A Simple Hebbian/Anti-Hebbian Network Learns the Sparse, Independent Components of Natural Images

Performance of biologically grounded models of the early visual system on standard object recognition tasks

Computational neuroscience models of vision and neural network models for object recognition are often framed by different research agendas. Computational neuroscience mainly aims at replicating experimental data, while (artificial) neural networks target high performance on classification tasks. However, we propose that models of vision should be validated on object recognition tasks. At some point, mechanisms of realistic neuro-computational models of the visual cortex have to convince in object recognition as well. In order to foster this idea, we report the recognition accuracy for two different neuro-computational models of the visual cortex on several object recognition datasets. The models were trained using unsupervised Hebbian learning rules on natural scene inputs for the emergence of receptive fields comparable to their biological counterpart. We assume that the emerged receptive fields result in a general codebook of features, which should be applicable to a variety of visual scenes. We report the performances on datasets with different levels of difficulty, ranging from the simple MNIST to the more complex CIFAR-10 or ETH-80. We found that both networks show good results on simple digit recognition, comparable with previously published biologically plausible models. We also observed that our deeper layer neurons provide for naturalistic datasets a better recognition codebook. As for most datasets, recognition results of biologically grounded models are not available yet, our results provide a broad basis of performance values to compare methodologically similar models.

Exploitation of image statistics with sparse coding in the case of stereo vision

The sparse coding algorithm has served as a model for early processing in mammalian vision. It has been assumed that the brain uses sparse coding to exploit statistical properties of the sensory stream. We hypothesize that sparse coding discovers patterns from the data set, which can be used to estimate a set of stimulus parameters by simple readout. In this study, we chose a model of stereo vision to test our hypothesis. We used the Locally Competitive Algorithm (LCA), followed by a naïve Bayes classifier, to infer stereo disparity. From the results we report three observations. First, disparity inference was successful with this naturalistic processing pipeline. Second, an expanded, highly redundant representation is required to robustly identify the input patterns. Third, the inference error can be predicted from the number of active coefficients in the LCA representation. We conclude that sparse coding can generate a suitable general representation for subsequent inference tasks.

Sparse coding with a somato-dendritic rule

Cortical neurons are silent most of the time: sparse activity enables low-energy computation in the brain, and promises to do the same in neuromorphic hardware. Beyond power efficiency, sparse codes have favourable properties for associative learning, as they can store more information than local codes but are easier to read out than dense codes. Auto-encoders with a sparse constraint can learn sparse codes, and so can single-layer networks that combine recurrent inhibition with unsupervised Hebbian learning. But the latter usually require fast homeostatic plasticity, which could lead to catastrophic forgetting in embodied agents that learn continuously. Here we set out to explore whether plasticity at recurrent inhibitory synapses could take up that role instead, regulating both the population sparseness and the firing rates of individual neurons. We put the idea to the test in a network that employs compartmentalised inputs to solve the task: rate-based dendritic compartments integrate the feedforward input, while spiking integrate-and-fire somas compete through recurrent inhibition. A somato-dendritic learning rule allows somatic inhibition to modulate nonlinear Hebbian learning in the dendrites. Trained on MNIST digits and natural images, the network discovers independent components that form a sparse encoding of the input and support linear decoding. These findings confirm that intrinsic homeostatic plasticity is not strictly required for regulating sparseness: inhibitory synaptic plasticity can have the same effect. Our work illustrates the usefulness of compartmentalised inputs, and makes the case for moving beyond point neuron models in artificial spiking neural networks.

A Gestalt inference model for auditory scene segregation

Our current understanding of how the brain segregates auditory scenes into meaningful objects is in line with a Gestaltism framework. These Gestalt principles suggest a theory of how different attributes of the soundscape are extracted then bound together into separate groups that reflect different objects or streams present in the scene. These cues are thought to reflect the underlying statistical structure of natural sounds in a similar way that statistics of natural images are closely linked to the principles that guide figure-ground segregation and object segmentation in vision. In the present study, we leverage inference in stochastic neural networks to learn emergent grouping cues directly from natural soundscapes including speech, music and sounds in nature. The model learns a hierarchy of local and global spectro-temporal attributes reminiscent of simultaneous and sequential Gestalt cues that underlie the organization of auditory scenes. These mappings operate at multiple time scales to analyze an incoming complex scene and are then fused using a Hebbian network that binds together coherent features into perceptually-segregated auditory objects. The proposed architecture successfully emulates a wide range of well established auditory scene segregation phenomena and quantifies the complimentary role of segregation and binding cues in driving auditory scene segregation.

