Large-scale Neural Models Research Articles

What must come down goes up - the effect of noise on weights in spike-timing-dependent plasticity

In large-scale spiking neural models that learn with spike-timing-dependent plasticity (STDP), it is a crucial, but difficult problem to balance synaptic potentiation (necessary to learn a task) and synaptic depression (necessary to counteract accidental weight increase and to balance overall firing rates). Adding random noise and choosing parameters such that inhibition slightly dominates excitation [1,2] is considered one way of accomplishing this. With the learning rule being standard STDP [2], the parameter that determines whether inhibition or excitation dominates is the so-called α-parameter, defined as the ratio between parameters that determine amount of depression and those that determine amount of potentiation [3]. If the α-parameter is set to a value greater than 1, the weights are assumed to go down. However, using numerical simulations, we demonstrate that this is not the case. Using standard leaky integrate-and-fire neurons, we ran 10 simulations per parameter set (250s each), with α-parameters A = {0.5, 0.9, 1.0, 1.1, 1.25, 1.5} and noise frequencies F = {5, 10, 20, 40, 80, 160}. We also varied the maximum weight and the initial weight of the synapse from 0 to 1 in small intervals. In contrast to our assumptions, we find a weight increase with α-parameters greater than 1 (see Figure ​Figure1).1). The complex weight dynamics observed can be explained by the interplay of 3 factors. 1. The α-parameter: if it is greater than 1 it tends to drive the weights down, given that firing is uncorrelated. 2. The impact of the pre-synaptic spike on the post-synaptic potential (determined by synaptic weight, capacitance of the neuron, or time constant of the synapse): if the impact is strong enough pre-synaptic spikes will cause post-synaptic spikes and firing will no longer be uncorrelated. Hence, weights will tend to go up (for example, if the initial values of the weights are too high). 3. Noise frequency: There is a positive interaction between noise frequency and α-parameter in driving weights down. If the frequency is high enough, the α-parameter can be low. If the frequency is low the α-parameter needs to be very high. High frequency drives the weights down, because the post-synaptic neuron operates in input averaging mode, firing more regular than the pre-synaptic spikes that evoked it, and, therefore, being decorrelated from it [2]. These findings give insights in how to set parameters (in particular α-parameter, initial weights and noise frequency) to achieve a desired weight dynamics in large-scale STDP models. Figure 1 A plot of the weight strength after training for different frequencies with an initial weight of 0.5. As expected, the weight increases for α-parameters smaller than 1. However, the weight also increases for α-parameters larger than 1 ...

Retraction: Benchmarking Neuromorphic Systems with Nengo.

[This retracts the article on p. 380 in vol. 9, PMID: 26539076.].

Benchmarking neuromorphic systems with Nengo.

Nengo is a software package for designing and simulating large-scale neural models. Nengo is architected such that the same Nengo model can be simulated on any of several Nengo backends with few to no modifications. Backends translate a model to specific platforms, which include GPUs and neuromorphic hardware. Nengo also contains a large test suite that can be run with any backend and focuses primarily on functional performance. We propose that Nengo's large test suite can be used to benchmark neuromorphic hardware's functional performance and simulation speed in an efficient, unbiased, and future-proof manner. We implement four benchmark models and show that Nengo can collect metrics across five different backends that identify situations in which some backends perform more accurately or quickly.

Magnetic resonance imaging of healthy and diseased brain networks.

