Institute for Theoretical Computer Science

We review different aspects of the simulation of spiking neural networks. We start by reviewing the different types of simulation strategies and algorithms that are currently implemented. We next review the precision of those simulation strategies, in particular in cases where plasticity depends on the exact timing of the spikes. We overview different simulators and simulation environments presently available (restricted to those freely available, open source and documented). For each simulation tool, its advantages and pitfalls are reviewed, with an aim to allow the reader to identify which simulator is appropriate for a given task. Finally, we provide a series of benchmark simulations of different types of networks of spiking neurons, including Hodgkin-Huxley type, integrate-and-fire models, interacting with current-based or conductance-based synapses, using clock-driven or event-driven integration strategies. The same set of models are implemented on the different simulators, and the codes are made available. The ultimate goal of this review is to provide a resource to facilitate identifying the appropriate integration strategy and simulation tool to use for a given modeling problem related to spiking neural networks.

Reward-modulated spike-timing-dependent plasticity (STDP) has recently emerged as a candidate for a learning rule that could explain how behaviorally relevant adaptive changes in complex networks of spiking neurons could be achieved in a self-organizing manner through local synaptic plasticity. However, the capabilities and limitations of this learning rule could so far only be tested through computer simulations. This article provides tools for an analytic treatment of rewardmodulated STDP, which allows us to predict under which conditions reward-modulated STDP will achieve a desired learning effect. These analytical results imply that neurons can learn through reward-modulated STDP to classify not only spatial but also temporal firing patterns of presynaptic neurons. They also can learn to respond to specific presynaptic firing patterns with particular spike patterns. Finally, the resulting learning theory predicts that even difficult credit-assignment problems, where it is very hard to tell which synaptic weights should be modified in order to increase the global reward for the system, can be solved in a self-organizing manner through reward-modulated STDP. This yields an explanation for a fundamental experimental result on biofeedback in monkeys by Fetz and Baker. In this experiment monkeys were rewarded for increasing the firing rate of a particular neuron in the cortex and were able to solve this extremely difficult credit assignment problem. Our model for this experiment relies on a combination of reward-modulated STDP with variable spontaneous firing activity. Hence it also provides a possible functional explanation for trial-to-trial variability, which is characteristic for cortical networks of neurons but has no analogue in currently existing artificial computing systems. In addition our model demonstrates that reward-modulated STDP can be applied to all synapses in a large recurrent neural network without endangering the stability of the network dynamics. Citation: Legenstein R, Pecevski D, Maass W (2008) A Learning Theory for Reward-Modulated Spike-Timing-Dependent Plasticity with Application to Biofeedback. PLoS Comput Biol 4(10): e1000180.

An important open problem of computational neuroscience is the generic organization of computations in networks of neurons in the brain. We show here through rigorous theoretical analysis that inherent stochastic features of spiking neurons, in combination with simple nonlinear computational operations in specific network motifs and dendritic arbors, enable networks of spiking neurons to carry out probabilistic inference through sampling in general graphical models. In particular, it enables them to carry out probabilistic inference in Bayesian networks with converging arrows (''explaining away'') and with undirected loops, that occur in many real-world tasks. Ubiquitous stochastic features of networks of spiking neurons, such as trial-to-trial variability and spontaneous activity, are necessary ingredients of the underlying computational organization. We demonstrate through computer simulations that this approach can be scaled up to neural emulations of probabilistic inference in fairly large graphical models, yielding some of the most complex computations that have been carried out so far in networks of spiking neurons.

NEVESIM is a software package for event-driven simulation of networks of spiking neurons with a fast simulation core in C++, and a scripting user interface in the Python programming language. It supports simulation of heterogeneous networks with different types of neurons and synapses, and can be easily extended by the user with new neuron and synapse types. To enable heterogeneous networks and extensibility, NEVESIM is designed to decouple the simulation logic of communicating events (spikes) between the neurons at a network level from the implementation of the internal dynamics of individual neurons. In this paper we will present the simulation framework of NEVESIM, its concepts and features, as well as some aspects of the object-oriented design approaches and simulation strategies that were utilized to efficiently implement the concepts and functionalities of the framework. We will also give an overview of the Python user interface, its basic commands and constructs, and also dis...