Slow feature analysis with spiking neurons and its application to audio stimuli

Extracting invariant features in an unsupervised manner is crucial to perform complex computation such as object recognition, analyzing music or understanding speech. While various algorithms have been proposed to perform such a task, Slow Feature Analysis (SFA) uses time as a means of detecting those invariants, extracting the slowly time-varying components in the input signals. In this work, we address the question of how such an algorithm can be implemented by neurons, and apply it in the context of audio stimuli. We propose a projected gradient implementation of SFA that can be adapted to a Hebbian like learning rule dealing with biologically plausible neuron models. Furthermore, we show that a Spike-Timing Dependent Plasticity learning rule, shaped as a smoothed second derivative, implements SFA for spiking neurons. The theory is supported by numerical simulations, and to illustrate a simple use of SFA, we have applied it to auditory signals. We show that a single SFA neuron can learn to extract the tempo in sound recordings.

Mirrored STDP Implements Autoencoder Learning in a Network of Spiking Neurons

The autoencoder algorithm is a simple but powerful unsupervised method for training neural networks. Autoencoder networks can learn sparse distributed codes similar to those seen in cortical sensory areas such as visual area V1, but they can also be stacked to learn increasingly abstract representations. Several computational neuroscience models of sensory areas, including Olshausen & Field’s Sparse Coding algorithm, can be seen as autoencoder variants, and autoencoders have seen extensive use in the machine learning community. Despite their power and versatility, autoencoders have been difficult to implement in a biologically realistic fashion. The challenges include their need to calculate differences between two neuronal activities and their requirement for learning rules which lead to identical changes at feedforward and feedback connections. Here, we study a biologically realistic network of integrate-and-fire neurons with anatomical connectivity and synaptic plasticity that closely matches that observed in cortical sensory areas. Our choice of synaptic plasticity rules is inspired by recent experimental and theoretical results suggesting that learning at feedback connections may have a different form from learning at feedforward connections, and our results depend critically on this novel choice of plasticity rules. Specifically, we propose that plasticity rules at feedforward versus feedback connections are temporally opposed versions of spike-timing dependent plasticity (STDP), leading to a symmetric combined rule we call Mirrored STDP (mSTDP). We show that with mSTDP, our network follows a learning rule that approximately minimizes an autoencoder loss function. When trained with whitened natural image patches, the learned synaptic weights resemble the receptive fields seen in V1. Our results use realistic synaptic plasticity rules to show that the powerful autoencoder learning algorithm could be within the reach of real biological networks.

In the brain areas responsible for sensory processing, neurons learn over time to respond to specific features in the external world. Here, we propose a new, biologically plausible model for how groups of neurons can learn which specific features to respond to. Our work connects theoretical arguments about the optimal forms of neuronal representations with experimental results showing how synaptic connections change in response to neuronal activity. Specifically, we show that biologically realistic neurons can implement an algorithm known as autoencoder learning, in which the neurons learn to form representations that can be used to reconstruct their inputs. Autoencoder networks can successfully model neuronal responses in early sensory areas, and they are also frequently used in machine learning for training deep neural networks. Despite their power and utility, autoencoder networks have not been previously implemented in a fully biological fashion. To perform the autoencoder algorithm, neurons must modify their incoming, feedforward synaptic connections as well as their outgoing, feedback synaptic connections—and the changes to both must depend on the errors the network makes when it tries to reconstruct its input. Here, we propose a model for activity in the network and show that the commonly used spike-timing-dependent plasticity paradigm will implement the desired changes to feedforward synaptic connection weights. Critically, we use recent experimental evidence to propose that feedback connections learn according to a temporally reversed plasticity rule. We show mathematically that the two rules combined can approximately implement autoencoder learning, and confirm our results using simulated networks of integrate-and-fire neurons. By showing that biological neurons can implement this powerful algorithm, our work opens the door for the modeling of many learning paradigms from both the fields of computational neuroscience and machine learning.

