Feedback-induced gain control in stochastic spiking networks

The joint influence of recurrent feedback and noise on gain control in a network of globally coupled spiking leaky integrate-and-fire neurons is studied theoretically and numerically. The context of our work is the origin of divisive versus subtractive gain control, as mixtures of these effects are seen in a variety of experimental systems. We focus on changes in the slope of the mean firing frequency-versus-input bias (f –I) curve when the gain control signal to the cells comes from the cells’ output spikes. Feedback spikes are modeled as alpha functions that produce an additive current in the current balance equation. For generality, they occur after a fixed minimum delay. We show that purely divisive gain control, i.e. changes in the slope of the f –I curve, arises naturally with this additive negative or positive feedback, due to a linearizing actions of feedback. Negative feedback alone lowers the gain, accounting in particular for gain changes in weakly electric fish upon pharmacological opening of the feedback loop as reported by Bastian (J Neurosci 6:553–562, 1986). When negative feedback is sufficiently strong it further causes oscillatory firing patterns which produce irregularities in the f –I curve. Small positive feedback alone increases the gain, but larger amounts cause abrupt jumps to higher firing frequencies. On the other hand, noise alone in open loop linearizes the f –I curve around threshold, and produces mixtures of divisive and subtractive gain control. With both noise and feedback, the combined gain control schemes produce a primarily divisive gain control shift, indicating the robustness of feedback gain control in stochastic networks. Similar results are found when the “input” parameter is the contrast of a time-varying signal rather than the bias current. Theoretical results are derived relating the slope of the f –I curve to feedback gain and noise strength. Good agreement with simulation results are found for inhibitory and excitatory feedback. Finally, divisive feedback is also found for conductance-based feedback (shunting or excitatory) with and without noise. This article is part of a special issue on Neuronal Dynamics of Sensory Coding.     

Diversity of Intrinsic Frequency Encoding Patterns in Rat Cortical Neurons—Mechanisms and Possible Functions

Extracellular recordings of single neurons in primary and secondary somatosensory cortices of monkeys in vivo have shown that their firing rate can increase, decrease, or remain constant in different cells, as the external stimulus frequency increases. We observed similar intrinsic firing patterns (increasing, decreasing or constant) in rat somatosensory cortex in vitro, when stimulated with oscillatory input using conductance injection (dynamic clamp). The underlying mechanism of this observation is not obvious, and presents a challenge for mathematical modelling. We propose a simple principle for describing this phenomenon using a leaky integrate-and-fire model with sinusoidal input, an intrinsic oscillation and Poisson noise. Additional enhancement of the gain of encoding could be achieved by local network connections amongst diverse intrinsic response patterns. Our work sheds light on the possible cellular and network mechanisms underlying these opposing neuronal responses, which serve to enhance signal detection.     

Action potentials in epithelial taste receptor cells induced by mucosal calcium

Summary Chemosensory cells in the taste bud of the tongue ofNecturus generate action potentials in response to electrical stimulation through a microelectrode, as recently described by Roper (Science,220:1311–1312, 1983). We report that the epithelial receptor cells also respond to 10mM CaCl₂, applied to the mucosal surface, with a depolarization which elicits action potentials when a threshold of –50 mV is reached. Since CaCl₂ is one of the taste stimuli in amphibia, the firing of action potentials by chemoreceptor cells may be part of the signal chain in gustatory reception of Ca ions.     