An important aspect of neuroscience is to characterize the underlying connectivity patterns of the human brain. Recent advances in magnetic resonance imaging (MRI) techniques (e.g., structural MRI, diffusion MRI, and functional MRI) and network analysis approaches such as graph theory have allowed to investigate the patterns of structural and functional connectivity of human brain (i.e., human connectome, Sporns et al., 2005). Using imaging connectomics, many studies have demonstrated that large-scale human brain networks have many non-trivial topological properties such as small-worldness, modular structure, and highly connected hubs. Moreover, these quantifiable network properties significantly correlate with behavioral, environmental, and genetic factors, and change with age, learning, and disease. Network analysis of neuroimaging data is opening up a new avenue of research into the understanding of the organizational principles of the brain that will be crucial for all basic scientists and clinical researchers. Such approaches are fascinating but there are a number of challenging issues in the extraction of reliable brain networks from various imaging modalities and the analyses of the topological properties, such as the definitions of network nodes and edges, and the reproducibility of the network analysis results. Nonetheless, the field of imaging connectomics has significantly advanced in the past several years, largely due to the rapid development of network models, tools and methodologies, and their widespread applications in cognitive and clinical research. In this research topic, we aimed at compiling works representing the state of the art in structural and functional brain networks in healthy and disease populations using neuroimaging data. We collected 29 articles that were from a number of internationally recognized scientists who have made significant contributions to this field. The article types are diverse and the topics covered various research directions including imaging techniques, computational modeling, network analysis approaches and tools, and applications. We assembled one hypothesis and theory article, one perspective article, and four review articles. In the hypothesis and theory article, Horwitz et al. (2013) highlighted recent efforts toward using large-scale neural modeling to explore the relationship between structural connectivity and functional/effective connectivity. Structural connectivity and functional/effective connectivity are the most fundamental concepts for the descriptions of brain networks, but their relationship is elusive. Horwitz et al. (2013) emphasized that the alteration of structural connectivity known in models does not necessarily result in matching changes in functional/effective connectivity and vice versa, suggesting that caution should be taken in the result interpretation of structural-functional connectivity relationship. Upon summarizing three commonly used strategies of imaging connectomics as biomarkers of brain diseases, Kaiser (2013) presented a novel fourth option for future disease biomarker studies, i.e., dynamic connectomes that use computational models of simulated brain activity based on structural connectivity rather than the structural connectome itself. Among the four review articles, Li et al. (2014) summarized brain connectivity studies related to the default-mode network (DMN) in the fields of social understanding: emotion perception, empathy, theory of mind, and morality, and suggested the vital roles of the DMN in this domain; Hoff et al. (2014) reviewed resting fMRI studies representing developmental changes in the functional brain networks from 20 weeks of gestation onwards, highlighting different developmental rates of network connectivity in different brain systems; De La Fuente et al. (2013) provided a systematic review regarding brain connectivity studies in attention-deficit/hyperactivity disorder and proposed that the roles of the subcortical structures and their structural/functional pathways in this developmental disorder should be further studied; Bernhardt et al. (2013) reviewed the application of multi-modal imaging techniques and brain connectivity approaches in temporal lobe epilepsy, specifically highlighting findings from graph-theoretical analysis that assessed the topological organization of brain networks. Together, these articles suggest that the combination of multi-modal imaging techniques and advanced network analysis approaches such as computational modeling and graph theory provides unique opportunities to enrich our understanding of biological mechanisms in healthy and diseased brains. In these articles, a number of important research directions were proposed for future brain network studies based on neuroimaging data. We assembled 23 original research articles, which can be roughly classified into imaging and analysis methodologies, tools, and applications in various domains.

Involuntary switching into the native language induced by electrocortical stimulation of the superior temporal gyrus: A multimodal mapping study

We describe involuntary language switching from L2 to L1 evoked by electro-stimulation in the superior temporal gyrus in a 30-year-old right-handed Serbian (L1) speaker who was also a late Italian learner (L2). The patient underwent awake brain surgery. Stimulation of other portions of the exposed cortex did not cause language switching as did not stimulation of the left inferior frontal gyrus, where we evoked a speech arrest. Stimulation effects on language switching were selective, namely, interfered with counting behaviour but not with object naming. The coordinates of the positive site were combined with functional and fibre tracking (DTI) data. Results showed that the language switching site belonged to a significant fMRI cluster in the left superior temporal gyrus/supramarginal gyrus found activated for both L1 and L2, and for both the patient and controls, and did not overlap with the inferior fronto-occipital fasciculus (IFOF), the inferior longitudinal fasciculus (ILF) and the superior longitudinal fasciculus (SLF). This area, also known as Stp, has a role in phonological processing.Language switching phenomenon we observed can be partly explained by transient dysfunction of the feed-forward control mechanism hypothesized by the DIVA (Directions Into Velocities of Articulators) model (Golfinopoulos, E., Tourville, J. A., & Guenther, F. H. (2010). The integration of large-scale neural network modeling and functional brain imaging in speech motor control. NeuroImage, 52, 862–874).

