Visualization of learning-induced synaptic plasticity in output neurons of the Drosophila mushroom body γ-lobe.

By learning, through experience, which stimuli coincide with dangers, it is possible to predict outcomes and act pre-emptively to ensure survival. In insects, this process is localized to the mushroom body (MB), the circuitry of which facilitates the coincident detection of sensory stimuli and punishing or rewarding cues and, downstream, the execution of appropriate learned behaviors. Here, we focused our attention on the mushroom body output neurons (MBONs) of the γ-lobes that act as downstream synaptic partners of the MB γ-Kenyon cells (KCs) to ask how the output of the MB γ-lobe is shaped by olfactory associative conditioning, distinguishing this from non-associative stimulus exposure effects, and without the influence of downstream modulation. This was achieved by employing a subcellularly localized calcium sensor to specifically monitor activity at MBON postsynaptic sites. Therein, we identified a robust associative modulation within only one MBON postsynaptic compartment (MBON-γ1pedc > α/ß), which displayed a suppressed postsynaptic response to an aversively paired odor. While this MBON did not undergo non-associative modulation, the reverse was true across the remainder of the γ-lobe, where general odor-evoked adaptation was observed, but no conditioned odor-specific modulation. In conclusion, associative synaptic plasticity underlying aversive olfactory learning is localized to one distinct synaptic γKC-to-γMBON connection.

Circuit and Cellular Mechanisms Facilitate the Transformation from Dense to Sparse Coding in the Insect Olfactory System.

Transformations between sensory representations are shaped by neural mechanisms at the cellular and the circuit level. In the insect olfactory system, the encoding of odor information undergoes a transition from a dense spatiotemporal population code in the antennal lobe to a sparse code in the mushroom body. However, the exact mechanisms shaping odor representations and their role in sensory processing are incompletely identified. Here, we investigate the transformation from dense to sparse odor representations in a spiking model of the insect olfactory system, focusing on two ubiquitous neural mechanisms: spike frequency adaptation at the cellular level and lateral inhibition at the circuit level. We find that cellular adaptation is essential for sparse representations in time (temporal sparseness), while lateral inhibition regulates sparseness in the neuronal space (population sparseness). The interplay of both mechanisms shapes spatiotemporal odor representations, which are optimized for the discrimination of odors during stimulus onset and offset. Response pattern correlation across different stimuli showed a nonmonotonic dependence on the strength of lateral inhibition with an optimum at intermediate levels, which is explained by two counteracting mechanisms. In addition, we find that odor identity is stored on a prolonged timescale in the adaptation levels but not in the spiking activity of the principal cells of the mushroom body, providing a testable hypothesis for the location of the so-called odor trace.

Winnerless competition in clustered balanced networks: inhibitory assemblies do the trick.

Balanced networks are a frequently employed basic model for neuronal networks in the mammalian neocortex. Large numbers of excitatory and inhibitory neurons are recurrently connected so that the numerous positive and negative inputs that each neuron receives cancel out on average. Neuronal firing is therefore driven by fluctuations in the input and resembles the irregular and asynchronous activity observed in cortical in vivo data. Recently, the balanced network model has been extended to accommodate clusters of strongly interconnected excitatory neurons in order to explain persistent activity in working memory-related tasks. This clustered topology introduces multistability and winnerless competition between attractors and can capture the high trial-to-trial variability and its reduction during stimulation that has been found experimentally. In this prospect article, we review the mean field description of balanced networks of binary neurons and apply the theory to clustered networks. We show that the stable fixed points of networks with clustered excitatory connectivity tend quickly towards firing rate saturation, which is generally inconsistent with experimental data. To remedy this shortcoming, we then present a novel perspective on networks with locally balanced clusters of both excitatory and inhibitory neuron populations. This approach allows for true multistability and moderate firing rates in activated clusters over a wide range of parameters. Our findings are supported by mean field theory and numerical network simulations. Finally, we discuss possible applications of the concept of joint excitatory and inhibitory clustering in future cortical network modelling studies.

Area-specific processing of cerebellar-thalamo-cortical information in primates.