Competition improves robustness against loss of information

A substantial number of works have aimed at modeling the receptive field properties of the primary visual cortex (V1). Their evaluation criterion is usually the similarity of the model response properties to the recorded responses from biological organisms. However, as several algorithms were able to demonstrate some degree of similarity to biological data based on the existing criteria, we focus on the robustness against loss of information in the form of occlusions as an additional constraint for better understanding the algorithmic level of early vision in the brain. We try to investigate the influence of competition mechanisms on the robustness. Therefore, we compared four methods employing different competition mechanisms, namely, independent component analysis, non-negative matrix factorization with sparseness constraint, predictive coding/biased competition, and a Hebbian neural network with lateral inhibitory connections. Each of those methods is known to be capable of developing receptive fields comparable to those of V1 simple-cells. Since measuring the robustness of methods having simple-cell like receptive fields against occlusion is difficult, we measure the robustness using the classification accuracy on the MNIST hand written digit dataset. For this we trained all methods on the training set of the MNIST hand written digits dataset and tested them on a MNIST test set with different levels of occlusions. We observe that methods which employ competitive mechanisms have higher robustness against loss of information. Also the kind of the competition mechanisms plays an important role in robustness. Global feedback inhibition as employed in predictive coding/biased competition has an advantage compared to local lateral inhibition learned by an anti-Hebb rule.

Deep learning in neural networks: an overview

In recent years, deep artificial neural networks (including recurrent ones) have won numerous contests in pattern recognition and machine learning. This historical survey compactly summarizes relevant work, much of it from the previous millennium. Shallow and Deep Learners are distinguished by the depth of their credit assignment paths, which are chains of possibly learnable, causal links between actions and effects. I review deep supervised learning (also recapitulating the history of backpropagation), unsupervised learning, reinforcement learning & evolutionary computation, and indirect search for short programs encoding deep and large networks.

Classification using sparse representations: a biologically plausible approach

Representing signals as linear combinations of basis vectors sparsely selected from an overcomplete dictionary has proven to be advantageous for many applications in pattern recognition, machine learning, signal processing, and computer vision. While this approach was originally inspired by insights into cortical information processing, biologically plausible approaches have been limited to exploring the functionality of early sensory processing in the brain, while more practical applications have employed non-biologically plausible sparse coding algorithms. Here, a biologically plausible algorithm is proposed that can be applied to practical problems. This algorithm is evaluated using standard benchmark tasks in the domain of pattern classification, and its performance is compared to a wide range of alternative algorithms that are widely used in signal and image processing. The results show that for the classification tasks performed here, the proposed method is competitive with the best of the alternative algorithms that have been evaluated. This demonstrates that classification using sparse representations can be performed in a neurally plausible manner, and hence, that this mechanism of classification might be exploited by the brain.

Finding a roadmap to achieve large neuromorphic hardware systems

Neuromorphic systems are gaining increasing importance in an era where CMOS digital computing techniques are reaching physical limits. These silicon systems mimic extremely energy efficient neural computing structures, potentially both for solving engineering applications as well as understanding neural computation. Toward this end, the authors provide a glimpse at what the technology evolution roadmap looks like for these systems so that Neuromorphic engineers may gain the same benefit of anticipation and foresight that IC designers gained from Moore's law many years ago. Scaling of energy efficiency, performance, and size will be discussed as well as how the implementation and application space of Neuromorphic systems are expected to evolve over time.

