Mixed Signal Learning by Spike Correlation Propagation in Feedback Inhibitory Circuits

The brain can learn and detect mixed input signals masked by various types of noise, and spike-timing-dependent plasticity (STDP) is the candidate synaptic level mechanism. Because sensory inputs typically have spike correlation, and local circuits have dense feedback connections, input spikes cause the propagation of spike correlation in lateral circuits; however, it is largely unknown how this secondary correlation generated by lateral circuits influences learning processes through STDP, or whether it is beneficial to achieve efficient spike-based learning from uncertain stimuli. To explore the answers to these questions, we construct models of feedforward networks with lateral inhibitory circuits and study how propagated correlation influences STDP learning, and what kind of learning algorithm such circuits achieve. We derive analytical conditions at which neurons detect minor signals with STDP, and show that depending on the origin of the noise, different correlation timescales are useful for learning. In particular, we show that non-precise spike correlation is beneficial for learning in the presence of cross-talk noise. We also show that by considering excitatory and inhibitory STDP at lateral connections, the circuit can acquire a lateral structure optimal for signal detection. In addition, we demonstrate that the model performs blind source separation in a manner similar to the sequential sampling approximation of the Bayesian independent component analysis algorithm. Our results provide a basic understanding of STDP learning in feedback circuits by integrating analyses from both dynamical systems and information theory.     

Learning shapes spontaneous activity itinerating over memorized states

Learning is a process that helps create neural dynamical systems so that an appropriate output pattern is generated for a given input. Often, such a memory is considered to be included in one of the attractors in neural dynamical systems, depending on the initial neural state specified by an input. Neither neural activities observed in the absence of inputs nor changes caused in the neural activity when an input is provided were studied extensively in the past. However, recent experimental studies have reported existence of structured spontaneous neural activity and its changes when an input is provided. With this background, we propose that memory recall occurs when the spontaneous neural activity changes to an appropriate output activity upon the application of an input, and this phenomenon is known as bifurcation in the dynamical systems theory. We introduce a reinforcement-learning-based layered neural network model with two synaptic time scales; in this network, I/O relations are successively memorized when the difference between the time scales is appropriate. After the learning process is complete, the neural dynamics are shaped so that it changes appropriately with each input. As the number of memorized patterns is increased, the generated spontaneous neural activity after learning shows itineration over the previously learned output patterns. This theoretical finding also shows remarkable agreement with recent experimental reports, where spontaneous neural activity in the visual cortex without stimuli itinerate over evoked patterns by previously applied signals. Our results suggest that itinerant spontaneous activity can be a natural outcome of successive learning of several patterns, and it facilitates bifurcation of the network when an input is provided.     

Exploring spike transfer through the thalamus using hybrid artificial-biological neuronal networks

We use dynamic clamp to construct "hybrid" thalamic circuits by connecting a biological neuron in situ to silicon- or software-generated "neurons" through artificial synapses. The purpose is to explore cellular sensory gating mechanisms that regulate the transfer efficiency of signals during different sleep-wake states. Hybrid technology is applied in vitro to different paradigms such as: (1) simulating interactions between biological thalamocortical neurons, artificial reticular thalamic inhibitory interneurons and a simulated sensory input, (2) grafting an artificial sensory input to a wholly biological thalamic network that generates spontaneous sleep-like oscillations, (3) injecting in thalamocortical neurons a background synaptic bombardment mimicking the activity of corticothalamic inputs. We show that the graded control of the strength of intrathalamic inhibition, combined with the membrane polarization and the fluctuating synaptic noise in thalamocortical neurons, is able to govern functional shifts between different input/output transmission states of the thalamic gate.     

An adaptive visual neuronal model implementing competitive, temporally asymmetric Hebbian learning

A novel depth-from-motion vision model based on leaky integrate-and-fire (I&F) neurons incorporates the implications of recent neurophysiological findings into an algorithm for object discovery and depth analysis. Pulse-coupled I&F neurons capture the edges in an optical flow field and the associated time of travel of those edges is encoded as the neuron parameters, mainly the time constant of the membrane potential and synaptic weight. Correlations between spikes and their timing thus code depth in the visual field. Neurons have multiple output synapses connecting to neighbouring neurons with an initial Gaussian weight distribution. A temporally asymmetric learning rule is used to adapt the synaptic weights online, during which competitive behaviour emerges between the different input synapses of a neuron. It is shown that the competition mechanism can further improve the model performance. After training, the weights of synapses sourced from a neuron do not display a Gaussian distribution, having adapted to encode features of the scenes to which they have been exposed.     