The cerebellar-thalamo-cortical (CTC) system plays a major role in controlling timing and coordination of voluntary movements. However, the functional impact of this system on motor cortical sites has not been documented in a systematic manner. We addressed this question by implanting a chronic stimulating electrode in the superior cerebellar peduncle (SCP) and recording evoked multiunit activity (MUA) and the local field potential (LFP) in the primary motor cortex ([Formula: see text]), the premotor cortex ([Formula: see text]) and the somatosensory cortex ([Formula: see text]). The area-dependent response properties were estimated using the MUA response shape (quantified by decomposing into principal components) and the time-dependent frequency content of the evoked LFP. Each of these signals alone enabled good classification between the somatosensory and motor sites. Good classification between the primary motor and premotor areas could only be achieved when combining features from both signal types. Topographical single-site representation of the predicted class showed good recovery of functional organization. Finally, the probability for misclassification had a broad topographical organization. Despite the area-specific response features to SCP stimulation, there was considerable site-to-site variation in responses, specifically within the motor cortical areas. This indicates a substantial SCP impact on both the primary motor and premotor cortex. Given the documented involvement of these cortical areas in preparation and execution of movement, this result may suggest a CTC contribution to both motor execution and motor preparation. The stimulation responses in the somatosensory cortex were sparser and weaker. However, a functional role of the CTC system in somatosensory computation must be taken into consideration.

Neural correlates of side-specific odour memory in mushroom body output neurons.

Humans and other mammals as well as honeybees learn a unilateral association between an olfactory stimulus presented to one side and a reward. In all of them, the learned association can be behaviourally retrieved via contralateral stimulation, suggesting inter-hemispheric communication. However, the underlying neuronal circuits are largely unknown and neural correlates of across-brain-side plasticity have yet not been demonstrated. We report neural plasticity that reflects lateral integration after side-specific odour reward conditioning. Mushroom body output neurons that did not respond initially to contralateral olfactory stimulation developed a unique and stable representation of the rewarded compound stimulus (side and odour) predicting its value during memory retention. The encoding of the reward-associated compound stimulus is delayed by about 40 ms compared with unrewarded neural activity, indicating an increased computation time for the read-out after lateral integration.

Rapid learning dynamics in individual honeybees during classical conditioning.

Associative learning in insects has been studied extensively by a multitude of classical conditioning protocols. However, so far little emphasis has been put on the dynamics of learning in individuals. The honeybee is a well-established animal model for learning and memory. We here studied associative learning as expressed in individual behavior based on a large collection of data on olfactory classical conditioning (25 datasets, 3298 animals). We show that the group-averaged learning curve and memory retention score confound three attributes of individual learning: the ability or inability to learn a given task, the generally fast acquisition of a conditioned response (CR) in learners, and the high stability of the CR during consecutive training and memory retention trials. We reassessed the prevailing view that more training results in better memory performance and found that 24 h memory retention can be indistinguishable after single-trial and multiple-trial conditioning in individuals. We explain how inter-individual differences in learning can be accommodated within the Rescorla-Wagner theory of associative learning. In both data-analysis and modeling we demonstrate how the conflict between population-level and single-animal perspectives on learning and memory can be disentangled.

Neural representation of calling songs and their behavioral relevance in the grasshopper auditory system.

Acoustic communication plays a key role for mate attraction in grasshoppers. Males use songs to advertise themselves to females. Females evaluate the song pattern, a repetitive structure of sound syllables separated by short pauses, to recognize a conspecific male and as proxy to its fitness. In their natural habitat females often receive songs with degraded temporal structure. Perturbations may, for example, result from the overlap with other songs. We studied the response behavior of females to songs that show different signal degradations. A perturbation of an otherwise attractive song at later positions in the syllable diminished the behavioral response, whereas the same perturbation at the onset of a syllable did not affect song attractiveness. We applied naïve Bayes classifiers to the spike trains of identified neurons in the auditory pathway to explore how sensory evidence about the acoustic stimulus and its attractiveness is represented in the neuronal responses. We find that populations of three or more neurons were sufficient to reliably decode the acoustic stimulus and to predict its behavioral relevance from the single-trial integrated firing rate. A simple model of decision making simulates the female response behavior. It computes for each syllable the likelihood for the presence of an attractive song pattern as evidenced by the population firing rate. Integration across syllables allows the likelihood to reach a decision threshold and to elicit the behavioral response. The close match between model performance and animal behavior shows that a spike rate code is sufficient to enable song pattern recognition.

Cellular adaptation facilitates sparse and reliable coding in sensory pathways.

