A Synchronization-Desynchronization Code for Natural Communication Signals

Neural heterogeneity and efficient population codes for communication signals

Efficient sensory coding implies that populations of neurons should represent information-rich aspects of a signal with little redundancy. Recent studies have shown that neural heterogeneity in higher brain areas enhances the efficiency of encoding by reducing redundancy across the population. Here, we study how neural heterogeneity in the early stages of sensory processing influences the efficiency of population codes. Through the analysis of in vivo recordings, we contrast the encoding of two types of communication signals of electric fishes in the most peripheral sensory area of the CNS, the electrosensory lateral line lobe (ELL). We show that communication signals used during courtship (big chirps) and during aggressive encounters (small chirps) are encoded by different populations of ELL pyramidal cells, namely I-cells and E-cells, respectively. Most importantly, we show that the encoding strategy differs for the two signals and we argue that these differences allow these cell types to encode specifically information-rich features of the signals. Small chirps are detected, and their timing is accurately signaled through stereotyped spike bursts, whereas the shape of big chirps is accurately represented by variable increases in firing rate. Furthermore, we show that the heterogeneity across I-cells enhances the efficiency of the population code and thus permits the accurate discrimination of different quality courtship signals. Our study shows the importance of neural heterogeneity early in a sensory system and that it initiates the sparsification of sensory representation thereby contributing to the efficiency of the neural code.

Burst-induced anti-Hebbian depression acts through short-term synaptic dynamics to cancel redundant sensory signals

Weakly electric fish can enhance the detection and localization of important signals such as those of prey in part by cancellation of redundant spatially diffuse electric signals due to, e.g., their tail bending. The cancellation mechanism is based on descending input, conveyed by parallel fibers emanating from cerebellar granule cells, that produces a negative image of the global low-frequency signals in pyramidal cells within the first-order electrosensory region, the electrosensory lateral line lobe (ELL). Here we demonstrate that the parallel fiber synaptic input to ELL pyramidal cell undergoes long-term depression (LTD) whenever both parallel fiber afferents and their target cells are stimulated to produce paired burst discharges. Paired large bursts (4-4) induce robust LTD over pre-post delays of up to +/-50 ms, whereas smaller bursts (2-2) induce weaker LTD. Single spikes (either presynaptic or postsynaptic) paired with bursts did not induce LTD. Tetanic presynaptic stimulation was also ineffective in inducing LTD. Thus, we have demonstrated a form of anti-Hebbian LTD that depends on the temporal correlation of burst discharge. We then demonstrated that the burst-induced LTD is postsynaptic and requires the NR2B subunit of the NMDA receptor, elevation of postsynaptic Ca(2+), and activation of CaMKIIbeta. A model incorporating local inhibitory circuitry and previously identified short-term presynaptic potentiation of the parallel fiber synapses further suggests that the combination of burst-induced LTD, presynaptic potentiation, and local inhibition may be sufficient to explain the generation of the negative image and cancellation of redundant sensory input by ELL pyramidal cells.

Kinetics of fast short-term depression are matched to spike train statistics to reduce noise

Short-term depression (STD) is observed at many synapses of the CNS and is important for diverse computations. We have discovered a form of fast STD (FSTD) in the synaptic responses of pyramidal cells evoked by stimulation of their electrosensory afferent fibers (P-units). The dynamics of the FSTD are matched to the mean and variance of natural P-unit discharge. FSTD exhibits switch-like behavior in that it is immediately activated with stimulus intervals near the mean interspike interval (ISI) of P-units (approximately 5 ms) and recovers immediately after stimulation with the slightly longer intervals (>7.5 ms) that also occur during P-unit natural and evoked discharge patterns. Remarkably, the magnitude of evoked excitatory postsynaptic potentials appear to depend only on the duration of the previous ISI. Our theoretical analysis suggests that FSTD can serve as a mechanism for noise reduction. Because the kinetics of depression are as fast as the natural spike statistics, this role is distinct from previously ascribed functional roles of STD in gain modulation, synchrony detection or as a temporal filter.

