic levels associated with NREM and REM sleep, respectively, may play critical, sequential roles in memory consolidation. The network dynamics that facilitate these roles arise through alteration of neural excitability and changes to network-wide excitatory/inhibitory balance. At low acetylcholine (ACh) levels, reduced activation of inhibitory neurons leads to network-wide disinhibition and bursts of synchronized activity led by engram neurons, driving recruitment of additional excitatory neurons into the engram. In contrast, at high ACh levels, increased network inhibition suppresses firing in all but the most strongly recruited excitatory neurons, pruning the expanded engram population. Together, these results provide a testable hypothesis regarding the role of sleep state-specific cholinergic modulation in the process of memory consolidation. Across animal species sleep stereotypically is composed of non-REM (non-rapid eye movement, NREM) and REM (rapid eye movement) phases, with NREM preceding REM. It remains not clear what differential functions those two phases play in sleep dependent memory consolidation – a process leading to better recollection of recently experienced events. Using a simplified computational model combined with analysis of in vivo recordings from mouse hippocampus, we show that changes in acetylcholine levels during NREM and REM sleep may drive these different roles in memory consolidation. Specifically, we show that low levels of acetylcholine during NREM disinhibit activity of otherwise quiescent subpopulations of excitatory neurons permitting the expansion of memory traces. In contrast, high levels of acetylcholine during REM create more competitive environment for neuronal activity leading to inhibition of less driven cells and preventing an overlap in memory representations when more than one memory traces are involved. We next further demonstrate advantages of the physiological structure of sleep in aiding this process.

Neuromodulatory processes in the brain can critically change signal processing on a cellular level, leading to dramatic changes in network level reorganization. Here, we use coupled nonidentical Kuramoto oscillators to investigate how changes in the shape of phase response curves from Type 1 to Type 2, mediated by varying ACh levels, coupled with activity-dependent plasticity may alter network reorganization. We first show that, when plasticity is absent, the Type 1 networks with symmetric adjacency matrix, as expected, exhibit asynchronous dynamics with oscillators of the highest natural frequency robustly evolving faster in terms of their phase dynamics. However, interestingly, Type 1 networks with an asymmetric connectivity matrix can produce stable synchrony (so-called splay states) with complex phase relationships. At the same time, Type 2 networks synchronize independent of the symmetry of their connectivity matrix, with oscillators locked so that those with higher natural frequency have a constant phase lead as compared to those with lower natural frequency. This relationship establishes a robust mapping between the frequency and oscillators' phases in the network, leading to structure and frequency mapping when plasticity is present. Finally, we show that biologically realistic, phase-locking dependent, connection plasticity naturally produces splay states in Type 1 networks that do not display the structure-frequency reorganization observed in synchronized Type II networks. These results indicate that the formation of splay states in the brain could be a common phenomenon.

Neuromodulatory processes in the brain can critically change signal processing on a cellular level leading to dramatic changes in network level reorganization. Here, we use coupled non-identical Kuramoto oscillators to investigate how changes in the shape of phase response curves from Type 1 to Type 2, mediated by varying ACh levels, coupled with activity dependent plasticity may alter network reorganization. We first show that when plasticity is absent, the Type 1 networks, as expected, exhibit asynchronous dynamics with oscillators of the highest natural frequency robustly evolving faster in terms of their phase dynamics. At the same time, the Type 2 networks synchronize, with oscillators locked so that the ones with higher natural frequency have a constant phase lead as compared to the ones with lower natural frequency. This relationship establishes a robust mapping between the frequency and oscillators’ phases in the network, leading to structure/frequency mapping when plasticity is present. Further we show that while connection plasticity can produce stable synchrony (so called splay states) in Type 1 networks, the structure/frequency reorganization observed in Type 2 networks is not present.

Forebrain acetylcholine (ACh) signaling has been shown to drive attention and learning. Recent experimental evidence of spatially and temporally constrained cholinergic signaling has sparked interest to investigate how it facilitates stimulus-induced learning. We use biophysical excitatory-inhibitory (E-I) multi-module neural network models to show that external stimuli and ACh signaling can mediate spatially constrained synaptic potentiation patterns. The effects of ACh on neural excitability are simulated by varying the conductance of a muscarinic receptor-regulated hyperpolarizing slow K+ current (m-current). Each network module consists of an E-I network with local excitatory connectivity and global inhibitory connectivity. The modules are interconnected with plastic excitatory synaptic connections, that change via a spike-timing-dependent plasticity (STDP) rule. Our results indicate that spatially constrained ACh release influences the information flow represented by network dynamics resulting in selective reorganization of inter-module interactions. Moreover the information flow depends on the level of synchrony in the network. For highly synchronous networks, the more excitable module leads firing in the less excitable one resulting in strengthening of the outgoing connections from the former and weakening of its incoming synapses. For networks with more noisy firing patterns, activity in high ACh regions is prone to induce feedback firing of synchronous volleys and thus strengthening of the incoming synapses to the more excitable region and weakening of outgoing synapses. Overall, these results suggest that spatially and directionally specific plasticity patterns, as are presumed necessary for feature binding, can be mediated by spatially constrained ACh release.