Adaptive filter model of the cerebellum

The Marr-Albus model of the cerebellum has been reformulated with linear system analysis. This adaptive linear filter model of the cerebellum performs a filtering action of a phase lead-lag compensator with learning capability, and will give an account for the phenomena which have been termed cerebellar compensation. It is postulated that a Golgi cell may act as a phase lag element; for example, as a leaky integrator with time constant about several seconds. Under this assumption, a mossy fiber-granule cell-Golgi cell input network functions as a phase lead-lag compensator. Output signals from Golgi-granule cell systems, namely, parallel fiber signals, are gathered together through variable synaptic connections to form a Purkinje cell output. From a general theory of adaptive linear filters, learning principles for these modifiable connections are derived. By these learning principles, a Purkinje cell output converges to the desired response to minimize the mean square error of the performance. In a more general sense, a Purkinje cell acquires a filtering function on the basis of multiple pairs of input signals and corresponding desired output signals. The mode of convergence of the output signal is described when the input signal is sinusoidal.     

Excitatory local circuits and their implications for olfactory processing in the fly antennal lobe

Conflicting views exist of how circuits of the antennal lobe, the insect equivalent of the olfactory bulb, translate input from olfactory receptor neurons (ORNs) into projection-neuron (PN) output. Synaptic connections between ORNs and PNs are one-to-one, yet PNs are more broadly tuned to odors than ORNs. The basis for this difference in receptive range remains unknown. Analyzing a Drosophila mutant lacking ORN input to one glomerulus, we show that some of the apparent complexity in the antennal lobe's output arises from lateral, interglomerular excitation of PNs. We describe a previously unidentified population of cholinergic local neurons (LNs) with multiglomerular processes. These excitatory LNs respond broadly to odors but exhibit little glomerular specificity in their synaptic output, suggesting that PNs are driven by a combination of glomerulus-specific ORN afferents and diffuse LN excitation. Lateral excitation may boost PN signals and enhance their transmission to third-order neurons in a mechanism akin to stochastic resonance.     

Interference in ballistic motor learning: specificity and role of sensory error signals

Humans are capable of learning numerous motor skills, but newly acquired skills may be abolished by subsequent learning. Here we ask what factors determine whether interference occurs in motor learning. We speculated that interference requires competing processes of synaptic plasticity in overlapping circuits and predicted specificity. To test this, subjects learned a ballistic motor task. Interference was observed following subsequent learning of an accuracy-tracking task, but only if the competing task involved the same muscles and movement direction. Interference was not observed from a non-learning task suggesting that interference requires competing learning. Subsequent learning of the competing task 4 h after initial learning did not cause interference suggesting disruption of early motor memory consolidation as one possible mechanism underlying interference. Repeated transcranial magnetic stimulation (rTMS) of corticospinal motor output at intensities below movement threshold did not cause interference, whereas suprathreshold rTMS evoking motor responses and (re)afferent activation did. Finally, the experiments revealed that suprathreshold repetitive electrical stimulation of the agonist (but not antagonist) peripheral nerve caused interference. The present study is, to our knowledge, the first to demonstrate that peripheral nerve stimulation may cause interference. The finding underscores the importance of sensory feedback as error signals in motor learning. We conclude that interference requires competing plasticity in overlapping circuits. Interference is remarkably specific for circuits involved in a specific movement and it may relate to sensory error signals.     

Temporal sub units in dendritic trees

This simulation study examines the possibility that dendritic sub units can be defined according to temporal aspects in the timing of populations of synaptic inputs. A two cell model with passive dendritic trees is used, which is subject to both common and independent synaptic inputs, the presence of common synaptic input results in a tendency for correlated firing in the two cell model. The strength of this correlation is used to measure the efficacy of the common synaptic inputs in modulating the output discharge of each neurone. Our results suggest that a small fraction of the total synaptic input can effectively modulate the timing of output spikes, this phenomenon is not dependent on the physical location of the inputs on the dendritic tree. This phenomenon depends on the presence of temporal correlation between the pre-synaptic spike trains that provide the common input. We propose to refer to these as temporal sub units.     