Physiologically inspired model for the visual recognition of transitive hand actions

The visual recognition of actions is an important visual function that is critical for motor learning and social communication. Action-selective neurons have been found in different cortical regions, including the superior temporal sulcus, parietal and premotor cortex. Among those are mirror neurons, which link visual and motor representations of body movements. While numerous theoretical models for the mirror neuron system have been proposed, the computational basis of the visual processing of goal-directed actions remains largely unclear. While most existing models focus on the possible role of motor representations in action recognition, we propose a model showing that many critical properties of action-selective visual neurons can be accounted for by well-established visual mechanisms. Our model accomplishes the recognition of hand actions from real video stimuli, exploiting exclusively mechanisms that can be implemented in a biologically plausible way by cortical neurons. We show that the model provides a unifying quantitatively consistent account of a variety of electrophysiological results from action-selective visual neurons. In addition, it makes a number of predictions, some of which could be confirmed in recent electrophysiological experiments.

Image segmentation using a sparse coding model of cortical area V1

Algorithms that encode images using a sparse set of basis functions have previously been shown to explain aspects of the physiology of a primary visual cortex (V1), and have been used for applications, such as image compression, restoration, and classification. Here, a sparse coding algorithm, that has previously been used to account for the response properties of orientation tuned cells in primary visual cortex, is applied to the task of perceptually salient boundary detection. The proposed algorithm is currently limited to using only intensity information at a single scale. However, it is shown to out-perform the current state-of-the-art image segmentation method (Pb) when this method is also restricted to using the same information.

Embedding responses in spontaneous neural activity shaped through sequential learning

Recent experimental measurements have demonstrated that spontaneous neural activity in the absence of explicit external stimuli has remarkable spatiotemporal structure. This spontaneous activity has also been shown to play a key role in the response to external stimuli. To better understand this role, we proposed a viewpoint, “memories-as-bifurcations,” that differs from the traditional “memories-as-attractors” viewpoint. Memory recall from the memories-as-bifurcations viewpoint occurs when the spontaneous neural activity is changed to an appropriate output activity upon application of an input, known as a bifurcation in dynamical systems theory, wherein the input modifies the flow structure of the neural dynamics. Learning, then, is a process that helps create neural dynamical systems such that a target output pattern is generated as an attractor upon a given input. Based on this novel viewpoint, we introduce in this paper an associative memory model with a sequential learning process. Using a simple Hebbian-type learning, the model is able to memorize a large number of input/output mappings. The neural dynamics shaped through the learning exhibit different bifurcations to make the requested targets stable upon an increase in the input, and the neural activity in the absence of input shows chaotic dynamics with occasional approaches to the memorized target patterns. These results suggest that these dynamics facilitate the bifurcations to each target attractor upon application of the corresponding input, which thus increases the capacity for learning. This theoretical finding about the behavior of the spontaneous neural activity is consistent with recent experimental observations in which the neural activity without stimuli wanders among patterns evoked by previously applied signals. In addition, the neural networks shaped by learning properly reflect the correlations of input and target-output patterns in a similar manner to those designed in our previous study.

The neural activity without explicit stimuli shows highly structured patterns in space and time, known as spontaneous activity. This spontaneous activity plays a key role in the behavior of the response to external stimuli generated by the interplay between the spontaneous activity and external input. Studying this interplay and how it is shaped by learning is an essential step toward understanding the principles of neural processing. To address this, we proposed a novel viewpoint, memories-as-bifurcations, in which the appropriate changes in the activity upon the input are embedded through learning. Based on this viewpoint, we introduce here an associative memory model with sequential learning by a simple Hebbian-type rule. In spite of its simplicity, the model memorizes the input/output mappings successively, as long as the input is sufficiently large and the synaptic change is slow. The spontaneous neural activity shaped after learning is shown to itinerate over the memorized targets in remarkable agreement with the experimental reports. These dynamics may prepare and facilitate to generate the learned response to the input. Our results suggest that this is the possible functional role of the spontaneous neural activity, while the uncovered network structure inspires a design principle for the memories-as-bifurcations.