<Emphasis Type="BoldItalic">In vivo</Emphasis> conditions influence the coding of stimulus features by bursts of action potentials

The functional role of burst firing (i.e. the firing of packets of action potentials followed by quiescence) in sensory processing is still under debate. Should bursts be considered as unitary events that signal the presence of a particular feature in the sensory environment or is information about stimulus attributes contained within their temporal structure? We compared the coding of stimulus attributes by bursts in vivo and in vitro of electrosensory pyramidal neurons in weakly electric fish by computing correlations between burst and stimulus attributes. Our results show that, while these correlations were strong in magnitude and significant in vitro, they were actually much weaker in magnitude if at all significant in vivo. We used a mathematical model of pyramidal neuron activity in vivo and showed that such a model could reproduce the correlations seen in vitro, thereby suggesting that differences in burst coding were not due to differences in bursting seen in vivo and in vitro. We next tested whether variability in the baseline (i.e. without stimulation) activity of ELL pyramidal neurons could account for these differences. To do so, we injected noise into our model whose intensity was calibrated to mimic baseline activity variability as quantified by the coefficient of variation. We found that this noise caused significant decreases in the magnitude of correlations between burst and stimulus attributes and could account for differences between in vitro and in vivo conditions. We then tested this prediction experimentally by directly injecting noise in vitro through the recording electrode. Our results show that this caused a lowering in magnitude of the correlations between burst and stimulus attributes in vitro and gave rise to values that were quantitatively similar to those seen under in vivo conditions. While it is expected that noise in the form of baseline activity variability will lower correlations between burst and stimulus attributes, our results show that such variability can account for differences seen in vivo. Thus, the high variability seen under in vivo conditions has profound consequences on the coding of information by bursts in ELL pyramidal neurons. In particular, our results support the viewpoint that bursts serve as a detector of particular stimulus features but do not carry detailed information about such features in their structure.     

Caffeine Modulates Vesicle Release and Recovery at Cerebellar Parallel Fibre Terminals,Independently of Calcium and Cyclic AMP Signalling

BackgroundCerebellar parallel fibres release glutamate at both the synaptic active zone and at extrasynaptic sites—a process known as ectopic release. These sites exhibit different short-term and long-term plasticity, the basis of which is incompletely understood but depends on the efficiency of vesicle release and recycling. To investigate whether release of calcium from internal stores contributes to these differences in plasticity, we tested the effects of the ryanodine receptor agonist caffeine on both synaptic and ectopic transmission.MethodsWhole cell patch clamp recordings from Purkinje neurons and Bergmann glia were carried out in transverse cerebellar slices from juvenile (P16-20) Wistar rats.ConclusionsWe conclude that caffeine increases release probability and inhibits vesicle recovery at parallel fibre synapses, independently of known pharmacological targets. This complex effect would lead to potentiation of transmission at fibres firing at low frequencies, but depression of transmission at high frequency connections.     

Slope-Based Stochastic Resonance: How Noise Enables Phasic Neurons to Encode Slow Signals

Fundamental properties of phasic firing neurons are usually characterized in a noise-free condition. In the absence of noise, phasic neurons exhibit Class 3 excitability, which is a lack of repetitive firing to steady current injections. For time-varying inputs, phasic neurons are band-pass filters or slope detectors, because they do not respond to inputs containing exclusively low frequencies or shallow slopes. However, we show that in noisy conditions, response properties of phasic neuron models are distinctly altered. Noise enables a phasic model to encode low-frequency inputs that are outside of the response range of the associated deterministic model. Interestingly, this seemingly stochastic-resonance (SR) like effect differs significantly from the classical SR behavior of spiking systems in both the signal-to-noise ratio and the temporal response pattern. Instead of being most sensitive to the peak of a subthreshold signal, as is typical in a classical SR system, phasic models are most sensitive to the signal''s rising and falling phases where the slopes are steep. This finding is consistent with the fact that there is not an absolute input threshold in terms of amplitude; rather, a response threshold is more properly defined as a stimulus slope/frequency. We call the encoding of low-frequency signals with noise by phasic models a slope-based SR, because noise can lower or diminish the slope threshold for ramp stimuli. We demonstrate here similar behaviors in three mechanistic models with Class 3 excitability in the presence of slow-varying noise and we suggest that the slope-based SR is a fundamental behavior associated with general phasic properties rather than with a particular biological mechanism.     