Computational neuroscience has produced a diversity of software for simulations of networks of spiking neurons, with both negative and positive consequences. On the one hand, each simulator uses its own programming or confi guration language, leading to considerable diffi culty in porting models from one simulator to another. This impedes communication between investigators and makes it harder to reproduce and build on the work of others. On the other hand, simulation results can be cross-checked between different simulators, giving greater confi dence in their correctness, and each simulator has different optimizations, so the most appropriate simulator can be chosen for a given modelling task. A common programming interface to multiple simulators would reduce or eliminate the problems of simulator diversity while retaining the benefi ts. PyNN is such an interface, making it possible to write a simulation script once, using the Python programming language, and run it without modifi cation on any supported simulator (currently NEURON, NEST, PCSIM, Brian and the Heidelberg VLSI neuromorphic hardware). PyNN increases the productivity of neuronal network modelling by providing high-level abstraction, by promoting code sharing and reuse, and by providing a foundation for simulator-agnostic analysis, visualization and data-management tools. PyNN increases the reliability of modelling studies by making it much easier to check results on multiple simulators. PyNN is open-source software and is available from http://neuralensemble.org/PyNN. compiler standards and simulators develop. Another is that model source code is often not written with reuse and extension in mind, and so considerable rewriting to modularize the code is necessary. Probably the most important barrier is that code written for one simulator is not compatible with any other simulator.

The means by which cortical neural networks are able to efficiently solve inference problems remains on open question in computational neuroscience. Recently, abstract models of Bayesian computation in neural circuits have been proposed, but they lack a mechanistic interpretation at the single-cell level. In this article, we describe a complete theoretical framework for building networks of leaky integrate-and-fire neurons that can sample from arbitrary probability distributions over binary random variables. We test our framework for a model inference task based on a psychophysical phenomenon (the Knill-Kersten optical illusion) and further assess its performance when applied to randomly generated distributions. As the local computations performed by the network strongly depend on the interaction between neurons, we compare several types of couplings mediated by either single synapses or interneuron chains. Due to its robustness to substrate imperfections such as parameter noise and background noise correlations, our model is particularly interesting for implementation on novel, neuro-inspired computing architectures, which can thereby serve as a fast, low-power substrate for solving real-world inference problems.

The Parallel Circuit SIMulator (PCSIM) is a software package for simulation of neural circuits. It is primarily designed for distributed simulation of large scale networks of spiking point neurons. Although its computational core is written in C++, PCSIM's primary interface is implemented in the Python programming language, which is a powerful programming environment and allows the user to easily integrate the neural circuit simulator with data analysis and visualization tools to manage the full neural modeling life cycle. The main focus of this paper is to describe PCSIM's full integration into Python and the benefi ts thereof. In particular we will investigate how the automatically generated bidirectional interface and PCSIM's object-oriented modular framework enable the user to adopt a hybrid modeling approach: using and extending PCSIM's functionality either employing pure Python or C++ and thus combining the advantages of both worlds. Furthermore, we describe several supplementary PCSIM packages written in pure Python and tailored towards setting up and analyzing neural simulations.

Processing and storage of sensory information is based on the interaction between different neural populations rather than the isolated activity of single neurons. In order to characterize the dynamic interaction and transient cooperation of sub-circuits within a neural network, multivariate autoregressive (MVAR) models have proven to be an important analysis tool. In this study, we apply directed functional coupling based on MVAR models and describe the temporal and spatial changes of functional coupling between simultaneously recorded local field potentials in extrastriate area V4 during visual memory. Specifically, we compare the strength and directional relations of coupling based on generalized partial directed coherence (GPDC) measures while two rhesus monkeys perform a visual short-term memory task. In both monkeys we find increases in theta power during the memory period that are accompanied by changes in directed coupling. These interactions are most prominent in the low frequency range encompassing the theta band (3–12 Hz) and, more importantly, are asymmetric between pairs of recording sites. Furthermore, we find that the degree of interaction decreases as a function of distance between electrode positions, suggesting that these interactions are a predominantly local phenomenon. Taken together, our results show that directed coupling measures based on MVAR models are able to provide important insights into the spatial and temporal formation of local functionally coupled ensembles during visual memory in V4. Moreover, our findings suggest that visual memory is accompanied not only by a temporary increase of oscillatory activity in the theta band, but by a direction-dependent change in theta coupling, which ultimately represents a change in functional connectivity within the neural circuit.