Most neurons in peripheral sensory pathways initially respond vigorously when a preferred stimulus is presented, but adapt as stimulation continues. It is unclear how this phenomenon affects stimulus coding in the later stages of sensory processing. Here, we show that a temporally sparse and reliable stimulus representation develops naturally in sequential stages of a sensory network with adapting neurons. As a modeling framework we employ a mean-field approach together with an adaptive population density treatment, accompanied by numerical simulations of spiking neural networks. We find that cellular adaptation plays a critical role in the dynamic reduction of the trial-by-trial variability of cortical spike responses by transiently suppressing self-generated fast fluctuations in the cortical balanced network. This provides an explanation for a widespread cortical phenomenon by a simple mechanism. We further show that in the insect olfactory system cellular adaptation is sufficient to explain the emergence of the temporally sparse and reliable stimulus representation in the mushroom body. Our results reveal a generic, biophysically plausible mechanism that can explain the emergence of a temporally sparse and reliable stimulus representation within a sequential processing architecture.

Natural image sequences constrain dynamic receptive fields and imply a sparse code.

In their natural environment, animals experience a complex and dynamic visual scenery. Under such natural stimulus conditions, neurons in the visual cortex employ a spatially and temporally sparse code. For the input scenario of natural still images, previous work demonstrated that unsupervised feature learning combined with the constraint of sparse coding can predict physiologically measured receptive fields of simple cells in the primary visual cortex. This convincingly indicated that the mammalian visual system is adapted to the natural spatial input statistics. Here, we extend this approach to the time domain in order to predict dynamic receptive fields that can account for both spatial and temporal sparse activation in biological neurons. We rely on temporal restricted Boltzmann machines and suggest a novel temporal autoencoding training procedure. When tested on a dynamic multi-variate benchmark dataset this method outperformed existing models of this class. Learning features on a large dataset of natural movies allowed us to model spatio-temporal receptive fields for single neurons. They resemble temporally smooth transformations of previously obtained static receptive fields and are thus consistent with existing theories. A neuronal spike response model demonstrates how the dynamic receptive field facilitates temporal and population sparseness. We discuss the potential mechanisms and benefits of a spatially and temporally sparse representation of natural visual input.

Critical song features for auditory pattern recognition in crickets.

Many different invertebrate and vertebrate species use acoustic communication for pair formation. In the cricket Gryllus bimaculatus, females recognize their species-specific calling song and localize singing males by positive phonotaxis. The song pattern of males has a clear structure consisting of brief and regular pulses that are grouped into repetitive chirps. Information is thus present on a short and a long time scale. Here, we ask which structural features of the song critically determine the phonotactic performance. To this end we employed artificial neural networks to analyze a large body of behavioral data that measured females' phonotactic behavior under systematic variation of artificially generated song patterns. In a first step we used four non-redundant descriptive temporal features to predict the female response. The model prediction showed a high correlation with the experimental results. We used this behavioral model to explore the integration of the two different time scales. Our result suggested that only an attractive pulse structure in combination with an attractive chirp structure reliably induced phonotactic behavior to signals. In a further step we investigated all feature sets, each one consisting of a different combination of eight proposed temporal features. We identified feature sets of size two, three, and four that achieve highest prediction power by using the pulse period from the short time scale plus additional information from the long time scale.

Parallel processing via a dual olfactory pathway in the honeybee.

In their natural environment, animals face complex and highly dynamic olfactory input. Thus vertebrates as well as invertebrates require fast and reliable processing of olfactory information. Parallel processing has been shown to improve processing speed and power in other sensory systems and is characterized by extraction of different stimulus parameters along parallel sensory information streams. Honeybees possess an elaborate olfactory system with unique neuronal architecture: a dual olfactory pathway comprising a medial projection-neuron (PN) antennal lobe (AL) protocerebral output tract (m-APT) and a lateral PN AL output tract (l-APT) connecting the olfactory lobes with higher-order brain centers. We asked whether this neuronal architecture serves parallel processing and employed a novel technique for simultaneous multiunit recordings from both tracts. The results revealed response profiles from a high number of PNs of both tracts to floral, pheromonal, and biologically relevant odor mixtures tested over multiple trials. PNs from both tracts responded to all tested odors, but with different characteristics indicating parallel processing of similar odors. Both PN tracts were activated by widely overlapping response profiles, which is a requirement for parallel processing. The l-APT PNs had broad response profiles suggesting generalized coding properties, whereas the responses of m-APT PNs were comparatively weaker and less frequent, indicating higher odor specificity. Comparison of response latencies within and across tracts revealed odor-dependent latencies. We suggest that parallel processing via the honeybee dual olfactory pathway provides enhanced odor processing capabilities serving sophisticated odor perception and olfactory demands associated with a complex olfactory world of this social insect.