Large-scale spiking circuit simulation of spatio-temporal dynamics in superior colliculus

We present computational simulations of the intrinsic circuitry of the superior colliculus using large-scale spiking neural circuit models. We reproduce recent results from slice experiments that showed different spatio-temporal patterns of interaction within the visual layers versus the eye-movement related layers of the superior colliculus. Specifically, the receptive fields of neurons in the visual layers implement a “center-surround” pattern of spatial competition, and furthermore additional input within the central region sums super-linearly. In contrast, the receptive fields of neurons in the motor-related regions implement spatially symmetric fields of overlaid excitation and inhibition, and additional inputs sum linearly. Our simulations investigate the circuit mechanisms and dynamics that differentiate the computational roles of these distinct but related regions. We constructed full-scale simulations of mouse brain slices using spiking neuron models. Connection parameters were then fit within physiological constraints to reproduce experimental data. Figure ​Figure1A1A shows results from a best fit of the superficial (visual) layers. Stimulation was applied at different horizontal distances from a central neuron in a tangential slice. The figure shows integrated post-synaptic potential (PSP) during stimulation at the indicated distance (purple). We also reproduced data from a “two-point” paradigm in which stimulation was simultaneously applied to the indicated electrode and the central electrode, simulating a “large” stimulus (red). The net response of the neuron sums super-linearly in comparison to the linear summation of the results from the two points independently (naive summation in black). Insets show corresponding slice data. Interestingly, anatomically distinct inhibitory and excitatory populations were not necessary to reproduce spatial asymmetry (“center-surround”) in the visual superficial layers. Rather, asymmetries in the temporal properties (synaptic dynamics) of connections were better able to account for observed data under all conditions. A further population of small disinhibitory neurons was necessary to maintain linear summation in the periphery while summing super-linearly in the center. Figure 1 This contrasts with the motor-related intermediate layers (Figure ​(Figure1B),1B), whose response to stimulation was best explained by both spatial and temporal symmetry between inhibitory and excitatory neural connections. Overall, the regions seem to have taken advantage of both spatial and temporal dynamics in their connections to specialize their computational function: The visual layers seem geared towards spatial competition and strengthening of salient stimuli, whereas the motor-related regions seem geared towards broad and long-term integration of input.

Neurogrid: A Mixed-Analog-Digital Multichip System for Large-Scale Neural Simulations

In this paper, we describe the design of Neurogrid, a neuromorphic system for simulating large-scale neural models in real time. Neuromorphic systems realize the function of biological neural systems by emulating their structure. Designers of such systems face three major design choices: 1) whether to emulate the four neural elements-axonal arbor, synapse, dendritic tree, and soma-with dedicated or shared electronic circuits; 2) whether to implement these electronic circuits in an analog or digital manner; and 3) whether to interconnect arrays of these silicon neurons with a mesh or a tree network. The choices we made were: 1) we emulated all neural elements except the soma with shared electronic circuits; this choice maximized the number of synaptic connections; 2) we realized all electronic circuits except those for axonal arbors in an analog manner; this choice maximized energy efficiency; and 3) we interconnected neural arrays in a tree network; this choice maximized throughput. These three choices made it possible to simulate a million neurons with billions of synaptic connections in real time-for the first time-using 16 Neurocores integrated on a board that consumes three watts.

Interaction of feedforward and feedback streams in visual cortex in a firing-rate model of columnar computations

Visual sensory input stimuli are rapidly processed along bottom-up feedforward cortical streams. Beyond such driving streams neurons in higher areas provide information that is re-entered into the representations and responses at the earlier stages of processing. The precise mechanisms and underlying functionality of such associative feedforward/feedback interactions are not resolved. This work develops a neuronal circuit at a level mimicking cortical columns with response properties linked to single cell recordings. The proposed model constitutes a coarse-grained model with gradual firing-rate responses which accounts for physiological in vitro recordings from mammalian cortical cells. It is shown that the proposed population-based circuit with gradual firing-rate dynamics generates responses like those of detailed biophysically realistic multi-compartment spiking models. The results motivate using a coarse-grained mechanism for large-scale neural network modeling and simulations of visual cortical mechanisms. They further provide insights about how local recurrent loops change the gain of modulating feedback signals.

Efficient Spiking Neural Network Model of Pattern Motion Selectivity in Visual Cortex

Simulating large-scale models of biological motion perception is challenging, due to the required memory to store the network structure and the computational power needed to quickly solve the neuronal dynamics. A low-cost yet high-performance approach to simulating large-scale neural network models in real-time is to leverage the parallel processing capability of graphics processing units (GPUs). Based on this approach, we present a two-stage model of visual area MT that we believe to be the first large-scale spiking network to demonstrate pattern direction selectivity. In this model, component-direction-selective (CDS) cells in MT linearly combine inputs from V1 cells that have spatiotemporal receptive fields according to the motion energy model of Simoncelli and Heeger. Pattern-direction-selective (PDS) cells in MT are constructed by pooling over MT CDS cells with a wide range of preferred directions. Responses of our model neurons are comparable to electrophysiological results for grating and plaid stimuli as well as speed tuning. The behavioral response of the network in a motion discrimination task is in agreement with psychophysical data. Moreover, our implementation outperforms a previous implementation of the motion energy model by orders of magnitude in terms of computational speed and memory usage. The full network, which comprises 153,216 neurons and approximately 40 million synapses, processes 20 frames per second of a 40 × 40 input video in real-time using a single off-the-shelf GPU. To promote the use of this algorithm among neuroscientists and computer vision researchers, the source code for the simulator, the network, and analysis scripts are publicly available.