Replicating receptive fields of simple and complex cells in primary visual cortex in a neuronal network model with temporal and population sparseness and reliability

We propose a new principle for replicating receptive field properties of neurons in the primary visual cortex. We derive a learning rule for a feedforward network, which maintains a low firing rate for the output neurons (resulting in temporal sparseness) and allows only a small subset of the neurons in the network to fire at any given time (resulting in population sparseness). Our learning rule also sets the firing rates of the output neurons at each time step to near-maximum or near-minimum levels, resulting in neuronal reliability. The learning rule is simple enough to be written in spatially and temporally local forms. After the learning stage is performed using input image patches of natural scenes, output neurons in the model network are found to exhibit simple-cell-like receptive field properties. When the output of these simple-cell-like neurons are input to another model layer using the same learning rule, the second-layer output neurons after learning become less sensitive to the phase of gratings than the simple-cell-like input neurons. In particular, some of the second-layer output neurons become completely phase invariant, owing to the convergence of the connections from first-layer neurons with similar orientation selectivity to second-layer neurons in the model network. We examine the parameter dependencies of the receptive field properties of the model neurons after learning and discuss their biological implications. We also show that the localized learning rule is consistent with experimental results concerning neuronal plasticity and can replicate the receptive fields of simple and complex cells.

Learning Invariance from Natural Images Inspired by Observations in the Primary Visual Cortex

The human visual system has the remarkable ability to largely recognize objects invariant of their position, rotation, and scale. A good interpretation of neurobiological findings involves a computational model that simulates signal processing of the visual cortex. In part, this is likely achieved step by step from early to late areas of visual perception. While several algorithms have been proposed for learning feature detectors, only few studies at hand cover the issue of biologically plausible learning of such invariance. In this study, a set of Hebbian learning rules based on calcium dynamics and homeostatic regulations of single neurons is proposed. Their performance is verified within a simple model of the primary visual cortex to learn so-called complex cells, based on a sequence of static images. As a result, the learned complex-cell responses are largely invariant to phase and position.

Contrast normalization contributes to a biologically-plausible model of receptive-field development in primary visual cortex (V1)

Neuronal populations in the primary visual cortex (V1) of mammals exhibit contrast normalization. Neurons that respond strongly to simple visual stimuli – such as sinusoidal gratings – respond less well to the same stimuli when they are presented as part of a more complex stimulus which also excites other, neighboring neurons. This phenomenon is generally attributed to generalized patterns of inhibitory connections between nearby V1 neurons. The Bienenstock, Cooper and Munro (BCM) rule is a neural network learning rule that, when trained on natural images, produces model neurons which, individually, have many tuning properties in common with real V1 neurons. However, when viewed as a population, a BCM network is very different from V1 – each member of the BCM population tends to respond to the same dominant features of visual input, producing an incomplete, highly redundant code for visual information. Here, we demonstrate that, by adding contrast normalization into the BCM rule, we arrive at a neurally-plausible Hebbian learning rule that can learn an efficient sparse, overcomplete representation that is a better model for stimulus selectivity in V1. This suggests that one role of contrast normalization in V1 is to guide the neonatal development of receptive fields, so that neurons respond to different features of visual input.

A sparse coding model with synaptically local plasticity and spiking neurons can account for the diverse shapes of V1 simple cell receptive fields

Sparse coding algorithms trained on natural images can accurately predict the features that excite visual cortical neurons, but it is not known whether such codes can be learned using biologically realistic plasticity rules. We have developed a biophysically motivated spiking network, relying solely on synaptically local information, that can predict the full diversity of V1 simple cell receptive field shapes when trained on natural images. This represents the first demonstration that sparse coding principles, operating within the constraints imposed by cortical architecture, can successfully reproduce these receptive fields. We further prove, mathematically, that sparseness and decorrelation are the key ingredients that allow for synaptically local plasticity rules to optimize a cooperative, linear generative image model formed by the neural representation. Finally, we discuss several interesting emergent properties of our network, with the intent of bridging the gap between theoretical and experimental studies of visual cortex.

In a sparse coding model, individual input stimuli are represented by the activities of model neurons, the majority of which are inactive in response to any particular stimulus. For a given class of stimuli, the neurons are optimized so that the stimuli can be faithfully represented with the minimum number of co-active units. This has been proposed as a model for visual cortex. While it has previously been demonstrated that sparse coding model neurons, when trained on natural images, learn to represent the same features as do neurons in primate visual cortex, it remains to be demonstrated that this can be achieved with physiologically realistic plasticity rules. In particular, learning in cortex appears to occur by the modification of synaptic connections between neurons, which must depend only on information available locally, at the synapse, and not, for example, on the properties of large numbers of distant cells. We provide the first demonstration that synaptically local plasticity rules are sufficient to learn a sparse image code, and to account for the observed response properties of visual cortical neurons: visual cortex actually could learn a sparse image code.