Factors affecting phase synchronization in integrate-and-fire oscillators

Step changes in input current are known to induce partial phase synchrony in ensembles of leaky integrate-and-fire neurons operating in the oscillatory or “regular firing” regime. An analysis of this phenomenon in the absence of noise is presented based on the probability flux within an ensemble of generalized integrate-and-fire neurons. It is shown that the induction of phase synchrony by a step input can be determined by calculating the ratio of the voltage densities obtained from fully desynchronized ensembles firing at the pre and post-step firing rates. In the limit of low noise and in the absence of phase synchrony, the probability density as a function of voltage is inversely proportional to the time derivative along the voltage trajectory. It follows that the magnitude of phase synchronization depends on the degree to which a change in input leads to a uniform multiplication of the voltage derivative over the range from reset to spike threshold. This analysis is used to investigate several factors affecting phase synchronization including high firing rates, inputs modeled as conductances rather than currents, peri-threshold sodium currents, and spike-triggered potassium currents. Finally, we show that without noise, the equilibrium ensemble density is proportional to the phase response curve commonly used to analyze oscillatory systems. Action Editor: John Rinzel     

No neural evidence for dynamic auditory tuning of the A1 receptor in the ear of the noctuid moth, <Emphasis Type="Italic">Noctua pronuba</Emphasis>

By examining the mechanical properties of the tympanum of the noctuid moth, Noctua pronuba, Windmill et al. (2006) suggested that this insect increases (up-tunes) the frequencies of its best hearing when exposed to high intensity sounds (HIS) resembling the echolocation calls of attacking bats. We tested whether this biophysical phenomenon was encoded in the neural responses of this moth’s most sensitive auditory receptor (A1 cell) before and after exposure to HIS. We measured: (1) the number of A1 action potentials (spikes) per stimulus pulse; (2) the proportion of A1 spike periods below that determined to elicit evasive flight maneuvers and, (3) the change in A1 cell firing (spike number, interspike interval, stimulus/spike latency) over a duration of time similar to that in which up-tuning lasts. We observed no significant spiking response changes in the predicted direction to any of the frequencies tested following exposure to HIS and we observed only two of the 24 predicted time-dependent changes to A1 firing. These results indicate that tympanal up-tuning does not result in a change to this moth’s auditory frequency sensitivity and we suggest either sensillar resonances or increases in thoracic muscle tension following exposure to HIS as alternative explanations.     

Some properties of auditory neuron’s model trained by firing caused by tones modulated by low-frequency noise

In many sensory systems the formation of burst firing can be observed along a way from the periphery to the central nuclei. We investigate the putative transformation of spontaneous activity in the auditory pathway using a neuron model trained by real firing recorded in the auditory nuclei of the frog. The model has 200 separate inputs (neuronal spines). It is supposed that every spine is a coincidence detector. Its output (synaptic potential) sharply increases at emergence of the precisely certain interpulse interval in an input pulse sequence. If the total synaptic potentials excess a threshold, the model generates output spike, which changes weight of all spines according to the simplified Hebb principle. The model was trained by real firing caused in the auditory nuclei of the frog by tones modulated by low-frequency noise in the frequency ranges of 0–15 Hz, 0–50 Hz or 0–150 Hz. After that training the synaptic weights of every spine essentially changed. Thus, along with some increase of weights of spines tuned to boundary frequencies of modulating noise, the most characteristic change was the emphasizing weights of spines tuned to short interpulse intervals. As a result the spontaneous activity passed through the trained model became much more bursting. Efficiency of a signal transmission in model was higher when input spontaneous activity of real cells contains bursts of spikes. Results of modeling are discussed in connection with modern physiological data demonstrating the functional advantage of bursting.     