Viewing strategy of Cebus monkeys during free exploration of natural images.

Humans and other primates move their eyes several times per second to foveate at different locations of a visual scene. What features of a scene guide eye movements in natural vision? We recorded eye movements of three monkeys during free exploration of natural scenes and propose a simple model to explain their dynamics. We use the spatial clustering of fixation positions to define the monkeys' subjective regions-of-interest (ROI) in natural scenes. For most images the subjective ROIs match significantly the computed saliency of the natural scene, except when the image contains human or primate faces. We also investigated the temporal sequence of eye movements by computing the probability that a fixation will be made inside or outside of the ROI, given the current fixation position. We fitted a Markov chain model to the sequence of fixation positions, and find that fixations made inside a ROI are more likely to be followed by another fixation in the same ROI. This is true, independent of the image saliency in the area of the ROI. Our results show that certain regions in a natural scene are explored locally before directing the focus to another local region. This strategy could allow for quick integration of the visual features that constitute an object, and efficient segmentation of objects from other objects and the background during free viewing of natural scenes.

Stereotypical spatiotemporal activity patterns during slow-wave activity in the neocortex.

Alternating epochs of activity and silence are a characteristic feature of neocortical networks during certain sleep cycles and deep states of anesthesia. The mechanism and functional role of these slow oscillations (<1 Hz) have not yet been fully characterized. Experimental and theoretical studies show that slow-wave oscillations can be generated autonomously by neocortical tissue but become more regular through a thalamo-cortical feedback loop. Evidence for a functional role of slow-wave activity comes from EEG recordings in humans during sleep, which show that activity travels as stereotypical waves over the entire brain, thought to play a role in memory consolidation. We used an animal model to investigate activity wave propagation on a smaller scale, namely within the rat somatosensory cortex. Signals from multiple extracellular microelectrodes in combination with one intracellular recording in the anesthetized animal in vivo were utilized to monitor the spreading of activity. We found that activity propagation in most animals showed a clear preferred direction, suggesting that it often originated from a similar location in the cortex. In addition, the breakdown of active states followed a similar pattern with slightly weaker direction preference but a clear correlation to the direction of activity spreading, supporting the notion of a wave-like phenomenon similar to that observed after strong sensory stimulation in sensory areas. Taken together, our findings support the idea that activity waves during slow-wave sleep do not occur spontaneously at random locations within the network, as was suggested previously, but follow preferred synaptic pathways on a small spatial scale.

Adaptation reduces variability of the neuronal population code.

Sequences of events in noise-driven excitable systems with slow variables often show serial correlations among their intervals of events. Here, we employ a master equation for generalized non-renewal processes to calculate the interval and count statistics of superimposed processes governed by a slow adaptation variable. For an ensemble of neurons with spike-frequency adaptation, this results in the regularization of the population activity and an enhanced postsynaptic signal decoding. We confirm our theoretical results in a population of cortical neurons recorded in vivo.

Beyond the cortical column: abundance and physiology of horizontal connections imply a strong role for inputs from the surround.

Current concepts of cortical information processing and most cortical network models largely rest on the assumption that well-studied properties of local synaptic connectivity are sufficient to understand the generic properties of cortical networks. This view seems to be justified by the observation that the vertical connectivity within local volumes is strong, whereas horizontally, the connection probability between pairs of neurons drops sharply with distance. Recent neuroanatomical studies, however, have emphasized that a substantial fraction of synapses onto neocortical pyramidal neurons stems from cells outside the local volume. Here, we discuss recent findings on the signal integration from horizontal inputs, showing that they could serve as a substrate for reliable and temporally precise signal propagation. Quantification of connection probabilities and parameters of synaptic physiology as a function of lateral distance indicates that horizontal projections constitute a considerable fraction, if not the majority, of inputs from within the cortical network. Taking these non-local horizontal inputs into account may dramatically change our current view on cortical information processing.

Dynamic encoding of movement direction in motor cortical neurons.

