Pattern recognition with figure-ground separation by generation of coherent oscillations

doi:10.1016/0893-6080(94)90055-8

Neural Networks

Volume 7, Issue 1, 1994, Pages 49-63

https://doi.org/10.1016/0893-6080(94)90055-8 Get rights and content

Abstract

For elucidating the computation theory underlying flexible recognition of visual patterns by an oscillating neural network of the brain, a model with two dynamical centers was presented and the capability of the model was demonstrated with computer experiments. One of the two dynamical centers, the center for figure organization, organizes figures from elementary visual signals by self-organizing coherent dynamics of neural oscillators, being separated from background represented by incoherent dynamics. The other center, the center for symbol formation, provides symbolic constraints to the dynamics of the former center to define the boundaries of the figures according to memory. Synchronized oscillations also emerge in the latter center by neural oscillators encoding elementary symbols constituting memory items. The linkers connecting the two centers are dynamically gated to generate correspondence between the symbol and the figure. Information circulation emerges between the two centers due to synchronization through linkers. It enables the pattern recognition in the presence of background independent of size, position, and some deformation.

References (36)

C.D. Gilbert
Horizontal integration in the neocortex
Trends in Neuroscience
(1985)
S. Grossberg et al.
Synchronized oscillations during cooperative feature linking in a cortical model of visual perception
Neural Networks
(1991)
Y. Kuramoto
Collective synchronization of pulse-coupled oscillators and excitable units
Physica D.
(1991)
M. Mishkin et al.
Object vision and spatial vision: Two cortical pathways
Trends in Neuroscience
(1983)
Yamaguchi, Y., & Shimizu, H. (in press). Self-organization of information in the brain. In Y. Sawada & T. Musya...
S. Amari
Characteristics of random nets of analog neuron-like elements
IEEE Transactions on Systems, Man, and Cybernetics
(1972)
R.A. Andersen et al.
Encoding spatial location by posterior parietal neurons
Science
(1985)
S.M. Ahn et al.
Steady-state and limit cycle activity of mass of neurons forming simple feedback loops (I): Lumped circuit model
Kybernetik
(1974)
J. Bullier et al.
Physiological studies on the feedback connection to the striate cortex from cortical areas 18 and 19 of the cat
Experimental Brain Research
(1988)
R. Eckhorn et al.
Coherent oscillations: A mechanism of feature linking in the visual cortex?
Biological Cybernetics
(1988)

R. Eckhorn et al.

Feature linking via stimulus-evoked oscillations: Experimental results from cat visual cortex and functional implications from a network model

R. Eckhorn et al.

Feature linking via synchronization among distributed assemblies: Simulations of results from cat visual cortex

Neural Computation

(1990)

A.K. Engel et al.

Synchronization of oscillatory neuronal responses between striate and extrastriate visual cortical areas of the cat

W.J. Freeman

Mass action in the nervous system

(1975)

K. Fukushima

A neural network model for selective attention in visual pattern recognition

Biological Cybernetics

(1986)

C.D. Gilbert et al.

Columnar specificity of intrinsic horizontal and coricocortical connections in cat visual cortex

Journal of Neuroscience

(1989)

C.M. Gray et al.

Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects global stimuli properties

Nature

(1989)

H. Haken

Synergetics

(1983)

Cited by (21)