Unsupervised learning of generative and discriminative weights encoding elementary image components in a predictive coding model of cortical function

A method is presented for learning the reciprocal feedforward and feedback connections required by the predictive coding model of cortical function. When this method is used, feedforward and feedback connections are learned simultaneously and independently in a biologically plausible manner. The performance of the proposed algorithm is evaluated by applying it to learning the elementary components of artificial and natural images. For artificial images, the bars problem is employed, and the proposed algorithm is shown to produce state-of-the-art performance on this task. For natural images, components resembling Gabor functions are learned in the first processing stage, and neurons responsive to corners are learned in the second processing stage. The properties of these learned representations are in good agreement with neurophysiological data from V1 and V2. The proposed algorithm demonstrates for the first time that a single computational theory can explain the formation of cortical RFs and also the response properties of cortical neurons once those RFs have been learned.

Independent component analysis in spiking neurons

Although models based on independent component analysis (ICA) have been successful in explaining various properties of sensory coding in the cortex, it remains unclear how networks of spiking neurons using realistic plasticity rules can realize such computation. Here, we propose a biologically plausible mechanism for ICA-like learning with spiking neurons. Our model combines spike-timing dependent plasticity and synaptic scaling with an intrinsic plasticity rule that regulates neuronal excitability to maximize information transmission. We show that a stochastically spiking neuron learns one independent component for inputs encoded either as rates or using spike-spike correlations. Furthermore, different independent components can be recovered, when the activity of different neurons is decorrelated by adaptive lateral inhibition.

Receptive Field Self-Organization in a Model of the Fine Structure in V1 Cortical Columns

We study a dynamical model of processing and learning in the visual cortex, which reflects the anatomy of V1 cortical columns and properties of their neuronal receptive fields. Based on recent results on the fine-scale structure of columns in V1, we model the activity dynamics in subpopulations of excitatory neurons and their interaction with systems of inhibitory neurons. We find that a dynamical model based on these aspects of columnar anatomy can give rise to specific types of computations that result in self-organization of afferents to the column. For a given type of input, self-organization reliably extracts the basic input components represented by neuronal receptive fields. Self-organization is very noise tolerant and can robustly be applied to different types of input. To quantitatively analyze the system's component extraction capabilities, we use two standard benchmarks: the bars test and natural images. In the bars test, the system shows the highest noise robustness reported so far. If natural image patches are used as input, self-organization results in Gabor-like receptive fields. In quantitative comparison with in vivo measurements, we find that the obtained receptive fields capture statistical properties of V1 simple cells that algorithms such as independent component analysis or sparse coding do not reproduce.

Spatio-temporal adaptation in the unsupervised development of networked visual neurons

There have been many computational models mimicking the visual cortex that are based on spatial adaptations of unsupervised neural networks. In this paper, we present a new model called neuronal cluster which includes spatial as well as temporal weights in its unified adaptation scheme. The "in-place" nature of the model is based on two biologically plausible learning rules, Hebbian rule and lateral inhibition. We present the mathematical demonstration that the temporal weights are derived from the delay in lateral inhibition. By training with the natural videos, this model can develop spatio-temporal features such as orientation selective cells, motion sensitive cells, and spatio-temporal complex cells. The unified nature of the adaptation scheme allows us to construct a multilayered and task-independent attention selection network which uses the same learning rule for edge, motion, and color detection, and we can use this network to engage in attention selection in both static and dynamic scenes.