Transient signals trigger synchronous bursts in an identified population of neurons

It is an important task in neuroscience to find general principles that relate neural codes to the structure of the signals they encode. The structure of sensory signals can be described in many ways, but one important categorization distinguishes continuous from transient signals. We used the communication signals of the weakly electric fish to reveal how transient signals (chirps) can be easily distinguished from the continuous signal they disrupt. These communication signals-low-frequency sinusoids interrupted by high-frequency transients-were presented to pyramidal cells of the electrosensory lateral line lobe (ELL) during in vivo recordings. We show that a specific population of electrosensory neurons encodes the occurrence of the transient signal by synchronously producing a burst of spikes, whereas bursting was neither common nor synchronous in response to the continuous signal. We also confirmed that burst can be triggered by low-frequency modulations typical of prey signals. However, these bursts are more common in a different segment of the ELL and during spatially localized stimulation. These localized stimuli will elicit synchronized bursting only in a restricted number of cells the receptive fields of which overlap the spatial extent of the stimulus. Therefore the number of cells simultaneously producing a burst and the ELL segment responding most strongly may carry the information required to disambiguate chirps from prey signals. Finally we show that the burst response to chirps is due to a biophysical mechanism previously characterized by in vitro studies of electrosensory neurons. We conclude that bursting and synchrony across cells are important mechanisms used by sensory neurons to carry the information about behaviorally relevant but transient signals.

Divisive gain modulation with dynamic stimuli in integrate-and-fire neurons

The modulation of the sensitivity, or gain, of neural responses to input is an important component of neural computation. It has been shown that divisive gain modulation of neural responses can result from a stochastic shunting from balanced (mixed excitation and inhibition) background activity. This gain control scheme was developed and explored with static inputs, where the membrane and spike train statistics were stationary in time. However, input statistics, such as the firing rates of pre-synaptic neurons, are often dynamic, varying on timescales comparable to typical membrane time constants. Using a population density approach for integrate-and-fire neurons with dynamic and temporally rich inputs, we find that the same fluctuation-induced divisive gain modulation is operative for dynamic inputs driving nonequilibrium responses. Moreover, the degree of divisive scaling of the dynamic response is quantitatively the same as the steady-state responses—thus, gain modulation via balanced conductance fluctuations generalizes in a straight-forward way to a dynamic setting.

Many neural computations, including sensory and motor processing, require neurons to control their sensitivity (often termed ‘gain’) to stimuli. One common form of gain manipulation is divisive gain control, where the neural response to a specific stimulus is simply scaled by a constant. Most previous theoretical and experimental work on divisive gain control have assumed input statistics to be constant in time. However, realistic inputs can be highly time-varying, often with time-varying statistics, and divisive gain control remains to be extended to these cases. A widespread mechanism for divisive gain control for static inputs is through an increase in stimulus independent membrane fluctuations. We address the question of whether this divisive gain control scheme is indeed operative for time-varying inputs. Using simplified spiking neuron models, we employ accurate theoretical methods to estimate the dynamic neural response. We find that gain control via membrane fluctuations does indeed extend to the time-varying regime, and moreover, the degree of divisive scaling does not depend on the timescales of the driving input. This significantly increases the relevance of this form of divisive gain control for neural computations where input statistics change in time, as expected during normal sensory and motor behavior.

Postsynaptic receptive field size and spike threshold determine encoding of high-frequency information via sensitivity to synchronous presynaptic activity

Parallel sensory streams carrying distinct information about various stimulus properties have been observed in several sensory systems, including the visual system. What remains unclear is why some of these streams differ in the size of their receptive fields (RFs). In the electrosensory system, neurons with large RFs have short-latency responses and are tuned to high-frequency inputs. Conversely, neurons with small RFs are low-frequency tuned and exhibit longer-latency responses. What principle underlies this organization? We show experimentally that synchronous electroreceptor afferent (P-unit) spike trains selectively encode high-frequency stimulus information from broadband signals. This finding relies on a comparison of stimulus-spike output coherence using output trains obtained by either summing pairs of recorded afferent spike trains or selecting synchronous spike trains based on coincidence within a small time window. We propose a physiologically realistic decoding mechanism, based on postsynaptic RF size and postsynaptic output rate normalization that tunes target pyramidal cells in different electrosensory maps to low- or high-frequency signal components. By driving realistic neuron models with experimentally obtained P-unit spike trains, we show that a small RF is matched with a postsynaptic integration regime leading to responses over a broad range of frequencies, and a large RF with a fluctuation-driven regime that requires synchronous presynaptic input and therefore selectively encodes higher frequencies, confirming recent experimental data. Thus our work reveals that the frequency content of a broadband stimulus extracted by pyramidal cells, from P-unit afferents, depends on the amount of feedforward convergence they receive.