Efficiency of neural transmission as a function of synaptic noise, threshold, and source characteristics

There has been a growing interest in the estimation of information carried by a single neuron and multiple single units or population of neurons to specific stimuli. In this paper we analyze, inspired by article of Levy and Baxter (2002), the efficiency of a neuronal communication by considering dendrosomatic summation as a Shannon-type channel (1948) and by considering such uncertain synaptic transmission as part of the dendrosomatic computation. Specifically, we study Mutual Information between input and output signals for different types of neuronal network architectures by applying efficient entropy estimators. We analyze the influence of the following quantities affecting transmission abilities of neurons: synaptic failure, activation threshold, firing rate and type of the input source. We observed a number of surprising non-intuitive effects. It turns out that, especially for lower activation thresholds, significant synaptic noise can lead even to twofold increase of the transmission efficiency. Moreover, the efficiency turns out to be a non-monotonic function of the activation threshold. We find a universal value of threshold for which a local maximum of Mutual Information is achieved for most of the neuronal architectures, regardless of the type of the source (correlated and non-correlated). Additionally, to reach the global maximum the optimal firing rates must increase with the threshold. This effect is particularly visible for lower firing rates. For higher firing rates the influence of synaptic noise on the transmission efficiency is more advantageous. Noise is an inherent component of communication in biological systems, hence, based on our analysis, we conjecture that the neuronal architecture was adjusted to make more effective use of this attribute.     

Predicting stimulus-locked single unit spiking from cortical local field potentials

The rapidly increasing use of the local field potential (LFP) has motivated research to better understand its relation to the gold standard of neural activity, single unit (SU) spiking. We addressed this in an in vivo, awake, restrained mouse auditory cortical electrophysiology preparation by asking whether the LFP could actually be used to predict stimulus-evoked SU spiking. Implementing a Bayesian algorithm to predict the likelihood of spiking on a trial by trial basis from different representations of the despiked LFP signal, we were able to predict, with high quality and fine temporal resolution (2 ms), the time course of a SU’s excitatory or inhibitory firing rate response to natural species-specific vocalizations. Our best predictions were achieved by representing the LFP by its wide-band Hilbert phase signal, and approximating the statistical structure of this signal at different time points as independent. Our results show that each SU’s action potential has a unique relationship with the LFP that can be reliably used to predict the occurrence of spikes. This “signature” interaction can reflect both pre- and post-spike neural activity that is intrinsic to the local circuit rather than just dictated by the stimulus. Finally, the time course of this “signature” may be most faithful when the full bandwidth of the LFP, rather than specific narrow-band components, is used for representation.     

The potassium A-current,low firing rates and rebound excitation in Hodgkin-Huxley models

It is widely believed, following the work of Connor and Stevens (1971,J. Physiol. Lond. 214, 31–53) that the ability to fire action potentials over a wide frequency range, especially down to very low rates, is due to the transient, potassium A-current (I _A). Using a reduction of the classical Hodgkin-Huxley model, we study the effects ofI _A on steady firing rate, especially in the near-threshold regime for the onset of firing. A minimum firing rate of zero corresponds to a homoclinic bifurcation of periodic solutions at a critical level of stimulating current. It requires that the membrane's steady-state current-voltage relation be N-shaped rather than monotonic. For experimentally based genericI _A parameters, the model does not fire at arbitrarily low rates, although it can for the more atypicalI _A parameters given by Connor and Stevens for the crab axon. When theI _A inactivation rate is slow, we find that the transient potassium current can mediate more complex firing patterns, such as periodic bursting in some parameter regimes. The number of spikes per burst increases asg _A decreases and as inactivation rate decreases. We also study howI _A affects properties of transient voltage responses, such as threshold and firing latency for anodal break excitation. We provide mathematical explanations for several of these dynamic behaviors using bifurcation theory and averaging methods.     

Threshold behavior in the initiation of hippocampal population bursts

Hippocampal population discharges such as sharp waves, epileptiform firing, and GDPs recur at long and variable intervals. The mechanisms for their precise timing are not well understood. Here, we show that population bursts in the disinhibited CA3 region are initiated at a threshold level of population firing after recovery from a previous event. Each population discharge follows an active buildup period when synaptic traffic and cell firing increase to threshold levels. Single-cell firing can advance burst onset by increasing population firing to suprathreshold values. Population synchrony is suppressed when threshold frequencies cannot be reached due to reduced cellular excitability or synaptic efficacy. Reducing synaptic strength reveals partially synchronous population bursts that are curtailed by GABA(B)-mediated conductances. Excitatory glutamatergic transmission and delayed GABA(B)-mediated signals have opposing feedback effects on CA3 cell firing and so determine threshold behavior for population synchrony.     

