Limitations of short range Mexican hat connection for driving target selection in a 2D neural field: activity suppression and deviation from input stimuli.

Geoffrey Mégardon,Christophe Tandonnet,Alain Guillaume,Petroc Sumner

doi:10.3389/fncom.2015.00128

Abstract

Dynamic Neural Field models (DNF) often use a kernel of connection with short range excitation and long range inhibition. This organization has been suggested as a model for brain structures or for artificial systems involved in winner-take-all processes such as saliency localization, perceptual decision or target/action selection. A good example of such a DNF is the superior colliculus (SC), a key structure for eye movements. Recent results suggest that the superficial layers of the SC (SCs) exhibit relatively short range inhibition with a longer time constant than excitation. The aim of the present study was to further examine the properties of a DNF with such an inhibition pattern in the context of target selection. First we tested the effects of stimulus size and shape on when and where self-maintained clusters of firing neurons appeared, using three variants of the model. In each model variant, small stimuli led to rapid formation of a spiking cluster, a range of medium sizes led to the suppression of any activity on the network and hence to no target selection, while larger sizes led to delayed selection of multiple loci. Second, we tested the model with two stimuli separated by a varying distance. Again single, none, or multiple spiking clusters could occur, depending on distance and relative stimulus strength. For short distances, activity attracted toward the strongest stimulus, reminiscent of well-known behavioral data for saccadic eye movements, while for larger distances repulsion away from the second stimulus occurred. All these properties predicted by the model suggest that the SCs, or any other neural structure thought to implement a short range MH, is an imperfect winner-take-all system. Although, those properties call for systematic testing, the discussion gathers neurophysiological and behavioral data suggesting that such properties are indeed present in target selection for saccadic eye movements.

Highlights

The ability to select important stimuli for further processing and action planning is a key function of brains of visually dominant animals
We constructed a Dynamic Neural Field (DNF)-Mexican hat (MH) integrating short range MH connections based on recent results obtained in the superficial layers of the superior colliculus (SC), and we tested how it performs in very simple target selection tasks: (1) the localization of a single stimulus of different sizes; (2) the selection and localization of the strongest of a pair of stimulations
Our work demonstrates that even a short range inhibition can enable a selection dynamic

Summary

Introduction

The ability to select important stimuli for further processing and action planning is a key function of brains of visually dominant animals. Since Koch and Ullman (1985) it is thought that potential points of interest are evaluated through early visual processing and converge on a saliency map It has been suggested for a long time that a connectivity pattern of short range excitation and long range inhibition in topographically organized visual structures could achieve saliency localization–see blob detection models for computer vision (Bretzner and Lindeberg, 1998; Lowe, 1999; Kong et al, 2013) and models of V1/LGN (Kang et al, 2003; Schwabe et al, 2006; Spratling, 2010; Zeng et al, 2011)–and target selection (Arai et al, 1994; Kopecz, 1995; Kopecz and Schöner, 1995; Trappenberg et al, 2001). The relevance of such organization has been underlined for action selection in artificial cognition (Erlhagen and Bicho, 2006; Richter et al, 2012; Sandamirskaya, 2014); hardware implementations have emerged (Millner et al, 2010) and are suggested to be an important milestone for developing complex cognition (Indiveri et al, 2009)

Objectives

Methods

Results

Discussion

Conclusion