Isotropic sequence order learning using a novel linear algorithm in a closed loop behavioural system

In this article, we present an isotropic algorithm for sequence order learning. Its central goal is to learn the causal relation between two (or more) inputs in order to react to the earliest incoming signal after successful learning (like in typical classical conditioning situations). We implement this algorithm in a behaving system (a robot) thereby creating a closed loop situation where the learner's actions influence its own sensor inputs to the end of creating an autonomous agent. Autonomous behaviour implies that learning goals are internally defined within the organism's capabilities. Standard learning models for sequence learning (e.g. temporal difference (TD)-learning) need an externally defined reward. This, however, is in conflict with the requirement of an implicitly defined internal goal in autonomous behaviour. Therefore, in this study we present a system in which the external reward is replaced by a reflex loop. This loop explicitly includes the environment. Every reflex loop has the inherent disadvantage, which is that its re-actions occur each time just after a reflex-eliciting sensor event and thus 'too late'. However, a reflex can serve as the internal reference for sequence order learning, which has the task of eliminating this disadvantage by creating earlier anticipatory actions. In our system learning is achieved by modifying synaptic weights of a linear neuron with a correlation based learning rule which involves the derivative of the neuron's output. All input lines are entirely isotropic. The synaptic weight change curve of this rule is strongly related to the temporal Hebb learning rule, which was found in spike timing experiments. We find that after learning the reflex loop is replaced in functional terms with an earlier anticipatory action (and pathway). In addition, we observed that the synaptic weights stabilise as soon as the reflex remains silent.     

A Model of Self-Organizing Head-Centered Visual Responses in Primate Parietal Areas

We present a hypothesis for how head-centered visual representations in primate parietal areas could self-organize through visually-guided learning, and test this hypothesis using a neural network model. The model consists of a competitive output layer of neurons that receives afferent synaptic connections from a population of input neurons with eye position gain modulated retinal receptive fields. The synaptic connections in the model are trained with an associative trace learning rule which has the effect of encouraging output neurons to learn to respond to subsets of input patterns that tend to occur close together in time. This network architecture and synaptic learning rule is hypothesized to promote the development of head-centered output neurons during periods of time when the head remains fixed while the eyes move. This hypothesis is demonstrated to be feasible, and each of the core model components described is tested and found to be individually necessary for successful self-organization.     

Automatic Adaptation to Fast Input Changes in a Time-Invariant Neural Circuit

Neurons must faithfully encode signals that can vary over many orders of magnitude despite having only limited dynamic ranges. For a correlated signal, this dynamic range constraint can be relieved by subtracting away components of the signal that can be predicted from the past, a strategy known as predictive coding, that relies on learning the input statistics. However, the statistics of input natural signals can also vary over very short time scales e.g., following saccades across a visual scene. To maintain a reduced transmission cost to signals with rapidly varying statistics, neuronal circuits implementing predictive coding must also rapidly adapt their properties. Experimentally, in different sensory modalities, sensory neurons have shown such adaptations within 100 ms of an input change. Here, we show first that linear neurons connected in a feedback inhibitory circuit can implement predictive coding. We then show that adding a rectification nonlinearity to such a feedback inhibitory circuit allows it to automatically adapt and approximate the performance of an optimal linear predictive coding network, over a wide range of inputs, while keeping its underlying temporal and synaptic properties unchanged. We demonstrate that the resulting changes to the linearized temporal filters of this nonlinear network match the fast adaptations observed experimentally in different sensory modalities, in different vertebrate species. Therefore, the nonlinear feedback inhibitory network can provide automatic adaptation to fast varying signals, maintaining the dynamic range necessary for accurate neuronal transmission of natural inputs.     

Sex Differences in High‐fat Diet‐induced Obesity,Metabolic Alterations and Learning,and Synaptic Plasticity Deficits in Mice

Obesity is a potential risk factor for cognitive deficits in the elder humans. Using a high‐fat diet (HFD)–induced obese mouse model, we investigated the impacts of HFD on obesity, metabolic and stress hormones, learning performance, and hippocampal synaptic plasticity. Both male and female C57BL/6J mice fed with HFD (3 weeks to 9–12 months) gained significantly more weights than the sex‐specific control groups. Compared with the obese female mice, the obese males had similar energy intake but developed more weight gains. The obese male mice developed hyperglycemia, hyperinsulinemia, hypercholesterolemia, and hyperleptinemia, but not hypertriglyceridemia. The obese females had less hyperinsulinemia and hypercholesterolemia than the obese males, and no hyperglycemia and hypertriglyceridemia. In the contextual fear conditioning and step‐down passive avoidance tasks, the obese male, but not female, mice showed poorer learning performance than their normal counterparts. These learning deficits were not due to sensorimotor impairment as verified by the open‐field and hot‐plate tests. Although, basal synaptic transmission characteristics (input–output transfer and paired‐pulse facilitation (PPF) ratio) were not significantly different between normal and HFD groups, the magnitudes of synaptic plasticity (long‐term potentiation (LTP) and long‐term depression (LTD)) were lower at the Schaffer collateral‐CA1 synapses of the hippocampal slices isolated from the obese male, but not female, mice, as compared with their sex‐specific controls. Our results suggest that male mice are more vulnerable than the females to the impacts of HFD on weight gains, metabolic alterations and deficits of learning, and hippocampal synaptic plasticity.     