Electrocommunication signals in free swimming brown ghost knifefish, Apteronotus leptorhynchus

Brown ghost knifefish, Apteronotus leptorhynchus, are a species of weakly electric fish that produce a continuous electric organ discharge (EOD) that is used in navigation, prey capture and communication. Stereotyped modulations of EOD frequency and amplitude are common in social situations and are thought to serve as communication signals. Of these modulations, the most commonly studied is the chirp. This study presents a quantitative analysis of chirp production in pairs of free-swimming, physically interacting male and female A. leptorhynchus. Under these conditions, we found that in addition to chirps, the fish commonly produce a second signal type, a type of frequency rise called abrupt frequency rises, AFRs. By quantifying the behaviours associated with signal production, we find that Type 2 chirps tend to be produced when the fish are apart, following periods of low aggression, whereas AFRs tend to be produced when the fish are aggressively attacking one another in close proximity. This study is the first to our knowledge that quantitatively describes both electrocommunication signalling and behavioural correlates on a subsecond time-scale in a wave-type weakly electric fish.

Ionic and neuromodulatory regulation of burst discharge controls frequency tuning

Sensory neurons encode natural stimuli by changes in firing rate or by generating specific firing patterns, such as bursts. Many neural computations rely on the fact that neurons can be tuned to specific stimulus frequencies. It is thus important to understand the mechanisms underlying frequency tuning. In the electrosensory system of the weakly electric fish, Apteronotus leptorhynchus, the primary processing of behaviourally relevant sensory signals occurs in pyramidal neurons of the electrosensory lateral line lobe (ELL). These cells encode low frequency prey stimuli with bursts of spikes and high frequency communication signals with single spikes. We describe here how bursting in pyramidal neurons can be regulated by intrinsic conductances in a cell subtype specific fashion across the sensory maps found within the ELL, thereby regulating their frequency tuning. Further, the neuromodulatory regulation of such conductances within individual cells and the consequences to frequency tuning are highlighted. Such alterations in the tuning of the pyramidal neurons may allow weakly electric fish to preferentially select for certain stimuli under various behaviourally relevant circumstances.

Intrinsic frequency tuning in ELL pyramidal cells varies across electrosensory maps

The tuning of neuronal responsiveness to specific stimulus frequencies is an important computation across many sensory modalities. The weakly electric fish Apteronotus leptorhynchus detects amplitude modulations of a self-generated quasi-sinusoidal electric organ discharge to sense its environment. These fish have to parse a complicated electrosensory environment with a wide range of possible frequency content. One solution has been to create multiple representations of the sensory input across distinct maps in the electrosensory lateral line lobe (ELL) that participate in distinct behavioral functions. E- and I-type pyramidal cells in the ELL that process sensory input further exhibit a preferred range of stimulus frequencies in relation to the different behaviors and sensory maps. We tested the hypothesis that variations in the intrinsic spiking mechanism of E- and I-type pyramidal cells contribute to map-specific frequency tuning. We find that E-cells exhibit a systematic change in their intrinsic spike characteristics and frequency tuning across sensory maps, whereas I-cells are constant in both spike characteristics and frequency tuning. As frequency tuning becomes more high-pass in E-cells, the refractory variables of spike half-width and afterhyperpolarization magnitude increase, spike threshold increases, adaptation becomes faster, and the gain of the spiking response decreases. These findings indicate that frequency tuning across sensory maps in the ELL is supported by differences in the intrinsic spike characteristics of pyramidal cells, revealing a link between cellular biophysical properties and signal processing in sensory maps with defined behavioral roles.

Correlation and synchrony transfer in integrate-and-fire neurons: basic properties and consequences for coding

We study how pairs of neurons transfer correlated input currents into correlated spikes. Over rapid time scales, correlation transfer increases with both spike time variability and rate; the dependence on variability disappears at large time scales. This persists for a nonlinear membrane model and for heterogeneous cell pairs, but strong nonmonotonicities follow from refractory effects. We present consequences for population coding and for the encoding of time-varying stimuli.