Across vertebrate species, sleep consists of repeating cycles of NREM followed by REM. However, the respective functions of NREM, REM, and their stereotypic cycling pattern are not well understood. Using a simplified biophysical network model, we show that NREM and REM sleep can play differential and critical roles in memory consolidation primarily regulated, based on state-specific changes in cholinergic signaling. Within this network, decreasing and increasing muscarinic acetylcholine (ACh) signaling during bouts of NREM and REM, respectively, differentially alters neuronal excitability and excitatory/inhibitory balance. During NREM, deactivation of inhibitory neurons leads to network-wide disinhibition and bursts of synchronized activity led by firing in engram neurons. These features strengthen connections from the original engram neurons to less-active network neurons. In contrast, during REM, an increase in network inhibition suppresses firing in all but the most-active excitatory neurons, leading to competitive strengthening/pruning of the memory trace. We tested the predictions of the model against in vivo recordings from mouse hippocampus during active sleep-dependent memory storage. Consistent with modeling results, we find that functional connectivity between CA1 neurons changes differentially at transition from NREM to REM sleep during learning. Returning to the model, we find that an iterative sequence of state-specific activations during NREM/REM cycling is essential for memory storage in the network, serving a critical role during simultaneous consolidation of multiple memories. Together these results provide a testable mechanistic hypothesis for the respective roles of NREM and REM sleep, and their universal relative timing, in memory consolidation. Significance statement: Using a simplified computational model and in vivo recordings from mouse hippocampus, we show that NREM and REM sleep can play differential roles in memory consolidation. The specific neurophysiological features of the two sleep states allow for expansion of memory traces (during NREM) and prevention of overlap between different memory traces (during REM). These features are likely essential in the context of storing more than one new memory simultaneously within a brain network.

Rhythmic synchronization of neuronal firing patterns is a widely present phenomenon in the brain—one that seems to be essential for many cognitive processes. A variety of mechanisms contribute to generation and synchronization of network oscillations, ranging from intrinsic cellular excitability to network mediated effects. However, it is unclear how these mechanisms interact together. Here, using computational modeling of excitatory-inhibitory neural networks, we show that different synchronization mechanisms dominate network dynamics at different levels of excitation and inhibition (i.e. E/I levels) as synaptic strength is systematically varied. Our results show that with low synaptic strength networks are sensitive to external oscillatory drive as a synchronizing mechanism—a hallmark of resonance. In contrast, in a strongly-connected regime, synchronization is driven by network effects via the direct interaction between excitation and inhibition, and spontaneous oscillations and cross-frequency coupling emerge. Unexpectedly, we find that while excitation dominates network synchrony at low excitatory coupling strengths, inhibition dominates at high excitatory coupling strengths. Together, our results provide novel insights into the oscillatory modulation of firing patterns in different excitation/inhibition regimes.

General anesthetics work through a variety of molecular mechanisms while resulting in the common end point of sedation and loss of consciousness. Generally, the administration of common anesthetics induces reduction in synaptic excitation while promoting synaptic inhibition. Exogenous modulation of the anesthetics’ synaptic effects can help determine the neuronal pathways involved in anesthesia. For example, both animal and human studies have shown that exogenously induced increases in acetylcholine in the brain can elicit wakeful-like behavior despite the continued presence of the anesthetic. However, the underlying mechanisms of anesthesia reversal at the cellular level have not been investigated. Here we apply a computational model of a network of excitatory and inhibitory neurons to simulate the network-wide effects of anesthesia, due to changes in synaptic inhibition and excitation, and their reversal by cholinergic activation through muscarinic receptors. We use a differential evolution algorithm to fit model parameters to match measures of spiking activity, neuronal connectivity, and network dynamics recorded in the visual cortex of rodents during anesthesia with desflurane in vivo. We find that facilitating muscarinic receptor effects of acetylcholine on top of anesthetic-induced synaptic changes predicts the reversal of anesthetic suppression of neurons’ spiking activity, functional connectivity, as well as pairwise and population interactions. Thus, our model predicts a specific neuronal mechanism for the cholinergic reversal of anesthesia consistent with experimental behavioral observations.