Hierarchical feature binding in a spiking neural network model of the primate ventral visual pathway
7 affiliations
- 1Department of Comparative Biomedical Sciences, School of Veterinary Medicine, University of Surrey, Guildford, United Kingdom
- 2Centre for Neural Circuits and Behaviour, University of Oxford, Oxford, United Kingdom
- 3Independent Researcher, Guildford, United Kingdom
- 4Department of Electrical and Electronic Engineering, Imperial College London, South Kensington, London, United Kingdom
- 5Department of Bioengineering, Imperial College London, South Kensington, London, United Kingdom
- 6The Surrey Institute for People-Centred AI, University of Surrey, Guildford, United Kingdom
- 7Centre for Theoretical Neuroscience and Artificial Intelligence, Department of Experimental Psychology, University of Oxford, Oxford, United Kingdom
Abstract
Feature binding - how the brain encodes which features are part of other features to form representations of the coherent objects we perceive - remains an unsolved problem in neuroscience. Despite progress towards a solution, major theories either lack detailed explanations at the neuronal level or rely on biologically unrealistic simplifications, and none adequately account for the representation of hierarchical information, which is crucial to our perception of the world. To address this, a solution termed binding by polychrony has been proposed to explain how hierarchical feature relationships may be encoded at the neuronal level in a biologically realistic system. This theory relies on a phenomenon known as polychronization, where groups of neurons fire in precisely coordinated, time-locked sequences, leading to the emergence of regularly repeating spatiotemporal patterns that might encode these relationships. In this study, we explore binding by polychrony through simulations of a spiking neural network that closely aligns with the structural organisation of the primate ventral visual pathway, incorporating bottom-up, top-down, and lateral synaptic connections. By exposing the network to collections of related 2D object shapes from ecologically realistic datasets and applying spike-timing-dependent plasticity, the network self-organises such that individual neurons respond selectively to specific shape features. Furthermore, the network exhibits polychronization, giving rise to repeating spatiotemporal patterns, some of which form circuits that encode hierarchical feature relationships. Notably, these circuits are robust, even with the randomised, Poisson-distributed spike timings that represent the visual stimuli in the input layer. These results provide evidence for binding by polychrony as a feasible solution to the feature binding problem, and characterise the mechanism by which it may function. This mechanism can guide experimentalists in identifying such circuits in vivo, and could also be utilised in computer vision systems to capture more information and improve robustness to adversarial inputs. Author summary: How the brain pieces together visual details into coherent objects, known as the feature binding problem, remains an unresolved conceptual challenge in neuroscience. Existing theories either lack biological realism or do not adequately capture the hierarchical relationships among visual features. To address this shortcoming, we conducted detailed simulations with biologically realistic neuron models organised into networks reflecting the structure and connectivity of the primate ventral visual pathway. When exposed to ecologically realistic datasets of 2D object shapes and trained through unsupervised competitive learning, the networks self-organised such that neurons became selectively responsive to specific shape features. Crucially, we observed that groups of neurons developed precise, regularly repeating spatiotemporal firing patterns, a phenomenon known as polychronization. These patterns support a novel and biologically plausible mechanism through which neurons encode hierarchical relationships among visual features represented across different spatial scales within the network. This hierarchical feature binding mechanism advances our understanding of how complex visual information may be assembled in the brain. These insights could also inform the development of AI vision systems that interpret natural scenes more robustly and efficiently, similar to biological visual systems.
Related
- A hierarchical framework of representational emergence: From feature binding to supramodal emergence
- Neurodynamical heterogeneous spiking neural network for spatiotemporal cognitive signal processing
- A Spiking Neural Network Model of Elementary Self-Consciousness via Endogenous Default Mode Network Dynamics
- Polyaxonal A1 amacrine cells form a spatially extended network in primate retina
- Phase-locked dual-band vibration bursts strengthen anti-kindling by vibrotactile coordinated reset stimulation in a spiking network model
BibTeX
@article{gardner2026hierarchical,
title = {{Hierarchical feature binding in a spiking neural network model of the primate ventral visual pathway}},
author = {Brian Gardner and Patrick T. McCarthy and Joseph Chrol-Cannon and Dan F. M. Goodman and Simon R. Schultz and Giovanni Lo Iacono and Simon M. Stringer},
journal = {PLOS Computational Biology},
year = {2026},
month = sep,
doi = {10.1371/journal.pcbi.1014752},
url = {https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014752},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Hierarchical feature binding in a spiking neural network model of the primate ventral visual pathway
AU - Brian Gardner
AU - Patrick T. McCarthy
AU - Joseph Chrol-Cannon
AU - Dan F. M. Goodman
AU - Simon R. Schultz
AU - Giovanni Lo Iacono
AU - Simon M. Stringer
PY - 2026
DA - 2026/09/16
JO - PLOS Computational Biology
DO - 10.1371/journal.pcbi.1014752
UR - https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014752
AB - Feature binding - how the brain encodes which features are part of other features to form representations of the coherent objects we perceive - remains an unsolved problem in neuroscience. Despite progress towards a solution, major theories either lack detailed explanations at the neuronal level or rely on biologically unrealistic simplifications, and none adequately account for the representation of hierarchical information, which is crucial to our perception of the world. To address this, a solution termed binding by polychrony has been proposed to explain how hierarchical feature relationships may be encoded at the neuronal level in a biologically realistic system. This theory relies on a phenomenon known as polychronization, where groups of neurons fire in precisely coordinated, time-locked sequences, leading to the emergence of regularly repeating spatiotemporal patterns that might encode these relationships. In this study, we explore binding by polychrony through simulations of a spiking neural network that closely aligns with the structural organisation of the primate ventral visual pathway, incorporating bottom-up, top-down, and lateral synaptic connections. By exposing the network to collections of related 2D object shapes from ecologically realistic datasets and applying spike-timing-dependent plasticity, the network self-organises such that individual neurons respond selectively to specific shape features. Furthermore, the network exhibits polychronization, giving rise to repeating spatiotemporal patterns, some of which form circuits that encode hierarchical feature relationships. Notably, these circuits are robust, even with the randomised, Poisson-distributed spike timings that represent the visual stimuli in the input layer. These results provide evidence for binding by polychrony as a feasible solution to the feature binding problem, and characterise the mechanism by which it may function. This mechanism can guide experimentalists in identifying such circuits in vivo, and could also be utilised in computer vision systems to capture more information and improve robustness to adversarial inputs. Author summary: How the brain pieces together visual details into coherent objects, known as the feature binding problem, remains an unresolved conceptual challenge in neuroscience. Existing theories either lack biological realism or do not adequately capture the hierarchical relationships among visual features. To address this shortcoming, we conducted detailed simulations with biologically realistic neuron models organised into networks reflecting the structure and connectivity of the primate ventral visual pathway. When exposed to ecologically realistic datasets of 2D object shapes and trained through unsupervised competitive learning, the networks self-organised such that neurons became selectively responsive to specific shape features. Crucially, we observed that groups of neurons developed precise, regularly repeating spatiotemporal firing patterns, a phenomenon known as polychronization. These patterns support a novel and biologically plausible mechanism through which neurons encode hierarchical relationships among visual features represented across different spatial scales within the network. This hierarchical feature binding mechanism advances our understanding of how complex visual information may be assembled in the brain. These insights could also inform the development of AI vision systems that interpret natural scenes more robustly and efficiently, similar to biological visual systems.
SN - 1553-7358
ER -