Detection of ultrasonic cavitation based on low-frequency analysis of acoustic signal

The acoustic cavitation phenomenon constitutes a potential hazard in ultrasound diagnostics and therapy so that early and effective detection of cavitation is of great interest. However, cavitation might even bring a higher risk especially when an echocontrast agent based on microbubbles is used. The major goal of the present work was to develop a cavitation detection method based on increased level of cavitation noise in the range of low frequencies (about 1 Hz). This method was applied in vitro using a model of body fluid containing a model echocontrast agent, such as 5% solution of lyophilized egg albumin, which was sonicated by ultrasound disintegrator. Ultrasound signal evokes cavitation in microbubble suspension accompanied by a certain level of cavitation acoustic noise. The level of noise voltage increased in the frequency range of 0.1 to 2 Hz in the presence of cavitation. Hence, this method makes it possible to determine the value of cavitation threshold. In addition, we examined how the cavitation threshold is affected by temperature and viscosity. It was found that the cavitation threshold decreased with growing temperature while it increased with growing viscosity.     

Properties of human motoneurones and their synaptic noise deduced from motor unit recordings with the aid of computer modelling.

This paper reviews two new facets of the behaviour of human motoneurones; these were demonstrated by modelling combined with analysis of long periods of low-frequency tonic motor unit firing (sub-primary range). 1) A novel transformation of the interval histogram has shown that the effective part of the membrane's post-spike voltage trajectory is a segment of an exponential (rather than linear), with most spikes being triggered by synaptic noise before the mean potential reaches threshold. The curvature of the motoneurone's trajectory affects virtually all measures of its behaviour and response to stimulation. The 'trajectory' is measured from threshold, and so includes any changes in threshold during the interspike interval. 2) A novel rhythmic stimulus (amplitude-modulated pulsed vibration) has been used to show that the motoneurone produces appreciable phase-advance during sinusoidal excitation. At low frequencies, the advance increases with rising stimulus frequency but then, slightly below the motoneurones mean firing rate, it suddenly becomes smaller. The gain has a maximum for stimuli at the mean firing rate (the 'carrier'). Such behaviour is functionally important since it affects the motoneurone's response to any rhythmic input, whether generated peripherally by the receptors (as in tremor) or by the CNS (as with cortical oscillations). Low mean firing rates favour tremor, since the high gain and reduced phase advance at the 'carrier' reduce the stability of the stretch reflex.     

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.     

Spatiotemporally controlled cardiac conduction block using high-frequency electrical stimulation

Background

Methods for the electrical inhibition of cardiac excitation have long been sought to control excitability and conduction, but to date remain largely impractical. High-amplitude alternating current (AC) stimulation has been known to extend cardiac action potentials (APs), and has been recently exploited to terminate reentrant arrhythmias by producing reversible conduction blocks. Yet, low-amplitude currents at similar frequencies have been shown to entrain cardiac tissues by generation of repetitive APs, leading in some cases to ventricular fibrillation and hemodynamic collapse in vivo. Therefore, an inhibition method that does not lead to entrainment – irrespective of the stimulation amplitude (bound to fluctuate in an in vivo setting) – is highly desirable.

Methodology/Principal Findings

We investigated the effects of broader amplitude and frequency ranges on the inhibitory effects of extracellular AC stimulation on HL-1 cardiomyocytes cultured on microelectrode arrays, using both sinusoidal and square waveforms. Our results indicate that, at sufficiently high frequencies, cardiac tissue exhibits a binary response to stimulus amplitude with either prolonged APs or no effect, thereby effectively avoiding the risks of entrainment by repetitive firing observed at lower frequencies. We further demonstrate the ability to precisely define reversible local conduction blocks in beating cultures without influencing the propagation activity in non-blocked areas. The conduction blocks were spatiotemporally controlled by electrode geometry and stimuli duration, respectively, and sustainable for long durations (300 s).

