Panagiotis C. Petrantonakis

A compressed sensing perspective of hippocampal function.

Hippocampus is one of the most important information processing units in the brain. Input from the cortex passes through convergent axon pathways to the downstream hippocampal subregions and, after being appropriately processed, is fanned out back to the cortex. Here, we review evidence of the hypothesis that information flow and processing in the hippocampus complies with the principles of Compressed Sensing (CS). The CS theory comprises a mathematical framework that describes how and under which conditions, restricted sampling of information (data set) can lead to condensed, yet concise, forms of the initial, subsampled information entity (i.e., of the original data set). In this work, hippocampus related regions and their respective circuitry are presented as a CS-based system whose different components collaborate to realize efficient memory encoding and decoding processes. This proposition introduces a unifying mathematical framework for hippocampal function and opens new avenues for exploring coding and decoding strategies in the brain.

Dendrites of dentate gyrus granule cells contribute to pattern separation by controlling sparsity.

The hippocampus plays a key role in pattern separation, the process of transforming similar incoming information to highly dissimilar, nonverlapping representations. Sparse firing granule cells (GCs) in the dentate gyrus (DG) have been proposed to undertake this computation, but little is known about which of their properties influence pattern separation. Dendritic atrophy has been reported in diseases associated with pattern separation deficits, suggesting a possible role for dendrites in this phenomenon. To investigate whether and how the dendrites of GCs contribute to pattern separation, we build a simplified, biologically relevant, computational model of the DG. Our model suggests that the presence of GC dendrites is associated with high pattern separation efficiency while their atrophy leads to increased excitability and performance impairments. These impairments can be rescued by restoring GC sparsity to control levels through various manipulations. We predict that dendrites contribute to pattern separation as a mechanism for controlling sparsity.

Dentate Gyrus Circuitry Features Improve Performance of Sparse Approximation Algorithms

Memory-related activity in the Dentate Gyrus (DG) is characterized by sparsity. Memory representations are seen as activated neuronal populations of granule cells, the main encoding cells in DG, which are estimated to engage 2–4% of the total population. This sparsity is assumed to enhance the ability of DG to perform pattern separation, one of the most valuable contributions of DG during memory formation. In this work, we investigate how features of the DG such as its excitatory and inhibitory connectivity diagram can be used to develop theoretical algorithms performing Sparse Approximation, a widely used strategy in the Signal Processing field. Sparse approximation stands for the algorithmic identification of few components from a dictionary that approximate a certain signal. The ability of DG to achieve pattern separation by sparsifing its representations is exploited here to improve the performance of the state of the art sparse approximation algorithm “Iterative Soft Thresholding” (IST) by adding new algorithmic features inspired by the DG circuitry. Lateral inhibition of granule cells, either direct or indirect, via mossy cells, is shown to enhance the performance of the IST. Apart from revealing the potential of DG-inspired theoretical algorithms, this work presents new insights regarding the function of particular cell types in the pattern separation task of the DG.

A Simple Method to Simultaneously Detect and Identify Spikes from Raw Extracellular Recordings

The ability to track when and which neurons fire in the vicinity of an electrode, in an efficient and reliable manner can revolutionize the neuroscience field. The current bottleneck lies in spike sorting algorithms; existing methods for detecting and discriminating the activity of multiple neurons rely on inefficient, multi-step processing of extracellular recordings. In this work, we show that a single-step processing of raw (unfiltered) extracellular signals is sufficient for both the detection and identification of active neurons, thus greatly simplifying and optimizing the spike sorting approach. The efficiency and reliability of our method is demonstrated in both real and simulated data.

An Architecture for the Acceleration of a Hybrid Leaky Integrate and Fire SNN on the Convey HC-2ex FPGA-Based Processor

Neuromorphic computing is expanding by leaps and bounds through custom integrated circuits (digital and analog), and large scale platforms developed by industry or government-funded projects (e.g. TrueNorth and BrainScaleS, respectively). Whereas the trend is for massive parallelism and neuromorphic computation in order to solve problems, such as those that may appear in machine learning and deep learning algorithms, there is substantial work on brain-like highly accurate neuromorphic computing in order to model the human brain. In such a form of computing, spiking neural networks (SNN) such as the Hodgkin and Huxley model are mapped to various technologies, including FPGAs. In this work, we present a highly efficient FPGA-based architecture for the detailed hybrid Leaky Integrate and Fire SNN that can simulate generic characteristics of neurons of the cerebral cortex. This architecture supports arbitrary, sparse O(n2) interconnection of neurons without need to re-compile the design, and plasticity rules, yielding on a four-FPGA Convey 2ex hybrid computer a speedup of 923x for a non-trivial data set on 240 neurons vs. the same model in the software simulator BRAIN on a Intel(R) Xeon(R) CPU E5-2620 v2 @ 2.10GHz, i.e. the reference state-of-the-art software. Although the reference, official software is single core, the speedup demonstrates that the application scales well among multiple FPGAs, whereas this would not be the case in general-purpose computers due to the arbitrary interconnect requirements. The FPGA-based approach leads to highly detailed models of parts of the human brain up to a few hundred neurons vs. a dozen or fewer neurons on the reference system.