Short-term memory requires communication between multiple brain regions that collectively mediate the encodingand maintenance of sensory information. It has been suggested that oscillatory synchronization underlies intercorticalcommunication. Yet, whether and how distant cortical areas cooperate during visual memory remains elusive. We examined neuralinteractions between visual area V4 and the lateral prefrontal cortex using simultaneous local field potential (LFP) recordingsand single-unit activity (SUA) in monkeys performing a visual short-term memory task. During the memory period, we observedenhanced between-area phase synchronization in theta frequencies (3–9 Hz) of LFPs together with elevated phase locking of SUAto theta oscillations across regions. In addition, we found that the strength of intercortical locking was predictive of the animals’behavioral performance. This suggests that theta-band synchronization coordinates action potential communication between V4and prefrontal cortex that may contribute to the maintenance of visual short-term memories.

Recurrent neural networks, such as Liquid State Machines and Echo State Net-works, that are driven by feedback from readout units with trained input weightshave proven to be computationally extremely powerful systems. Traditionally,readouts are trained using supervised batch learning techniques and teacher forc-ing of the feedback during training. More recently, a supervised online learningmechanism forthe training of readouts has been proposedthat uses the actual out-putofthereadoutduringtraining. We extendandmodifythis approachbytrainingthe weights of the readout units using a simple online reward-modulated Hebbianlearning rule. In contrast to supervised learning, the readouts do not have explicitinformation on the error, and learning only incorporates local information on thepre- and postsynaptic activity and a global reward signal. In our experiments, weshow that this method enables a readout to generate periodic patterns out of aninitially chaotic recurrent circuit. To accomplish this, it drives the network into astable state such that the periodic pattern can be reliably produced after learning.Moreover, we show that two readouts can be trained using a common reward sig-nal in order to produce a simple memory of the input. Furthermore, our resultsare in line with the hypothesis that noise plays a crucial role for learning insteadof only acting as a disadvantageous disturbance.

This paper addresses the question how generic microcircuits of neurons in different parts of the cortex can attain and maintain different computational specializations. We show that if stochastic variations in the dynamics of local microcircuits are correlated with signals related to functional improvements of the brain (e.g. in the control of behavior), the computational operation of these microcircuits can become optimized for specific tasks such as the generation of specific periodic signals and task-dependent routing of information. Furthermore, we show that working memory can autonomously emerge through reward-modulated Hebbian learning, if needed for specific tasks. Altogether, our results suggest that reward-modulated synaptic plasticity can not only optimize the network parameters for specific computational tasks, but also initiate a functional rewiring that re-programs microcircuits, thereby generating diverse computational functions in different generic cortical microcircuits. On a more general level, this work provides a new perspective for a standard model for computations in generic cortical microcircuits (liquid computing model). It shows that the arguably most problematic assumption of this model, the postulate of a teacher that trains neural readouts through supervised learning, can be eliminated. We show that generic networks of neurons can learn numerous biologically relevant computations through trial and error.

According to Sandro Nannini's Time and Consciousness in Cognitive Naturalism, we can draw an analogy between the shift in the conception of time that occurred in physics with the introduction of relativity theory and a shift towards a scientifically more graspable functional concept of phenomenal consciousness. This analogy is meant to persuade us of the eliminative materialist view that we should abandon our folk psychological concept of consciousness. In my commentary, I examine the naturaliza-tion procedure underlying Nannini's cognitive naturalism, argue for its inability to account for the phenomenal feel of conscious states, and point to some important differences between the conceptual change in the case of time and the intended change in the case of consciousness.

Physicalism is a metaphysical thesis easily presented in slogan form – there is nothing over and above the physical – but notoriously difficult to formulate precisely. 'Understanding physicalism' combines insights from contemporary philosophy of mind and metaphysics to present a new account of physical properties and metaphysical dependence and, on this foundation, develop a more rigorous and illuminating formulation of the thesis of physicalism.

Abstract: According to Sandro Nannini’s Time and Consciousness in Cognitive Naturalism , we can draw an analogy between the shift in the conception of time that occurred in physics with the introduction of relativity theory and a shift towards a scientifically more graspable functional concept of phenomenal consciousness. This analogy is meant to persuade us of the eliminative materialist view that we should abandon our folk psychological concept of consciousness. In my commentary, I examine the naturalization procedure underlying Nannini’s cognitive naturalism, argue for its inability to account for the phenomenal feel of conscious states, and point to some important differences between the conceptual change in the case of time and the intended change in the case of consciousness. Keywords:  Eliminative Materialism; Phenomenal Consciousness; Explanatory Gap; Hard Problem; Cognitive Naturalism. Naturalismo cognitivo e sensazioni fenomeniche Riassunto:  In Time and Consciousness in C...