HRLSim: A High Performance Spiking Neural Network Simulator for GPGPU Clusters

Modeling of large-scale spiking neural models is an important tool in the quest to understand brain function and subsequently create real-world applications. This paper describes a spiking neural network simulator environment called HRL Spiking Simulator (HRLSim). This simulator is suitable for implementation on a cluster of general purpose graphical processing units (GPGPUs). Novel aspects of HRLSim are described and an analysis of its performance is provided for various configurations of the cluster. With the advent of inexpensive GPGPU cards and compute power, HRLSim offers an affordable and scalable tool for design, real-time simulation, and analysis of large-scale spiking neural networks.

Systematic approximations of neural fields through networks of neural masses in the virtual brain

Full brain network models comprise a large-scale connectivity (the connectome) and neural mass models as the network's nodes. Neural mass models absorb implicitly a variety of properties in their constant parameters to achieve a reduction in complexity. In situations, where the local network connectivity undergoes major changes, such as in development or epilepsy, it becomes crucial to model local connectivity explicitly. This leads naturally to a description of neural fields on folded cortical sheets with local and global connectivities. The numerical approximation of neural fields in biologically realistic situations as addressed in Virtual Brain simulations (see http://thevirtualbrain.org/app/ (version 1.0)) is challenging and requires a thorough evaluation if the Virtual Brain approach is to be adapted for systematic studies of disease and disorders. Here we analyze the sampling problem of neural fields for arbitrary dimensions and provide explicit results for one, two and three dimensions relevant to realistically folded cortical surfaces. We characterize (i) the error due to sampling of spatial distribution functions; (ii) useful sampling parameter ranges in the context of encephalographic (EEG, MEG, ECoG and functional MRI) signals; (iii) guidelines for choosing the right spatial distribution function for given anatomical and geometrical constraints.

Mozaik: a framework for model construction, simulation, data analysis and visualization for large-scale spiking neural circuit models

Event Abstract Back to Event Mozaik: a framework for model construction, simulation, data analysis and visualization for large-scale spiking neural circuit models Jan Antolik1 and Andrew P. Davison1* 1 CNRS, UNIC, France The increasing amount of computational resources becoming available is causing a shift in computational neuroscience towards increasingly heterogeneous models of neural circuits and brain regions, and more importantly, employment of increasingly complex stimulation and experimental protocols in an effort to bridge the gap between simulations and biological experiments. This growth in complexity of virtual experiments poses a challenge for the existing tool-chains, as the set of tools that are involved in typical modeller's workflow is expanding, with a growing amount and complexity of metadata, describing the experimental context, flowing between them. A plethora of tools is currently available covering different parts of the workflow; however, numerous areas lack dedicated tools, while integration and interoperability of existing tools is limited. This forces modellers to either handle the workflow manually, leading to errors, or to write substantial amounts of code to automate parts of the workflow, in both cases hindering their productivity. To address these issues, we have developed Mozaik: an integrated workflow system for spiking neuronal network simulations written in Python. Mozaik integrates the model, experiment and stimulation specification, simulation execution, data storage, data analysis and visualization into a single automated workflow, ensuring that all relevant metadata are available to each of the workflow components. It is based on several widely used Python tools, including PyNN [1], Neo [2] and Matplotlib [3]. It offers scientists a declarative way of specifying models and recording configurations, using hierarchically organized configuration files. The models are constructed using topologically organized sheets of neurons and projections between such sheets of neurons. After specifying the model, a user can define the list of experimental protocols to be executed with the model. Each experiment defines a list of stimuli that will be presented to the model, and optionally can also involve direct stimulation of neurons in the model. Using a single command, a user can instruct Mozaik to create an instance of the model and execute the specified experiments. Mozaik will automatically record all data into a datastore and, importantly, also save all the relevant metadata describing the experimental context. The datastore can be accessed via a query interface, allowing for easy and efficient data manipulation. This query system is extensible and already offers a number of operations that are commonly used in analysis or visualization. The analysis and visualization modules operate over the datastore, using the query system. They are written in a modular and extensible way, utilizing the saved metadata to automate numerous processes. This ensures an efficient way to add new analysis and visualization methods. Mozaik is written in a modular way, making it easy to replace any of its components, while the existing modules are designed to be extensible with minimal programming effort. Mozaik increases the productivity of running virtual experiments on highly structured neural networks, by significantly automating the whole experimental cycle, while increasing the reliability of modeling studies by relieving the user from manually handling the flow of metadata between the individual stages of the workflow. Acknowledgements Supported by the European Union through project number FP7-269921 (BrainScaleS).