Social modulation of learned behavior by dopamine in the basal ganglia: Insights from songbirds

Dysfunction of the dopaminergic system leads to motor, cognitive, and motivational symptoms in brain disorders such as Parkinson’s disease. The basal ganglia (BG) are involved in sensorimotor learning and receive a strong dopaminergic signal, shown to play an important role in social interactions. The function of the dopaminergic input to the BG in the integration of social cues during sensorimotor learning remains however largely unexplored. Songbirds use learned vocalizations to communicate during courtship and aggressive behaviors. Like language learning in humans, song learning strongly depends on social interactions. In songbirds, a specialized BG–thalamo-cortical loop devoted to song is particularly tractable for elucidating the signals carried by dopamine in the BG, and the function of dopamine signaling in mediating social cues during skill learning and execution. Here, I review experimental findings uncovering the physiological effects and function of the dopaminergic signal in the songbird BG, in light of our knowledge of the BG–dopamine interactions in mammals. Interestingly, the compact nature of the striato-pallidal circuits in birds led to new insight on the physiological effects of the dopaminergic input on the BG network as a whole. In singing birds, D1-like receptor agonist and antagonist can modulate the spectral variability of syllables bi-directionally, suggesting that social context-dependent changes in spectral variability are triggered by dopaminergic input through D1-like receptors. As variability is crucial for exploration during motor learning, but must be reduced after learning to optimize performance, I propose that, the dopaminergic input to the BG could be responsible for the social-dependent regulation of the exploration/exploitation balance in birdsong, and possibly in learned skills in other vertebrates.     

The neural substrate of spectral preference in Drosophila

Drosophila vision is mediated by inputs from three types of photoreceptor neurons; R1-R6 mediate achromatic motion detection, while R7 and R8 constitute two chromatic channels. Neural circuits for processing chromatic information are not known. Here, we identified the first-order interneurons downstream of the chromatic channels. Serial EM revealed that small-field projection neurons Tm5 and Tm9 receive direct synaptic input from R7 and R8, respectively, and indirect input from R1-R6, qualifying them to function as color-opponent neurons. Wide-field Dm8 amacrine neurons receive input from 13-16 UV-sensing R7s and provide output to projection neurons. Using a combinatorial expression system to manipulate activity in different neuron subtypes, we determined that Dm8 neurons are necessary and sufficient for flies to exhibit phototaxis toward ultraviolet instead of green light. We propose that Dm8 sacrifices spatial resolution for sensitivity by relaying signals from multiple R7s to projection neurons, which then provide output to higher visual centers.     

Feedforward Inhibition and Synaptic Scaling – Two Sides of the Same Coin?

Feedforward inhibition and synaptic scaling are important adaptive processes that control the total input a neuron can receive from its afferents. While often studied in isolation, the two have been reported to co-occur in various brain regions. The functional implications of their interactions remain unclear, however. Based on a probabilistic modeling approach, we show here that fast feedforward inhibition and synaptic scaling interact synergistically during unsupervised learning. In technical terms, we model the input to a neural circuit using a normalized mixture model with Poisson noise. We demonstrate analytically and numerically that, in the presence of lateral inhibition introducing competition between different neurons, Hebbian plasticity and synaptic scaling approximate the optimal maximum likelihood solutions for this model. Our results suggest that, beyond its conventional use as a mechanism to remove undesired pattern variations, input normalization can make typical neural interaction and learning rules optimal on the stimulus subspace defined through feedforward inhibition. Furthermore, learning within this subspace is more efficient in practice, as it helps avoid locally optimal solutions. Our results suggest a close connection between feedforward inhibition and synaptic scaling which may have important functional implications for general cortical processing.     

