My name is Niceto Rafael Luque and I am currently a Marie-curie Post-Doc Fellow at the Institute of Vision, University Pierre et Marie Curie
I received my B.S in Electronics Engineering and an M.S. in Automatics and Industrial Electronics from the University of Cordoba (Spain) in 2003 and 2006, respectively. In April 2007 I officially joined to the University of Granada with a National Grant as a researcher of the European Project SENSOPAC .I also received my M.S. in Computer Architecture and Networks from the University of Granada in 2007.
Finally, I received my Doctorate from the University of Granada in 2013 in Control Engineering and Computer Science.
From 2012 to 2014 I participated in an EU project related to adaptive learning mechanisms and bio-inspired control REALNET. In August 2014, I officially joined the Human Brain Project (HBP); a ten-year, large-scale European research initiative whose goal is to better understand the human brain and its diseases and ultimately to emulate its computational capabilities.Finally, in 2015 I obtained an IF Marie Curie Post-Doc Fellowship from the EU Comission. My main research interests include biologically processing control schemes, light weight robots, and cerebellar spiking neural networks.
Experimental studies about the Central Nervous System (CNS) in all levels (sub cellular, cellular and at system level) are performed in order to obtain a better understanding of its anatomic structures and the physiological processes that the CNS seems to possess. Nevertheless, the observations to be done with that aim must be managed within a representative scenario where the functional description of the CNS is available. This is possible just in case when all the needed conceptual elements that properly describe the CNS functionality are available too. Both Physiologists and Neurophysiologists have traditionally used the performance (or the lack of performance in presence of pathologies), as the basis for the functional assessment of the CNS components, thus producing useful qualitative and phenomenal models. Although these models are often more than enough for clinical issues they do not provide a detailed comprehension of the whole CNS.
The current technology allows a restricted in vivo access to the CNS (mainly to the more external areas) by means of functional magnetic resonance imaging and magnetoencephalography. Similarly, it is of common use, recordings by means of electrode matrices; however these recordings just allow extracellular access of barely a hundred neurons at best.
But most of the functional neural networks related to the hippocampus and the cerebellum (two of the best-known regions) are sized from just a hundred of thousand to several millions of cells. The information process within these neural networks occurs thanks to the self-organized dynamic patterns of the neural activity that covers a large proportion of the nervous system. These emerging patterns can be hardly understood taking into account just individual activities of individual cells ( or even hundred of cells) in the same way that it is tough to understand a book just taken into account individual words. Even the data collected from very large scale studies do not present the necessary resolution for observing these patterns and making the corresponding cellular interaction matches.
The biologically plausible computational models (cerebellum, inferior olive nucleus, cuneate nucleus ...etc) give answer to this problem allowing the study of neural network models " as large as it is needed" using neuronal models that have been developed according to experimental cellular data. These neural network models can be computationally simulated in pretty different conditions and circumstances to give a pretty consistent idea about how the CNS neural networks may operate. In many cases, these models are becoming a fundamental tool in the neuroscience hypothesis-experimentation cycle. The computational models allow researchers to test their "what's up when ...?" in simulation. This fact leads to make better hypothesis and better experiments designed with greater probability of success.
From this perspective, and thanks to the exponential computational performance evolution, Computational Neuroscience has positioned over the last years as a promising sub-field in neuroscience. Computational Neuroscience must not be considered as just a tool to better understand the behavior of a functional neural network within the CNS by using a mathematical analysis and massive computational simulations but also as a fundamental element to determine 1) what the different parts of the CNS do 2) How these different parts do what they do
In such scenario I have been developing my research during these years in the framework of two European projects (SENSOPAC and REALNET) helping to develop different models of diverse nervous system elements(cerebellum, inferior olive nucleus and cuneate nucleus) in cooperation with different research groups from neurophysiology(Egidio D'Angelo) trough computational neurobiology(Angelo Arleo) to robotics(Patrick van der Smagt).
Understanding how the brain processes and represents information is at the core of experimental studies of the Central Nervous System (CNS). A network of brain subsystems mediates information processing through distributed neural computation and dynamic patterns of neural activity. Over the last decades, studying how these patterns are elicited in the CNS under specific behavioural tasks has become a break through research topic in integrative neuroscience. These specific tasks are related to the concept of embodied cognition, according to which the primary goal of the CNS is to solve and facilitate the body-environment interaction. This project focuses on the cerebellum, a brain region that plays a crucial role in body-environment interaction, with a primary function related to adaptive motor control and coordination. The functional characteristics of the cerebellum make it a perfect candidate to start modelling and building an embodied nervous system. The cerebellar capability of performing adaptive information processing mediating sensorimotor control will be evaluated in specific tasks. Additionally, the emergence of cognitive-like representations will be studied by focusing on how models of the environment/tools can be acquired through a closed-loop sensorimotor interaction. This project sets forth a multidisciplinary methodology combining neuromimetic models and embodied neurorobotics. Simulated neural models and robotic experiments will guarantee full access to the system properties, which will be assessed through both qualitative and quantitative performance indicators to facilitate a constructive cross-validation against neurophysiological data. This approach will also allow us to predict new functional roles of specific cell/network/topology properties. The goal of this project lies on moving forward the knowledge frontiers in integrative neuroscience and biological control.