A Novel and Simple Spike Sorting Implementation

Monitoring the activity of multiple, individual neurons that fire spikes in the vicinity of an electrode, namely perform a Spike Sorting (SS) procedure, comprises one of the most important tools for contemporary neuroscience in order to reverse-engineer the brain. As recording electrodes’ technology rabidly evolves by integrating thousands of electrodes in a confined spatial setting, the algorithms that are used to monitor individual neurons from recorded signals have to become even more reliable and computationally efficient. In this work, we propose a novel framework of the SS approach in which a single-step processing of the raw (unfiltered) extracellular signal is sufficient for both the detection and sorting of the activity of individual neurons. Despite its simplicity, the proposed approach exhibits comparable performance with state-of-the-art approaches, especially for spike detection in noisy signals, and paves the way for a new family of SS algorithms with the potential for multi-recording, fast, on-chip implementations.

Dendrites of DG granule cells contribute to pattern separation by controlling sparsity

The hippocampus plays a key role in pattern separation, namely the process of transforming similar incoming information to highly dissimilar, non-overlapping representations. Sparse firing granule cells in the dentate gyrus have been proposed to undertake this computation, but little is known about which of their properties influence pattern separation. Dendritic atrophy and spine loss have been reported in diseases associated with pattern separation deficits, suggesting a possible role for dendrites in this phenomenon. To investigate whether and how the dendrites of granule cells contribute to pattern separation, we build a simplified, biologically relevant, computational model of the dentate gyrus. Our model suggests that the presence of granule cell dendrites is associated with high pattern separation efficiency while their atrophy leads to increased excitability and performance impairments that cannot be explained by input resistance changes. These impairments, however, can be rescued by a range of manipulations that restore network sparsity to control levels. Thus, our model suggests that the contribution of dendrites to pattern separation amounts to one of many ways for controlling sparsity. We provide a number of testable predictions that can help investigate this proposition experimentally.

Dendritic nonlinearities enable PFC microcircuits to serve as predictive modules of persistent activity

The ability to monitor and probe the activity of large neuronal networks both in vivo and in vitro has recently established that neurons of various brain regions are organized into spatially restricted clusters (or small assemblies) that are bi-directionally connected, share common inputs and are co-activated during behavioral tasks [1,2]. Investigations regarding the functional implications of such neuronal clustering have proposed that this modularity may underlie the spiking irregularities seen in cortical activity in vivo [3] or code for the execution of a voluntary movement [4]. In the prefrontal cortex (PFC), such microcircuits are proposed to support the spontaneous emergence of Up and Down states [5], a phenomenon linked to persistent activity, which is the cellular correlate of working memory. In this work we investigate the functional role of PFC microcircuits in the expression of persistent activity, focusing on the contribution of nonlinear dendritic properties to the induction, termination, and coding of upcoming state transitions. Towards this goal we developed a layer V PFC microcircuit consisting of 7 pyramidal neurons and 2 interneurons implemented in the NEURON simulation environment. Modelling equations for the biophysical mechanisms used have been reported in [6,7]. All neuron models were biophysically detailed but morphologically simplified and were validated regarding their intrinsic, synaptic and connectivity properties (e.g. number of synapses, latencies etc). Our results show that the non-linear integration of synaptic inputs at the basal dendrites of pyramidal neurons, mediated by the induction of NMDA-spikes, is imperative for the emergence of the persistent state in the microcircuit, but this necessity disappears when increasing the network size. Moreover dendritic versus somatic specific alterations of ionic currents (such as the R type VGCCs) differentially modulate persistent activity induction, substantiating the critical role of location specific effects of various neuromodulators. Finally, we find that several features of the network activity prior to the induction and/or termination of persistent firing contain predictive information of the upcoming state-transition(s), which is readily available to downstream neurons. These findings suggest that PFC microcircuits may serve as tunable and predictive modules of persistent activity and subsequently working memory.