Conclusion/Significance

Inhibition of cardiac excitation induced by high-frequency AC stimulation exhibits a binary response to amplitude above a threshold frequency, enabling the generation of reversible conduction blocks without the risks of entrainment. This inhibition method could yield novel approaches for arrhythmia modeling in vitro, as well as safer and more efficacious tools for in vivo cardiac mapping and radio-frequency ablation guidance applications.     

Frequency switching in a two-compartmental model of the dopaminergic neuron

Mid-brain dopaminergic (DA) neurons display two functionally distinct modes of electrical activity: low- and high-frequency firing. The high-frequency firing is linked to important behavioral events in vivo. However, it cannot be elicited by standard manipulations in vitro. We had suggested a two-compartmental model of the DA cell that united data on firing frequencies under different experimental conditions. We now analyze dynamics of this model. The analysis was possible due to introduction of timescale separation among variables. We formulate the requirements for low and high frequencies. We found that the modulation of the SK current gating controls the frequency rise under applied depolarization. This provides a new mechanism that limits the frequency in the control conditions and allows high-frequency responses to depolarization if the SK current gating is downregulated. The mechanism is based on changing Ca^2 + balance and can also be achieved by direct modulation of the balance. Interestingly, such changes do not affect the high-frequency oscillations under NMDA. Therefore, altering Ca^2 + balance allows combining the high-frequency response to NMDA activation with the inability of other treatments to effectively elevate the frequency. We conclude that manipulations affecting Ca^2 + balance are most effective in controlling the frequency range. This modeling prediction gives a clue to the mechanism of the high-frequency firing in the DA neuron in vivo and in vitro.     

Synchronous Clusters in a Noisy Inhibitory Neural Network

We study the stability and information encoding capacity of synchronized states in a neuronal network model that represents part of thalamic circuitry. Our model neurons have a Hodgkin-Huxley-type low-threshold calcium channel, display postinhibitory rebound, and are connected via GABAergic inhibitory synapses.We find that there is a threshold in synaptic strength, _c, below which there are no stable spiking network states. Above threshold the stable spiking state is a cluster state, where different groups of neurons fire consecutively, and each neuron fires with the same cluster each time. Weak noise destabilizes this state, but stronger noise drives the system into a different, self-organized, stochastically synchronized state. Neuronal firing is still organized in clusters, but individual neurons can hop from cluster to cluster. Noise can actually induce and sustain such a state below the threshold of synaptic strength. We do find a qualitative difference in the firing patterns between small (10 neurons) and large (1000 neurons) networks.We determine the information content of the spike trains in terms of two separate contributions: the spike-time jitter around cluster firing times, and the hopping from cluster to cluster. We quantify the information loss due to temporally correlated interspike intervals. Recent experiments on the locust olfactory system and striatal neurons suggest that the nervous system may actually use these two channels to encode separate and unique information.     

On the stationary state of a network of inhibitory spiking neurons

The background activity of a cortical neural network is modeled by a homogeneous integrate-and-fire network with unreliable inhibitory synapses. For the case of fast synapses, numerical and analytical calculations show that the network relaxes into a stationary state of high attention. The majority of the neurons has a membrane potential just below the threshold; as a consequence the network can react immediately – on the time scale of synaptic transmission- on external pulses. The neurons fire with a low rate and with a broad distribution of interspike intervals. Firing events of the total network are correlated over short time periods. The firing rate increases linearly with external stimuli. In the limit of infinitely large networks, the synaptic noise decreases to zero. Nevertheless, the distribution of interspike intervals remains broad. Action Editor: Misha Tsodyks