Francesco Frisone

Coordinate-free sensorimotor processing: computing with population codes

The purpose of the study is to outline a computational architecture for the intelligent processing of sensorimotor patterns. The focus is on the nature of the internal representations of the outside world which are necessary for planning and other goal-oriented functions. A model of cortical map dynamics and self-organization is proposed that integrates a number of concepts and methods partly explored in the field. The novelty and the biological plausibility is related to the global architecture which allows one to deal with sensorimotor patterns in a coordinate-free way, using population codes as distributed internal representations of external variables and the coupled dynamics of cortical maps as a general tool of trajectory formation. The basic computational features of the model are demonstrated in the case of articulatory speech synthesis and some of the metric properties are evaluated by means of simple simulation studies.

Cortical Maps of Sensorimotor Spaces

Abstract The chapter overviews a computational framework for characterizing the cortical representations and processes which underly the kinematic invariances of movements (the notion of spatial control), also taking into account the new understanding of the cortex as a continuously adapting dynamical system, shaped by competitive and cooperative lateral connections. We show how a coordinate-free representation of sensorimotor spaces can emerge from self-organized learning which builds a topological representing structure, thereby defining the concept of cortical map. This also implies a mixture of local and long-range lateral connections, consistent with known anatomical facts, thus allowing the representation of high-dimensional spaces in an apparently flat anatomy. The dynamics of cortical maps is analyzed taking into account the excitatory nature of the majority of cortical synapses and the puzzling presence of long-range competition without long-range inhibition. A model is proposed which combines a process of diffusion (via the excitatory topologically organized connections) and a process of competitive distribution of activation which tends to sharpen the active map region. The result is a propagating waveform attracted by a target-coding broad input pattern. This is the basis for a field-computing architecture of the interacting cortical processes which underly motor planning and control. We also address the the emergence of a representation of external 3-D space in a multimodal cortical map, possibly allocated in posterior parietal cortex.

Extending the TRN Model in a Biologically Plausible Way

The Topology Representing Network (TRN) model is extended by using an activation dynamics which implicitly orders the neurons according to the distance from the input pattern. This allows to apply the same Hebbian learning method to the thalamo-cortical and cortico-cortical connections. The model proposed combines a process of diffusion (via the excitatory topologically organized connections) and a process of competitive distribution of activation which tends to sharpen the active map region. The dynamics is analyzed taking into account the excitatory nature of the majority of cortical synapses and the puzzling presence of long-range competition without long-range inhibition. The model is shown to be more consistent than TRN or other self-organizing paradigms with a number of neurophysiological facts.

Can the synchronization of cortical areas be evidenced by fMRI

Abstract The goal of this study is to investigate the possibility of analyzing spatio-temporal organization of the human cortical activity during different complex tasks, by means of fMRI. To evidence cortical areas synchronization we propose a computational approach based on a self-organizing neural networks (“neural gas”) that detects time-dependent alterations in the regional intensity of the functional signal. Results of the application of such approach are reported and are compared with the results obtained with a standard statistical package (SPM96). Future experimental investigations will be aimed at the analysis of spatio-temporal structures of cortical activity in pathological conditions, such as epilepsy.

Representing Multidimensional Stimuli on the Cortex

A computational paradigm based on distributed spatial representation provides an unifying framework for dealing with open issues in modelling cortical maps such as the representation of multidimensional stimuli. This paper describes a computational architecture, based on two overlayed Topology Representing Networks, which is shown to reproduce artificially the ocular dominance bands observed from tangential sections of a monkeys right occipital lobe.

Analysis of fMRI time series with mixtures of Gaussians

In this paper, we discuss the application of the mixtures of Gaussians model for density estimation to the analysis of fMRI time series. We show that, in a classical sensorimotor paradigm (finger-tapping), the performance of the proposed method (in terms of number and location of the detected activity-related voxels) is very similar to that of voxel-by-voxel linear regression, but does not require an explicit model of the activation pattern and/or of the hemodynamic response. In addition, if the number of mixture elements is increased, our method is capable of detecting additional activity-related areas.

Soft-competitive versus EM learning in cortical map modeling

Starting from the problem of density estimation, it is shown that EM learning can be considered as a Hebbian mechanism. From this it is possible to outline a theory of self-organization of cortical maps which is based on a well defined optimization process and still preserves biologically desirable characteristics: local computation and uniform treatment of input and lateral connections.

Advances in modeling cortical maps

Publisher Summary The purpose of this chapter is to explore the hypothesis that lateral connections in cortical maps are used to build topological internal representations, and propose that the latter are particularly suitable for the processing of high-dimensional “spatial” quantities, like sensorimotor information. It presents the previously formulated hypothesis that lateral connections in cortical maps are used to build topological internal representations suitable for processing sensorimotor information. The logical development can proceed in two directions. At the theoretical level, it can attempt a unification of the learning mechanism for the thalamo-cortical and cortico-coartical connections, thus substituting the crisp set of lateral connections, with a fuzzier but more robust set. From the neurobiological point of view, it is very important to correlate dynamic models of cortical maps with dynamic brain imaging.

Fast Learning of Dynamic Compensation in Motor Control

In the framework of the theory equilibrium-point control, a model for learning the compensation of dynamic loads is presented. It is self-supervised but non Hebbian and can compensate unexpected load variations in 1–2 repeated trials. A preliminary study is presented as regards the generalisation across tasks and the role of the cerebellar circuitry is discussed as a dynamic co-processor capable to implement part of the required computations.

Topologic organization of context fields for sensorimotor coordination

In field computing a topologic organization of CFs is necessary to support sensorimotor planning. A simple model of cortical dynamics can exploit such topologic organization.