A large-scale neural network model of the influence of neuromodulatory levels on working memory and behavior

The dorsolateral prefrontal cortex (dlPFC), which is regarded as the primary site for visuospatial working memory in the brain, is significantly modulated by dopamine (DA) and norepinephrine (NE). DA and NE originate in the ventral tegmental area (VTA) and locus coeruleus (LC), respectively, and have been shown to have an “inverted-U” dose-response profile in dlPFC, where the level of arousal and decision-making performance is a function of DA and NE concentrations. Moreover, there appears to be a sweet spot, in terms of the level of DA and NE activation, which allows for optimal working memory and behavioral performance. When either DA or NE is too high, input to the PFC is essentially blocked. When either DA or NE is too low, PFC network dynamics become noisy and activity levels diminish. Mechanisms for how this is occurring have been suggested, however, they have not been tested in a large-scale model with neurobiologically plausible network dynamics. Also, DA and NE levels have not been simultaneously manipulated experimentally, which is not realistic in vivo due to strong bi-directional connections between the VTA and LC. To address these issues, we built a spiking neural network model that includes D1, α2A, and α1 receptors. The model was able to match the inverted-U profiles that have been shown experimentally for differing levels of DA and NE. Furthermore, we were able to make predictions about what working memory and behavioral deficits may occur during simultaneous manipulation of DA and NE outside of their optimal levels. Specifically, when DA levels were low and NE levels were high, cues could not be held in working memory due to increased noise. On the other hand, when DA levels were high and NE levels were low, incorrect decisions were made due to weak overall network activity. We also show that lateral inhibition in working memory may play a more important role in increasing signal-to-noise ratio than increasing recurrent excitatory input.

Perceptual Decision Making “Through the Eyes” of a Large-Scale Neural Model of V1

Sparse coding has been posited as an efficient information processing strategy employed by sensory systems, particularly visual cortex. Substantial theoretical and experimental work has focused on the issue of sparse encoding, namely how the early visual system maps the scene into a sparse representation. In this paper we investigate the complementary issue of sparse decoding, for example given activity generated by a realistic mapping of the visual scene to neuronal spike trains, how do downstream neurons best utilize this representation to generate a “decision.” Specifically we consider both sparse (L1-regularized) and non-sparse (L2 regularized) linear decoding for mapping the neural dynamics of a large-scale spiking neuron model of primary visual cortex (V1) to a two alternative forced choice (2-AFC) perceptual decision. We show that while both sparse and non-sparse linear decoding yield discrimination results quantitatively consistent with human psychophysics, sparse linear decoding is more efficient in terms of the number of selected informative dimension.

Interpreting the effects of altered brain anatomical connectivity on fMRI functional connectivity: a role for computational neural modeling.

Recently, there have been a large number of studies using resting state fMRI to characterize abnormal brain connectivity in patients with a variety of neurological, psychiatric, and developmental disorders. However, interpreting what the differences in resting state fMRI functional connectivity (rsfMRI-FC) actually reflect in terms of the underlying neural pathology has proved to be elusive because of the complexity of brain anatomical connectivity. The same is the case for task-based fMRI studies. In the last few years, several groups have used large-scale neural modeling to help provide some insight into the relationship between brain anatomical connectivity and the corresponding patterns of fMRI-FC. In this paper we review several efforts at using large-scale neural modeling to investigate the relationship between structural connectivity and functional/effective connectivity to determine how alterations in structural connectivity are manifested in altered patterns of functional/effective connectivity. Because the alterations made in the anatomical connectivity between specific brain regions in the model are known in detail, one can use the results of these simulations to determine the corresponding alterations in rsfMRI-FC. Many of these simulation studies found that structural connectivity changes do not necessarily result in matching changes in functional/effective connectivity in the areas of structural modification. Often, it was observed that increases in functional/effective connectivity in the altered brain did not necessarily correspond to increases in the strength of the anatomical connection weights. Note that increases in rsfMRI-FC in patients have been interpreted in some cases as resulting from neural plasticity. These results suggest that this interpretation can be mistaken. The relevance of these simulation findings to the use of functional/effective fMRI connectivity as biomarkers for brain disorders is also discussed.