Lee-Associator-a chaotic auto-associative network for progressive memory recalling
2006, Neural Networks
In the past few decades, neural networks have been extensively adopted in various applications ranging from simple synaptic memory coding to sophisticated pattern recognition problems such as scene analysis and robot vision. Moreover, current studies on neuroscience and physiology have reported that in a typical scene segmentation problem our major senses of perception (e.g. vision, olfaction, etc.) are highly chaotic and involved non-linear neural dynamics and oscillations. In this paper, the author proposes an innovative chaotic neural oscillator—namely the Lee-oscillator (Lee's Chaotic Neural Oscillator) to provide a chaotic neural coding and information processing scheme. To illustrate the capability of Lee-oscillators upon pattern association, a chaotic auto-associative network, namely Lee-Associator (Lee's Chaotic Auto-associator) is constructed.
Different from classical auto-associators such as the celebrated Hopfield network, which provides time-independent and static pattern association scheme, the Lee-Associator provides a remarkable progressive memory association scheme (what the author called ‘Progressive Memory Recalling Scheme, PMRS’) during the chaotic memory association. This is exactly consistent with the latest research in psychiatry and perception psychology on dynamic memory recalling schemes, as well as the implications and analogues to human perception as illustrated by the remarkable Rubin-vase experiment on visual psychology.
Dynamical cell assembly hypothesis - Theoretical possibility of spatio-temporal coding in the cortex
1996, Neural Networks
This paper is an attempt to understand how knowledge and events are represented and processed in the brain. An important point is the question of what carries information in the brain — the mean firing rate or the timing of spikes? The idea we want to pursue is that, contrary to the traditional view, the brain might use higher order statistics, which means in essence that timing of spikes plays a critical role in encoding, representing, and processing knowledge and events in the brain.
A recently revealed salient nature of cortical pyramidal cells, i.e., the high variability of inter-spike intervals suggests that a cortical neuron may function effectively as a coincidence detector. At the same time, non-classical experimental phenomena of task-related, short time-scaled dynamical modulations of temporal correlations between neurons suggest a non-classical view on the dynamics working in the brain. In response to contexts or external events, a group of neurons, a dynamical cell assembly, spontaneously organizes, linked temporarily by coincident timing of incident spikes, showing correlated firing with each other. This is an emergent property of neuronal populations in the cortex.
We make a theoretical exploration on issues as (1) the description of such emergent dynamics of dynamical cell assemblies based on the working hypothesis that a cortica neuron functions effectively as a coincidence detector, and (2) the principle of spatio-temporal coding based on the hypothetical emergent dynamics. Note that the conventional rate coding hypothesis does not give satisfactory answers to fundamental questions on the representation and processing of knowledge or events in the brain, e.g., the questions of cross-modular integration of information or the binding problem, and representation of hierarchical knowledge etc.
The first goal is to give a non-encyclopedic review on (1) the temporal structure of spike sequences, focusing on the question of the basic code in the brain; (2) the paradigms on representation of knowledge and events proposed from a theoretical or experimental basis. The classical paradigms of Hebb and Barlow with their experimental and theoretical critiques, and more recently proposed experiment-based paradigms are reviewed. Also a review is given on (3) the experimentally observed spatio-temporal structure of spike dynamics.
The second goal is to give a description of the dynamical cell assembly — the central concept in this paper. Aside from the question of physiological basis, we make a theoretical study, under a working hypothesis that a cortical neuron functions effectively as a coincidence detector, on the emergent dynamics of cell assemblies, and also examine how the observed experimental data could be explained within this theoretical setting.
We also try to give the principle of spatio-temporal coding based on the dynamical cell assembly framework. A key concept is the internal mechanism of “dialogue” among neuronal pools in the brain. This provides a dynamical foundation of bi-directional interactions for the linkage of distant modules to create integrated information. We present a simple model in order to illustrate the working principle of coincidence detector systems. Relations with other temporal coding paradigms are also discussed. Copyright © 1996 Elsevier Science Ltd.
Interareal synchronization in the visual cortex
1996, Behavioural Brain Research
The primary visual cortex (VI) is part of a highly interconnected network of cortical areas, hierarchically organized but operating concurrently across hierarchical levels. The high degree of reciprocal interconnection among visual cortical areas provides a framework for their interaction during the performance of visual scene analysis. The functional interdependency of visual cortical areas which develops during scene analysis can be investigated by techniques which measure interareal correlated activity. Evidence from monkeys performing a visual pattern discrimination suggests that synchronization of aperiodic activity from neuronal ensembles in cortical areas at different hierarchical levels is a relevant aspect of visual function. The near-periodic nature of the synchronized response to moving light bars in earlier studies may have been a result of the type of stimulus used. Various models of visual cortex are discussed in which interareal synchronization plays a functional role.
Dynamic linking among neural oscillators leads to flexible pattern recognition with figure-ground separation
1996, Neural Networks
The basic process of perceptual organization in vision is figure-ground separation. Gestalt psychologists have revealed that this separation is the product of several local coherence criteria. However, relevant information at the global level is also believed to participate in the process, although the neural basis remains unclear. To elucidate the underlying principle of figure-ground separation, we have proposed a neural network model of visual pattern recognition. This model consists of heterogeneous centers composed of nonlinear neural oscillators: a lower center for figure formation, and a higher center for symbol representation. A dynamic linking architecture generates correspondence between the two centers. Pattern recognition is attained as the formation of synchronization of neural oscillators in the two centers. In the present paper, we propose an extended version of our model. Its major improvement is that the synaptic efficacy in dynamic linking is now reversible. Dynamic linking between the two centers occurs via the reversible synapses in the form of transient temporal correlation among neural oscillators. This feature assures further flexibility and diversity of pattern recognition in the present model. Moreover, to elucidate the functional relationships between the local and global levels, bifurcation of the model's behavior was examined in terms of information flow between the two centers. The results suggest that reciprocal information flow between the two heterogeneous centers plays a signiificant role in pattern recognition.
Cortical Synchrony as a Mechanism of Collinear Facilitation and Suppression in Early Visual Cortex
2021, Frontiers in Systems Neuroscience
Neural network model for path-finding problems with the self-recovery property
2019, Physical Review E

View all citing articles on Scopus

View full text

Contributed articlePattern recognition with figure-ground separation by generation of coherent oscillations

Abstract

Trends in Neuroscience

Neural Networks

Physica D.

Trends in Neuroscience

Characteristics of random nets of analog neuron-like elements

IEEE Transactions on Systems, Man, and Cybernetics

Encoding spatial location by posterior parietal neurons

Science

Steady-state and limit cycle activity of mass of neurons forming simple feedback loops (I): Lumped circuit model

Kybernetik

Physiological studies on the feedback connection to the striate cortex from cortical areas 18 and 19 of the cat

Experimental Brain Research

Coherent oscillations: A mechanism of feature linking in the visual cortex?

Biological Cybernetics

Feature linking via stimulus-evoked oscillations: Experimental results from cat visual cortex and functional implications from a network model

Feature linking via synchronization among distributed assemblies: Simulations of results from cat visual cortex

Neural Computation

Synchronization of oscillatory neuronal responses between striate and extrastriate visual cortical areas of the cat

Mass action in the nervous system

A neural network model for selective attention in visual pattern recognition

Biological Cybernetics

Columnar specificity of intrinsic horizontal and coricocortical connections in cat visual cortex

Journal of Neuroscience

Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects global stimuli properties

Nature

Synergetics

Contributed article
Pattern recognition with figure-ground separation by generation of coherent oscillations