Efficient coding correlates with spatial frequency tuning in a model of V1 receptive field organization

Efficient coding has been proposed to play an essential role in early visual processing. While several approaches used an objective function to optimize a particular aspect of efficient coding, such as the minimization of mutual information or the maximization of sparseness, we here explore how different estimates of efficient coding in a model with nonlinear dynamics and Hebbian learning determine the similarity of model receptive fields to V1 data with respect to spatial tuning. Our simulation results indicate that most measures of efficient coding correlate with the similarity of model receptive field data to V1 data, that is, optimizing the estimate of efficient coding increases the similarity of the model data to experimental data. However, the degree of the correlation varies with the different estimates of efficient coding, and in particular, the variance in the firing pattern of each cell does not predict a similarity of model and experimental data.

Innate visual learning through spontaneous activity patterns

Patterns of spontaneous activity in the developing retina, LGN, and cortex are necessary for the proper development of visual cortex. With these patterns intact, the primary visual cortices of many newborn animals develop properties similar to those of the adult cortex but without the training benefit of visual experience. Previous models have demonstrated how V1 responses can be initialized through mechanisms specific to development and prior to visual experience, such as using axonal guidance cues or relying on simple, pairwise correlations on spontaneous activity with additional developmental constraints. We argue that these spontaneous patterns may be better understood as part of an "innate learning" strategy, which learns similarly on activity both before and during visual experience. With an abstraction of spontaneous activity models, we show how the visual system may be able to bootstrap an efficient code for its natural environment prior to external visual experience, and we continue the same refinement strategy upon natural experience. The patterns are generated through simple, local interactions and contain the same relevant statistical properties of retinal waves and hypothesized waves in the LGN and V1. An efficient encoding of these patterns resembles a sparse coding of natural images by producing neurons with localized, oriented, bandpass structure-the same code found in early visual cortical cells. We address the relevance of higher-order statistical properties of spontaneous activity, how this relates to a system that may adapt similarly on activity prior to and during natural experience, and how these concepts ultimately relate to an efficient coding of our natural world.

A sparse generative model of V1 simple cells with intrinsic plasticity

Current models for learning feature detectors work on two timescales: on a fast timescale, the internal neurons' activations adapt to the current stimulus; on a slow timescale, the weights adapt to the statistics of the set of stimuli. Here we explore the adaptation of a neuron's intrinsic excitability, termed intrinsic plasticity, which occurs on a separate timescale. Here, a neuron maintains homeostasis of an exponentially distributed firing rate in a dynamic environment. We exploit this in the context of a generative model to impose sparse coding. With natural image input, localized edge detectors emerge as models of V1 simple cells. An intermediate timescale for the intrinsic plasticity parameters allows modeling aftereffects. In the tilt aftereffect, after a viewer adapts to a grid of a certain orientation, grids of a nearby orientation will be perceived as tilted away from the adapted orientation. Our results show that adapting the neurons' gain-parameter but not the threshold-parameter accounts for this effect. It occurs because neurons coding for the adapting stimulus attenuate their gain, while others increase it. Despite its simplicity and low maintenance, the intrinsic plasticity model accounts for more experimental details than previous models without this mechanism.

Hebbian learning in a model with dynamic rate-coded neurons: an alternative to the generative model approach for learning receptive fields from natural scenes

Most computational models of coding are based on a generative model according to which the feedback signal aims to reconstruct the visual scene as close as possible. We here explore an alternative model of feedback. It is derived from studies of attention and thus, probably more flexible with respect to attentive processing in higher brain areas. According to this model, feedback implements a gain increase of the feedforward signal. We use a dynamic model with presynaptic inhibition and Hebbian learning to simultaneously learn feedforward and feedback weights. The weights converge to localized, oriented, and bandpass filters similar as the ones found in V1. Due to presynaptic inhibition the model predicts the organization of receptive fields within the feedforward pathway, whereas feedback primarily serves to tune early visual processing according to the needs of the task.

Implicit exploitation of regularities: novel correlations in images quickly alter visual perception

Viewing a set of two thousand images in which the contrasts of a central and a surrounding pattern were highly correlated increased the suppressive influence of the surround on the perceived contrast of the central pattern. The apparent increase in inhibition supports the operation of an anti-Hebbian mechanism between the two groups of cells excited by the patterns (one group by the central pattern and the other by the surround). According to this mechanism, inhibitory connections between nearby cells increase in efficacy according to a simple function of the correlations between the cells' activities.