Nexa: A scalable neural simulator with integrated analysis

Large-scale neural simulations encompass challenges in simulator design, data handling and understanding of simulation output. As the computational power of supercomputers and the size of network models increase, these challenges become even more pronounced. Here we introduce the experimental scalable neural simulator Nexa, for parallel simulation of large-scale neural network models at a high level of biological abstraction and for exploration of the simulation methods involved. It includes firing-rate models and capabilities to build networks using machine learning inspired methods for e.g. self-organization of network architecture and for structural plasticity. We show scalability up to the size of the largest machines currently available for a number of model scenarios. We further demonstrate simulator integration with online analysis and real-time visualization as scalable solutions for the data handling challenges.

Dynamics of large-scale neuronal networks of the human cortex functional connectivity

Spatio-temporally organized low-frequency fluctuations (<0.1 Hz) of blood-oxygen-level-dependent (BOLD) fMRI signal have been intensively investigated as a measure of functional connectivity (FC) between region pairs in the whole brain [1]. Resting state FC is commonly assumed to be shaped by the underlying anatomical connectivity (AC). Furthermore, it has been suggested that the strength, persistence, and spatial properties of FC are constrained by the large-scale anatomical structure of the cortex [2]. However, strong resting state FC is often observed between pairs of remote cortical regions, even without apparent direct anatomical connections [3]. Mechanisms generating resting state FC are largely unknown, and it has been contended that indirect connections, interregional distance, and collective effects governed by network properties of the cortex play significant role. In addition, some theoretical studies on large-scale brain networks demonstrated the importance of time delays in networks dynamics for the generation of resting state FC fluctuations [4,5]. To address these questions we investigate large-scale neural network model of human cortex FC. Our model is based on an empirically derived resting state FC network consisting of 64 region of interest (ROIs) (network nodes), which are chosen from all over the cortex. The ROIs are adapted from a study of functional segmentation of the brain cortex using high-model-order independent component analysis (ICA) [6]. There are 30 pairs of inter-hemispheric homologues, and 4 additional ROIs are chosen along the midline. The activity of each node is described by FitzHugh-Nagumo neurons. Network dynamics is modelled with different parameters for each node and different time delays to account for the finite signal propagation times between the nodes.

Using large-scale neural models to interpret connectivity measures of cortico-cortical dynamics at millisecond temporal resolution

Over the last two decades numerous functional imaging studies have shown that higher order cognitive functions are crucially dependent on the formation of distributed, large-scale neuronal assemblies (neurocognitive networks), often for very short durations. This has fueled the development of a vast number of functional connectivity measures that attempt to capture the spatiotemporal evolution of neurocognitive networks. Unfortunately, interpreting the neural basis of goal directed behavior using connectivity measures on neuroimaging data are highly dependent on the assumptions underlying the development of the measure, the nature of the task, and the modality of the neuroimaging technique that was used. This paper has two main purposes. The first is to provide an overview of some of the different measures of functional/effective connectivity that deal with high temporal resolution neuroimaging data. We will include some results that come from a recent approach that we have developed to identify the formation and extinction of task-specific, large-scale neuronal assemblies from electrophysiological recordings at a ms-by-ms temporal resolution. The second purpose of this paper is to indicate how to partially validate the interpretations drawn from this (or any other) connectivity technique by using simulated data from large-scale, neurobiologically realistic models. Specifically, we applied our recently developed method to realistic simulations of MEG data during a delayed match-to-sample (DMS) task condition and a passive viewing of stimuli condition using a large-scale neural model of the ventral visual processing pathway. Simulated MEG data using simple head models were generated from sources placed in V1, V4, IT, and prefrontal cortex (PFC) for the passive viewing condition. The results show how closely the conclusions obtained from the functional connectivity method match with what actually occurred at the neuronal network level.

Large scale neural model of the human cortex functional connectivity based on empirically derived resting state networks

Large scale neural model of the human cortex functional connectivity based on empirically derived resting state networks

A role for neural modeling in the study of brain disorders.

A role for neural modeling in the study of brain disorders.