A large-scale neurocomputational model of spatial cognition integrating memory with vision.

We introduce a large-scale neurocomputational model of spatial cognition called 'Spacecog', which integrates recent findings from mechanistic models of visual and spatial perception. As a high-level cognitive ability, spatial cognition requires the processing of behaviourally relevant features in complex environments and, importantly, the updating of this information during processes of eye and body movement. The Spacecog model achieves this by interfacing spatial memory and imagery with mechanisms of object localisation, saccade execution, and attention through coordinate transformations in parietal areas of the brain. We evaluate the model in a realistic virtual environment where our neurocognitive model steers an agent to perform complex visuospatial tasks. Our modelling approach opens up new possibilities in the assessment of neuropsychological data and human spatial cognition.

Auto-Selection of an Optimal Sparse Matrix Format in the Neuro-Simulator ANNarchy.

Modern neuro-simulators provide efficient implementations of simulation kernels on various parallel hardware (multi-core CPUs, distributed CPUs, GPUs), thereby supporting the simulation of increasingly large and complex biologically realistic networks. However, the optimal configuration of the parallel hardware and computational kernels depends on the exact structure of the network to be simulated. For example, the computation time of rate-coded neural networks is generally limited by the available memory bandwidth, and consequently, the organization of the data in memory will strongly influence the performance for different connectivity matrices. We pinpoint the role of sparse matrix formats implemented in the neuro-simulator ANNarchy with respect to computation time. Rather than asking the user to identify the best data structures required for a given network and platform, such a decision could also be carried out by the neuro-simulator. However, it requires heuristics that need to be adapted over time for the available hardware. The present study investigates how machine learning methods can be used to identify appropriate implementations for a specific network. We employ an artificial neural network to develop a predictive model to help the developer select the optimal sparse matrix format. The model is first trained offline using a set of training examples on a particular hardware platform. The learned model can then predict the execution time of different matrix formats and decide on the best option for a specific network. Our experimental results show that using up to 3,000 examples of random network configurations (i.e., different population sizes as well as variable connectivity), our approach effectively selects the appropriate configuration, providing over 93% accuracy in predicting the suitable format on three different NVIDIA devices.

BOLD Monitoring in the Neural Simulator ANNarchy.

Multi-scale network models that simultaneously simulate different measurable signals at different spatial and temporal scales, such as membrane potentials of single neurons, population firing rates, local field potentials, and blood-oxygen-level-dependent (BOLD) signals, are becoming increasingly popular in computational neuroscience. The transformation of the underlying simulated neuronal activity of these models to simulated non-invasive measurements, such as BOLD signals, is particularly relevant. The present work describes the implementation of a BOLD monitor within the neural simulator ANNarchy to allow an on-line computation of simulated BOLD signals from neural network models. An active research topic regarding the simulation of BOLD signals is the coupling of neural processes to cerebral blood flow (CBF) and cerebral metabolic rate of oxygen (CMRO2). The flexibility of ANNarchy allows users to define this coupling with a high degree of freedom and thus, not only allows to relate mesoscopic network models of populations of spiking neurons to experimental BOLD data, but also to investigate different hypotheses regarding the coupling between neural processes, CBF and CMRO2 with these models. In this study, we demonstrate how simulated BOLD signals can be obtained from a network model consisting of multiple spiking neuron populations. We first demonstrate the use of the Balloon model, the predominant model for simulating BOLD signals, as well as the possibility of using novel user-defined models, such as a variant of the Balloon model with separately driven CBF and CMRO2 signals. We emphasize how different hypotheses about the coupling between neural processes, CBF and CMRO2 can be implemented and how these different couplings affect the simulated BOLD signals. With the BOLD monitor presented here, ANNarchy provides a tool for modelers who want to relate their network models to experimental MRI data and for scientists who want to extend their studies of the coupling between neural processes and the BOLD signal by using modeling approaches. This facilitates the investigation and model-based analysis of experimental BOLD data and thus improves multi-scale understanding of neural processes in humans.

A computational model-based analysis of basal ganglia pathway changes in Parkinson's disease inferred from resting-state fMRI.

Previous computational model-based approaches for understanding the dynamic changes related to Parkinson's disease made particular assumptions about Parkinson's disease-related activity changes or specified dopamine-dependent activation or learning rules. Inspired by recent model-based analysis of resting-state fMRI, we have taken a data-driven approach. We fit the free parameters of a spiking neuro-computational model to match correlations of blood oxygen level-dependent signals between different basal ganglia nuclei and obtain subject-specific neuro-computational models of two subject groups: Parkinson patients and matched controls. When comparing mean firing rates at rest and connectivity strengths between the control and Parkinsonian model groups, several significant differences were found that are consistent with previous experimental observations. We discuss the implications of our approach and compare its results also with the popular "rate model" of the basal ganglia. Our study suggests that a model-based analysis of imaging data from healthy and Parkinsonian subjects is a promising approach for the future to better understand Parkinson-related changes in the basal ganglia and corresponding treatments.

Comparison of GPU- and CPU-implementations of mean-firing rate neural networks on parallel hardware.

Modern parallel hardware such as multi-core processors (CPUs) and graphics processing units (GPUs) have a high computational power which can be greatly beneficial to the simulation of large-scale neural networks. Over the past years, a number of efforts have focused on developing parallel algorithms and simulators best suited for the simulation of spiking neural models. In this article, we aim at investigating the advantages and drawbacks of the CPU and GPU parallelization of mean-firing rate neurons, widely used in systems-level computational neuroscience. By comparing OpenMP, CUDA and OpenCL implementations towards a serial CPU implementation, we show that GPUs are better suited than CPUs for the simulation of very large networks, but that smaller networks would benefit more from an OpenMP implementation. As this performance strongly depends on data organization, we analyze the impact of various factors such as data structure, memory alignment and floating precision. We then discuss the suitability of the different hardware depending on the networks' size and connectivity, as random or sparse connectivities in mean-firing rate networks tend to break parallel performance on GPUs due to the violation of coalescence.