Amongst computational models of various brain regions, the well-organized structure of cerebellum, has received special attention from researchers belonging to very different fields. On the one hand, neurophysiologists have studied and proposed detailed models and descriptions according to experimentally recorded cells and synaptic properties. However, these current detailed models are not meant to perform specific tasks at a behavioral or cognitive level (equivalent to awake animal protocols) as in functional experiments to control real agents (as front-end body). On the other hand, engineers have proposed machine-like systems that try to solve particular biological problems from a systemic point of view. Based on these opposed approaches, several cerebellar modeling frameworks have been proposed:
In state-generator models, the granule cell layer presents on/off type “granule” entities that provide a sparse coding of the state space (Marr-Albus Model(Marr, 1969;Albus, 1971) ,CMAC(Albus, 1975) model, or Yamazaki and Tanaka model(Yamazaki and Tanaka, 2005;2007a;2009;Yamazaki and Nagao, 2012)). These models succeed in explaining some traditional cerebellum-involving tasks such as eyelid conditioning(Yamazaki and Tanaka, 2007b) or motor control tasks(Manoonpong et al., 2007;Luque et al., 2011b;a). In functional models, only the functional abstraction of specific cerebellar operations are considered (MPFIM model(Wolpert and Kawato, 1998), Adaptive Filter model(Fujita, 1982;Porrill and Dean, 2007;Dean and Porril, 2008;Dean and Porrill, 2010;Dean et al., 2010), APG model(Houk et al., 1996), or LWPR model(Tolu et al., 2012)). This kind of approximation derives from an engineering point of view and can solve most of the tasks performed by common cerebellar models, such as eyelid conditioning, the vestibule ocular reflex (VOR), or movement correction. Finally, cellular-level models capture the biophysical features of cerebellar neuronal processing, and can be evaluated in the framework of neurophysiological experiments. These models are highly biologically plausible, but their application in the context of large-scale cerebellar modeling and computation remains limited. The very first approximations in this field were developed based on the Schweighofer–Arbib model(Schweighofer et al., 1998;Schweighofer et al., 1998b). Therefore, it can be noticed that there is a gap to be covered where neurophysiologists and system engineers need to work hand in hand towards a better understanding of how specific problems are solved using biologically-plausible computational principles(De Schutter, 2008).
As aforementioned, computational models of various brain regions have been developed and studied for more than thirty years in order to analyze central brain functions. Computational neuroscience (CN) is the natural complement of experimental brain research, since it provides a synthetic explanation about specific mechanisms and models which are only partially observed using anatomy, physiology or behavioral experimentation.
Fig. 1 Cerebellar microcircuitry. Distributed plasticity within the cerebellar microcircuitry is indicated by the red ovals.
It is a fact that, many computational models about the cerebro-cerebellar loop have been proposed since Marr and Albus thus providing elegant explanations about the core of the forward controller operation that cerebro-cerebellar loop seems to carry out. Nevertheless, these computational theories tend to focus in one part of the cerebellar circuitry and then extrapolating the obtained conclusions to the whole cerebro-cerebellar system. It is also true that functional features are not either suddenly going to emerge from modeling all the cerebellar parts together since small deviations in many of the estimated network parameters (Sporns, 2006) can cause large deviations in resultant global behavior. But, it is also true that simulating nervous systems "connected" to a body (agent or robot with sensors and actuators) could be of great interest for studying how certain capabilities of the nervous system (e.g. the role of the cerebellum in coordinated movements and object manipulation) are based on cellular characteristics, nervous system topology or local synaptic adaptation mechanisms. This project represents an integrative approach which builds the bridge between task specific experimentation (equivalent to “awake animal experimentation”) and Systems Neuroscience models.
This allows studying the role of certain nervous systems under what it is called "behavioral/cognitive tasks". This is closely related to the concept of "embodied cognition" (Pfeifer and Bongard, 2007), in which the main aim of the CNS is to solve and facilitate the body interaction with the environment. Therefore, it is crucial to study nervous system models within the framework of its interaction with a body (sensors and actuators) and environment.