Recent developments in neuromorphic hardware engineering make mixed-signal VLSI neural network models promising candidates for neuroscientific research tools and massively parallel computing devices, especially for tasks which exhaust the computing power of software simulations. Still, like all analog hardware systems, neuromorphic models suffer from a constricted configurability and production-related fluctuations of device characteristics. Since also future systems, involving ever-smaller structures, will inevitably exhibit such inhomogeneities on the unit level, self-regulation properties become a crucial requirement for their successful operation. By applying a cortically inspired self-adjusting network architecture, we show that the activity of generic spiking neural networks emulated on a neuromorphic hardware system can be kept within a biologically realistic firing regime and gain a remarkable robustness against transistorlevel variations. As a first approach of this kind in engineering practice, the short-term synaptic depression and facilitation mechanisms implemented within an analog VLSI model of I&F neurons are functionally utilized for the purpose of network level stabilization. We present experimental data acquired both from the hardware model and from comparative software simulations which prove the applicability of the employed paradigm to neuromorphic VLSI devices.

The Parallel Circuit SIMulator (PCSIM) is a software package for simulation of neural circuits. It is primarily designed for distributed simulation of large scale networks of spiking point neurons. Although its computational core is written in C++, PCSIM's primary interface is implemented in the Python programming language, which is a powerful programming environment and allows the user to easily integrate the neural circuit simulator with data analysis and visualization tools to manage the full neural modeling life cycle. The main focus of this paper is to describe PCSIM's full integration into Python and the benefi ts thereof. In particular we will investigate how the automatically generated bidirectional interface and PCSIM's object-oriented modular framework enable the user to adopt a hybrid modeling approach: using and extending PCSIM's functionality either employing pure Python or C++ and thus combining the advantages of both worlds. Furthermore, we describe several supplementary PCSIM packages written in pure Python and tailored towards setting up and analyzing neural simulations.

Recent developments in neuromorphic hardware engineering make mixed-signal VLSI neural network models promising candidates for neuroscientific research tools and massively parallel computing devices, especially for tasks which exhaust the computing power of software simulations. Still, like all analog hardware systems, neuromorphic models suffer from a constricted configurability and production-related fluctuations of device characteristics. Since also future systems, involving ever-smaller structures, will inevitably exhibit such inhomogeneities on the unit level, self-regulation properties become a crucial requirement for their successful operation. By applying a cortically inspired self-adjusting network architecture, we show that the activity of generic spiking neural networks emulated on a neuromorphic hardware system can be kept within a biologically realistic firing regime and gain a remarkable robustness against transistorlevel variations. As a first approach of this kind in engineering practice, the short-term synaptic depression and facilitation mechanisms implemented within an analog VLSI model of I&F neurons are functionally utilized for the purpose of network level stabilization. We present experimental data acquired both from the hardware model and from comparative software simulations which prove the applicability of the employed paradigm to neuromorphic VLSI devices.

A major goal of computational neuroscience is the creation of computer models for cortical areas whose response to sensory stimuli resembles that of cortical areas in vivo in important aspects. It is seldom considered whether the simulated spiking activity is realistic (in a statistical sense) in response to natural stimuli. Because certain statistical properties of spike responses were suggested to facilitate computations in the cortex, acquiring a realistic firing regimen in cortical network models might be a prerequisite for analyzing their computational functions. We present a characterization and comparison of the statistical response properties of the primary visual cortex (V1) in vivo and in silico in response to natural stimuli. We recorded from multiple electrodes in area V1 of 4 macaque monkeys and developed a large state-of-the-art network model for a 5 × 5-mm patch of V1 composed of 35,000 neurons and 3.9 million synapses that integrates previously published anatomical a...

The Functional Mock-up Interface (FMI) (Modelica Association 2021b) is a tool independent standard for the exchange of dynamic models and for co-simulation. FMI 2.0, released in 2014, is recognized as the de-facto standard in industry for exchanging models and tool coupling, and is currently supported by more than 160 simulation tools. Version 3.0 of the standard brings many new features that allow for advanced co-simulation algorithms and new use cases such as packaging and simulation of highly accurate virtual Electronic Control Units (vECUs). Besides Model-Exchange and Co-Simulation, a third interface type, Scheduled Execution, is defined for purely discrete, RTOS-like, simulation and supports preemption. Clocks allow the synchronization of events between Functional Mock-up Units (FMUs) and the importer. There is better support for data types including binary data and arrays. Advanced co-simulation approaches are enabled by intermediate variable access between communication point...