Effect of synaptic plasticity on sensory coding and steady-state filtering properties in the electric sense

Our modeling study examines short-term plasticity at the synapse between afferents from electroreceptors and pyramidal cells in the electrosensory lateral lobe (ELL) of the weakly electric fish Apteronotus leptorhynchus. It focusses on steady-state filtering and coherence-based coding properties. While developed for electroreception, our study exposes general functional features for different mixtures of depression and facilitation. Our computational model, constrained by the available in vivo and in vitro data, consists of a synapse onto a deterministic leaky integrate-and-fire (LIF) neuron. The synapse is either depressing (D), facilitating (F) or both (FD), and is driven by a sinusoidally or randomly modulated Poisson process. Due to nonlinearity, numerically computed input-output transfer functions are used to determine the filtering properties. The gain of the response at each sinusoidally modulated frequency is computed by dividing the fitted amplitudes of the input and output cycle histograms of the LIF models. While filtering is always low-pass for F alone, D alone exhibits a gain resonance (non-monotonicity) at a frequency that decreases with increasing recovery time constant of synaptic depression (tau(d)). This resonance is mitigated by the presence of F. For D, F and FD, coherence improves as the synaptic conductance time constant (tau(g)) increases, yet the mutual information per spike decreases. The information per spike for D and F follows opposite trends as their respective time constants increase. The broadband but non-monotonic gain and coherence functions seen in vivo suggest that D and perhaps FD dynamics are involved at this synapse. Our results further predict that the likely synaptic configuration is a slower tau(g), e.g. via a mixture of AMPA and NMDA synapses, and a relatively smaller synaptic facilitation time constant (tau(f)) and larger tau(d) (with tau(f) smaller than tau(d) and tau(g)). These results are compatible with known physiology.

Phylogenetic comparative analysis of electric communication signals in ghost knifefishes (Gymnotiformes: Apteronotidae)

Electrocommunication signals in electric fish are diverse, easily recorded and have well-characterized neural control. Two signal features, the frequency and waveform of the electric organ discharge (EOD), vary widely across species. Modulations of the EOD (i.e. chirps and gradual frequency rises) also function as active communication signals during social interactions, but they have been studied in relatively few species. We compared the electrocommunication signals of 13 species in the largest gymnotiform family,Apteronotidae. Playback stimuli were used to elicit chirps and rises. We analyzed EOD frequency and waveform and the production and structure of chirps and rises. Species diversity in these signals was characterized with discriminant function analyses, and correlations between signal parameters were tested with phylogenetic comparative methods. Signals varied markedly across species and even between congeners and populations of the same species. Chirps and EODs were particularly evolutionarily labile, whereas rises differed little across species. Although all chirp parameters contributed to species differences in these signals, chirp amplitude modulation, frequency modulation (FM) and duration were particularly diverse. Within this diversity,however, interspecific correlations between chirp parameters suggest that mechanistic trade-offs may shape some aspects of signal evolution. In particular, a consistent trade-off between FM and EOD amplitude during chirps is likely to have influenced the evolution of chirp structure. These patterns suggest that functional or mechanistic linkages between signal parameters(e.g. the inability of electromotor neurons increase their firing rates without a loss of synchrony or amplitude of action potentials) constrain the evolution of signal structure.

Spike-Frequency Adapting Neural Ensembles: Beyond Mean Adaptation and Renewal Theories

We propose a Markov process model for spike-frequency adapting neural ensembles that synthesizes existing mean-adaptation approaches, population density methods, and inhomogeneous renewal theory, resulting in a unified and tractable framework that goes beyond renewal and mean-adaptation theories by accounting for correlations between subsequent interspike intervals. A method for efficiently generating inhomogeneous realizations of the proposed Markov process is given, numerical methods for solving the population equation are presented, and an expression for the first-order interspike interval correlation is derived. Further, we show that the full five-dimensional master equation for a conductance-based integrate-and-fire neuron with spike-frequency adaptation and a relative refractory mechanism driven by Poisson spike trains can be reduced to a two-dimensional generalization of the proposed Markov process by an adiabatic elimination of fast variables. For static and dynamic stimulation, negative serial interspike interval correlations and transient population responses, respectively, of Monte Carlo simulations of the full five-dimensional system can be accurately described by the proposed two-dimensional Markov process.