When we perform a skilled movement such as reaching for an object, we can make use of prior information, for example about the location of the object in space. This helps us to prepare the movement, and we gain improved accuracy and speed during movement execution. Here, we investigate how prior information affects the motor cortical representation of movements during preparation and execution. We trained two monkeys in a delayed reaching task and provided a varying degree of prior information about the final target location. We decoded movement direction from multiple single-unit activity recorded from M1 (primary motor cortex) in one monkey and from PMd (dorsal premotor cortex) in a second monkey. Our results demonstrate that motor cortical cells in both areas exhibit individual encoding characteristics that change dynamically in time and dependent on prior information. On the population level, the information about movement direction is at any point in time accurately represented in a neuronal ensemble of time-varying composition. We conclude that movement representation in the motor cortex is not a static one, but one in which neurons dynamically allocate their computational resources to meet the demands defined by the movement task and the context of the movement. Consequently, we find that the decoding accuracy decreases if the precise task time, or the previous information that was available to the monkey, were disregarded in the decoding process. An optimal strategy for the readout of movement parameters from motor cortex should therefore take into account time and contextual parameters.

Serial correlation in neural spike trains: experimental evidence, stochastic modeling, and single neuron variability.

The activity of spiking neurons is frequently described by renewal point process models that assume the statistical independence and identical distribution of the intervals between action potentials. However, the assumption of independent intervals must be questioned for many different types of neurons. We review experimental studies that reported the feature of a negative serial correlation of neighboring intervals, commonly observed in neurons in the sensory periphery as well as in central neurons, notably in the mammalian cortex. In our experiments we observed the same short-lived negative serial dependence of intervals in the spontaneous activity of mushroom body extrinsic neurons in the honeybee. To model serial interval correlations of arbitrary lags, we suggest a family of autoregressive point processes. Its marginal interval distribution is described by the generalized gamma model, which includes as special cases the log-normal and gamma distributions, which have been widely used to characterize regular spiking neurons. In numeric simulations we investigated how serial correlation affects the variance of the neural spike count. We show that the experimentally confirmed negative correlation reduces single-neuron variability, as quantified by the Fano factor, by up to 50%, which favors the transmission of a rate code. We argue that the feature of a negative serial correlation is likely to be common to the class of spike-frequency-adapting neurons and that it might have been largely overlooked in extracellular single-unit recordings due to spike sorting errors.

FIND--a unified framework for neural data analysis.

The complexity of neurophysiology data has increased tremendously over the last years, especially due to the widespread availability of multi-channel recording techniques. With adequate computing power the current limit for computational neuroscience is the effort and time it takes for scientists to translate their ideas into working code. Advanced analysis methods are complex and often lack reproducibility on the basis of published descriptions. To overcome this limitation we develop FIND (Finding Information in Neural Data) as a platform-independent, open source framework for the analysis of neuronal activity data based on Matlab (Mathworks). Here, we outline the structure of the FIND framework and describe its functionality, our measures of quality control, and the policies for developers and users. Within FIND we have developed a unified data import from various proprietary formats, simplifying standardized interfacing with tools for analysis and simulation. The toolbox FIND covers a steadily increasing number of tools. These analysis tools address various types of neural activity data, including discrete series of spike events, continuous time series and imaging data. Additionally, the toolbox provides solutions for the simulation of parallel stochastic point processes to model multi-channel spiking activity. We illustrate two examples of complex analyses with FIND tools: First, we present a time-resolved characterization of the spiking irregularity in an in vivo extracellular recording from a mushroom-body extrinsic neuron in the honeybee during odor stimulation. Second, we describe layer specific input dynamics in the rat primary visual cortex in vivo in response to visual flash stimulation on the basis of multi-channel spiking activity.

G-Node: an integrated tool-sharing platform to support cellular and systems neurophysiology in the age of global neuroinformatics.

The global scale of neuroinformatics offers unprecedented opportunities for scientific collaborations between and among experimental and theoretical neuroscientists. To fully harvest these possibilities, a set of coordinated activities is required that will improve three key ingredients of neuroscientific research: data access, data storage, and data analysis, together with supporting activities for teaching and training. Focusing on the development of tools aiming at neurophysiological data, the newly established German Neuroinformatics Node (G-Node) aims at addressing these aspects as part of the International Neuroinformatics Coordination Facility (INCF). Based on its technical and scientific scope, the Node could play a substantial role for cellular and systems neurophysiology as well as for the neuroscience community at large.