Neural circuit flexibility in a small sensorimotor system

Neuronal circuits underlying rhythmic behaviors (central pattern generators: CPGs) can generate rhythmic motor output without sensory input. However, sensory input is pivotal for generating behaviorally relevant CPG output. Here we discuss recent work in the decapod crustacean stomatogastric nervous system (STNS) identifying cellular and synaptic mechanisms whereby sensory inputs select particular motor outputs from CPG circuits. This includes several examples in which sensory neurons regulate the impact of descending projection neurons on CPG circuits. This level of analysis is possible in the STNS due to the relatively unique access to identified circuit, projection, and sensory neurons. These studies are also revealing additional degrees of freedom in sensorimotor integration that underlie the extensive flexibility intrinsic to rhythmic motor systems.     

Synaptic modifications depend on synapse location and activity: a biophysical model of STDP

In spike-timing-dependent plasticity (STDP) the synapses are potentiated or depressed depending on the temporal order and temporal difference of the pre- and post-synaptic signals. We present a biophysical model of STDP which assumes that not only the timing, but also the shapes of these signals influence the synaptic modifications. The model is based on a Hebbian learning rule which correlates the NMDA synaptic conductance with the post-synaptic signal at synaptic location as the pre- and post-synaptic quantities. As compared to a previous paper [Saudargiene, A., Porr, B., Worgotter, F., 2004. How the shape of pre- and post-synaptic signals can influence stdp: a biophysical model. Neural Comp.], here we show that this rule reproduces the generic STDP weight change curve by using real neuronal input signals and combinations of more than two (pre- and post-synaptic) spikes. We demonstrate that the shape of the STDP curve strongly depends on the shape of the depolarising membrane potentials, which induces learning. As these potentials vary at different locations of the dendritic tree, model predicts that synaptic changes are location dependent. The model is extended to account for the patterns of more than two spikes of the pre- and post-synaptic cells. The results show that STDP weight change curve is also activity dependent.     

Songbird: a unique animal model for studying the molecular basis of disorders of vocal development and communication

Like humans, songbirds are one of the few animal groups that learn vocalization. Vocal learning requires coordination of auditory input and vocal output using auditory feedback to guide one’s own vocalizations during a specific developmental stage known as the critical period. Songbirds are good animal models for understand the neural basis of vocal learning, a complex form of imitation, because they have many parallels to humans with regard to the features of vocal behavior and neural circuits dedicated to vocal learning. In this review, we will summarize the behavioral, neural, and genetic traits of birdsong. We will also discuss how studies of birdsong can help us understand how the development of neural circuits for vocal learning and production is driven by sensory input (auditory information) and motor output (vocalization).     

Logarithmic amplifiers

Logarithmic amplifiers (log amps), which produce an output signal proportional to the logarithm of the input signal, are widely used in cytometry for measurements of parameters that vary over a wide dynamic range, e.g., cell surface immunofluorescence. Existing log amp circuits all deviate to some extent from ideal performance with respect to dynamic range and fidelity to the logarithmic curve; accuracy in quantitative analysis using log amps therefore requires that log amps be individually calibrated. However, accuracy and precision may be limited by photon statistics and system noise when very low level input signals are encountered.     

The chronotron: a neuron that learns to fire temporally precise spike patterns

In many cases, neurons process information carried by the precise timings of spikes. Here we show how neurons can learn to generate specific temporally precise output spikes in response to input patterns of spikes having precise timings, thus processing and memorizing information that is entirely temporally coded, both as input and as output. We introduce two new supervised learning rules for spiking neurons with temporal coding of information (chronotrons), one that provides high memory capacity (E-learning), and one that has a higher biological plausibility (I-learning). With I-learning, the neuron learns to fire the target spike trains through synaptic changes that are proportional to the synaptic currents at the timings of real and target output spikes. We study these learning rules in computer simulations where we train integrate-and-fire neurons. Both learning rules allow neurons to fire at the desired timings, with sub-millisecond precision. We show how chronotrons can learn to classify their inputs, by firing identical, temporally precise spike trains for different inputs belonging to the same class. When the input is noisy, the classification also leads to noise reduction. We compute lower bounds for the memory capacity of chronotrons and explore the influence of various parameters on chronotrons' performance. The chronotrons can model neurons that encode information in the time of the first spike relative to the onset of salient stimuli or neurons in oscillatory networks that encode information in the phases of spikes relative to the background oscillation. Our results show that firing one spike per cycle optimizes memory capacity in neurons encoding information in the phase of firing relative to a background rhythm.