Stimulus frequency differentially affects chirping in two species of weakly electric fish: implications for the evolution of signal structure and function

During social interactions, apteronotid electric fish modulate their electric organ discharges (EODs) to produce transient communication signals known as chirps. Chirps vary widely across species and sex in both number and structure. In Apteronotus leptorhynchus, males chirp far more than females and their chirps have greater frequency modulation than those of females. High-frequency chirps are produced by males most often in response to female-like electric signals. As such, they have been hypothesized to function in courtship. The more common low-frequency chirps, produced by both males and females in response to same-sex signals, are hypothesized to function as aggressive signals. To determine whether the two chirp types in the closely related Apteronotus albifrons have similar functions, we stimulated chirping in male and female A. leptorhynchus and A. albifrons with playbacks simulating the EODs of same-sex versusopposite-sex conspecifics. As in A. leptorhynchus, male and female A. albifrons produced low-frequency chirps most often to same-sex signals. Unlike A. leptorhynchus, however, A. albifrons also produced more high-frequency chirps to same-sex stimuli than to opposite-sex stimuli. This suggests that high-frequency chirps in A. albifrons,unlike those in A. leptorhynchus, may not function as courtship signals and that the function of similar chirp types has diversified in Apteronotus. Examples such as this, in which the function of a communication signal has changed in closely related species, are rare. The electrocommunication signals of apteronotids may thus provide a remarkable opportunity to investigate the evolutionary interactions of signal structure and function.

Limits of linear rate coding of dynamic stimuli by electroreceptor afferents

We estimated the frequency-intensity (f-I) curves of P-unit electroreceptors using 4-Hz random amplitude modulations (RAMs) and using the covariance method (50-Hz RAMs). Both methods showed that P units are linear encoders of stimulus amplitude with additive noise; the gain of the f-I curve was, on average, 0.32 and 2.38 spikesxs(-1)xmicroV(-1) for the low- and high-frequency cutoffs, respectively. There were two sources of apparent noise in the encoding process: the first was the variability of baseline P-unit discharge and the second was the variation of receptor discharge due to variability of the stimulus slope independent of its intensity. The covariance method showed that a linear combination of eigenvectors representing the time-weighted stimulus intensity (E1) and its derivative (E2) could account for, on average, 92% of the total response variability; E1 by itself accounted for 76% of the variability. The low gain of the low-frequency f-I curve implies that detection of small (1 microV) signals would require integration over many receptors ( approximately 1,200) and time (200 ms); even then, signals that elicit behavioral responses could not be detected using rate coding with the estimated gain and noise levels. Weak signals at the limit of behavioral thresholds could be detected if the animal were able to extract E1 from the population of responding P units; we propose a tentative mechanism for this operation although there is no evidence as to whether it is actually implemented in the nervous system of these fish.

Neural strategies for optimal processing of sensory signals

The electrosensory system is used for both spatial navigation tasks and communication. An electric organ generates a sinusoidal electric field and cutaneous electroreceptors respond to this field. Objects such as prey or rocks cause a local low-frequency modulation of the electric field; this cue is used by electric fish for navigation and prey capture. The interference of the electric fields of conspecifics produces beats, often with high frequencies, that are also sensed by the electroreceptors; furthermore, these electric fish can transiently modulate their electric discharge as a communication signal. Thus these fish must therefore detect a variety of low-intensity signals that differ greatly in their spatial extent, frequency, and duration. Behavioral studies suggest that they are highly adapted to these tasks. Experimental and theoretical analyses of the neural circuitry for the electrosense has demonstrated many commonalities with the more common senses, e.g., topographic mapping and receptive fields with On or Off centers and surround inhibition. The integration of computational and experimental analyses has demonstrated novel mechanisms that appear to optimize weak signal detection in the electrosense including: noise shaping by correlations within single spike trains, induction of oscillations by delayed feedback inhibition, the requirement for maps with differing receptive field sizes tuned for different stimulus parameters, and the role of non-plastic feedback for adaptive cancellation of redundant signals. It is likely that these mechanisms will also be operative in other sensory systems.