Fig. 2. Cerebellar control loop. The cerebellum provides corrective commands to compensate the mismatch between the inverse dynamic model and the real robot.
The present project focuses on an integrated approach to the cerebellar circuit modeling within real time “behavioral and cognitive tasks”. This is a compromise approach based on the assumption that most cerebellar functions involve the cooperative computation of several neural subcircuits and circuits
Albus, J.S. (1971). A theory of cerebellar function. Math Biosci. 10, 25-61.
Albus, J.S. (1975). Data storage in the cerebellar model articulation controller (CMAC). J. Dyn. Syst. Meas. Contr. ASME 3, 228–233.
De Schutter, E. (2008). Why Are Computational Neuroscience and Systems Biology So Separate? PLoS Comput Biol 4, e1000078.
Dean, P., and Porril, J. (2008). Adaptive filter Models of the Cerebellum. Computational Analysis. The Cerebellum 7, 567-571.
Dean, P., and Porrill, J. (2010). The cerebellum as an adaptive filter: a general model? Funct. Neurol. 25, 173-180.
Dean, P., Porrill, J., Ekerot, C.F., and Jörntel, H. (2010). The cerebellar microcircuit as an adaptive filter: experimental and computational evidence. Nat. Rev. Neurosci. 11, 30-34.
Fujita, M. (1982). Adaptive filter model of the cerebellum. Biol. Cybern. 45, 195-206.
Houk, J.C., Buckingham, J.T., and Barto, A.G. (1996). Models of cerebellum and motor learning. Behavioral and Brain Sciences 19, 369-383.
Luque, N.R., Garrido, J.A., Carrillo, R.R., Coenen, O.J.M.D., and Ros, E. (2011a). Cerebellar Input Configuration Toward Object Model Abstraction in Manipulation Tasks. Neural Networks, IEEE Trans. 22, 1321-1328.
Luque, N.R., Garrido, J.A., Carrillo, R.R., Coenen, O.J.M.D., and Ros, E. (2011b). Cerebellarlike Corrective Model Inference Engine for Manipulation Tasks. Systems, Man, and Cybernetics, Part B: Cybernetics, IEEE Trans. 41, 1299-1312.
Manoonpong, P., Geng, T., Kulvicius, T., Porr, B., and Wörgötter, F. (2007). Adaptive, Fast Walking in a Biped Robot under Neuronal Control and Learning. PLoS Comput Biol 3, e134.
Marr, D. (1969). A theory of cerebellar cortex. J. Physiol. 202, 437-470.
Pfeifer, R., and Bongard, J. (2007). How the body shapes the way we think - a new view on intelligence.
Porrill, J., and Dean, P. (2007). Cerebellar Motor Learning: When Is Cortical Plasticity Not Enough? PLoS Comput Biol 3, e197.
Schweighofer, N., Arbib, M.A., and Kawato, M. (1998). Role of the cerebellum in reaching movements in human. I. Distributed Inverse dynamics control. European J. Neurosci. 10, 86-94.
Schweighofer, N., Spoelstra, J., Arbib, M.A., and Kawato, M. (1998b). Role of the cerebellum in reaching movements in human. II. A neural model of the intermediate cerebellum. Eur.J. Neurosci. 10, 95-105.
Sporns, O. (2006). Small-world connectivity, motif composition, and complexity of fractal neuronal connections. Dedicated to the Memory of Ray Paton 85, 55-64.
Tolu, S., Vanegas, M., Luque, N.R., Garrido, J.A., and Ros, E. (2012). Bio-inspired Adaptive Feedback Error Learning Architecture for Motor Control. Biol. Cybern. 106, 507-522.
Wolpert, W., and Kawato, M. (1998). Multiple paired forward and inverse models for motor control. Neural Networks 11, 1317-1329.
Yamazaki, T., and Nagao, S. (2012). A Computational Mechanism for Unified Gain and Timing Control in the Cerebellum. PLoS ONE 3, e33319.
Yamazaki, T., and Tanaka, S. (2005). Neural modeling of an internal clock. Neural Comp. 17, 1032-1058.
Yamazaki, T., and Tanaka, S. (2007a). The cerebellum as a liquid state machine. Neural Networks 20, 290-297.
Yamazaki, T., and Tanaka, S. (2007b). A spiking network model for passage-of-time representation in the cerebellum. Eur. J. Neurosci. 26, 2279-2292.
Yamazaki, T., and Tanaka, S. (2009). Computational models of timing mechanisms in the cerebellar granular layer. Cerebellum 8, 423-432.
Hopefully any questions you had about me will have been answered clearly now you've looked at this site. If there are still any unanswered questions, please feel free to use the contact form below..