Learning Peri-saccadic Remapping of Receptive Field from Experience in Lateral Intraparietal Area

Xiao Wang, Yan Wu, Mingsha Zhang, Si Wu

Introduction

The human visual system maintains a stable perception of the world despite frequent, rapid eye movements known as saccades. These movements, occurring several times per second, cause the retinal image to shift abruptly across the sensor. Perceptual stability is attributed to predictive remapping, a process where neurons in the Lateral Intraparietal (LIP) area begin responding to stimuli in their future receptive fields (FRF) even before a saccade is completed.

While the "Cortical Wave" (CW) model has been proposed to explain this phenomenon through activity propagation across the cortex, the developmental origins of the necessary neural circuitry remain unclear. This study investigates whether these structured connections can emerge through visual experience using biologically plausible learning rules.

Methods and Model Architecture

The researchers developed a Spiking Neural Network (SNN) to simulate the development of remapping circuits. The model consists of two primary layers: the LIP Layer, representing the parietal cortex where visual remapping occurs, and the SC Layer, representing the Superior Colliculus, which provides the Corollary Discharge (CD)—an internal copy of the motor command.

The neural architecture of the Cortical Wave model, showing the interaction between LIP neurons, Superior Colliculus (SC) input, and visual input.

The model utilized Adaptive Exponential Integrate-and-Fire (AdEx IAF) neurons and incorporated Short-Term Plasticity (STP) to stabilize the emergent cortical waves. The network begins with "tabula rasa" connectivity—random and bidirectional connections in the LIP layer and non-selective connections from the SC. Learning is driven by Spike-Timing-Dependent Plasticity (STDP), which modifies synaptic strength based on the relative timing of pre- and post-synaptic firing. Training involves simulated "visual sweeps," where the movement of a static image across the retina during a saccade provides a consistent temporal sequence of neuronal activation.

Key Insights: The Role of Visual Experience

A central finding of the paper is that the "passive sweep" of images across the retina during eye movements serves as the primary training signal. Because the image always moves in the opposite direction of the saccade, the consistent timing allows STDP to prune random connections into the unidirectional paths required for the CW model.

The Corollary Discharge (CD) signal acts as a temporal gate, ensuring that synaptic strengthening only occurs during actual eye movements. This prevents the network from being erroneously modified by objects moving independently in the environment. Furthermore, the model demonstrates that the duration of the CD signal naturally encodes saccade amplitude, determining how far the cortical wave propagates.

Key Results

The model successfully learned the unidirectional connectivity patterns necessary for predictive remapping in both 1D and 2D simulations. Key findings include:

  • Receptive Field Elongation: The model successfully reproduced the "transient elongation" of receptive fields seen in biological experiments. During the peri-saccadic window, neurons responded to stimuli across a stretched area covering the current (CRF), future (FRF), and intermediate locations (IML).
Conceptual diagram of elongated remapping, showing the transition from the current receptive field (CRF) to the future receptive field (FRF).
  • Temporal Alignment: Remapping in the model occurred within a window of approximately 100ms around saccade onset, matching biological data observed in macaque monkeys.
  • Directional Generalization: The 2D model demonstrated that the brain can learn to remap in any direction (up, down, left, right) by strengthening specific sub-networks corresponding to different saccade vectors.

Conclusion and Future Directions

This research demonstrates that the complex circuitry required for visual stability does not need to be entirely hardwired; it can emerge from the statistical regularities of visual experience. However, the model currently relies on a pre-structured CD signal and a simplified two-layer architecture. Future work should explore how motor signals are calibrated in tandem with visual remapping and how this system operates in dynamic environments where objects may move independently of eye movements.

Summary updated

Abstract

Our eyes move constantly at a frequency of 3–5 times per second. These movements, called saccades, induce the sweeping of visual images on the retina, yet we perceive the world as stable. It has been suggested that the brain achieves this visual stability via predictive remapping of neuronal receptive field (RF). A recent experimental study disclosed details of this remapping process in the lateral intraparietal area (LIP), that is, about the time of the saccade, the neuronal RF expands along the saccadic trajectory temporally, covering the current RF (CRF), the future RF (FRF), and the region the eye will sweep through during the saccade. A cortical wave (CW) model was also proposed, which attributes the RF remapping as a consequence of neural activity propagating in the cortex, triggered jointly by a visual stimulus and the corollary discharge (CD) signal responsible for the saccade. In this study, we investigate how this CW model is learned naturally from visual experiences at the development of the brain. We build a two-layer network, with one layer consisting of LIP neurons and the other superior colliculus (SC) neurons. Initially, neuronal connections are random and non-selective. A saccade will cause a static visual image to sweep through the retina passively, creating the effect of the visual stimulus moving in the opposite direction of the saccade. According to the spiking-time-dependent-plasticity rule, the connection path in the opposite direction of the saccade...
BibTeX
@article{wang2017learning,
  author = {Xiao Wang and Yan Wu and Mingsha Zhang and Si Wu},
  title = {{Learning Peri-saccadic Remapping of Receptive Field from Experience in Lateral Intraparietal Area}},
  journal = {Frontiers Comput. Neurosci.},
  year = {2017},
  doi = {10.3389/fncom.2017.00110},
}
Figure