Visual Experience Reduces the Spatial Redundancy Between Cortical Feedback Inputs and Primary Visual Cortex Neurons
In a study titled “Visual Experience Reduces the Spatial Redundancy between Cortical Feedback Inputs and Primary Visual Cortex Neurons,” Rodrigo F. Dias, Radhika Rajan, and their team explored how visual experience influences the spatial redundancy of cortical feedback pathways. This research was carried out by scientists from the Champalimaud Neuroscience Programme at the Champalimaud Foundation in Lisbon, Portugal, and published in the journal Neuron on October 9, 2024. The study focused on the feedback circuits within the visual cortex of mice, investigating how visual experience alters the input organization from higher visual areas (lateromedial region, LM) to the primary visual cortex (V1).
Background and Motivation
Visual cognition is a complex process encompassing the processing of external perceptions and the integration of higher cognitive feedback. Current research suggests that feedback from higher cortical areas can modulate neural activity in the primary visual cortex according to specific contexts and expectations. Visual perception is believed to be achieved through a hierarchical network involving multi-level processing and feedback, yet the integration of information between different levels and the mechanisms for achieving this remain incompletely understood. Existing literature indicates that feedback information from the LM area can accurately match certain neural activity patterns in the primary visual cortex V1 and is influenced by experience. The topological structure of these feedback circuits is related to visual experience, but its mechanisms are unclear. Through this study, Dias et al. aimed to reveal the effects of visual experience on the organizational structure of the feedback pathways between the LM area and V1 in mice, addressing how higher cortical areas regulate the perception of the primary cortex based on visual experience.
Research Process and Methods
To explore the impact of visual experience on LM to V1 feedback organization, the research team divided mice into different rearing environments: completely reared in darkness (dark-reared postnatal day 0, DRP0), exposed to light conditions for the first 21 days post-birth, then shifted to darkness (dark-reared postnatal day 21, DRP21), and reared under normal light cycles (normally reared, NR). Utilizing fluorescence labeling and two-photon microscopy, the research team recorded the activity of LM neuronal axons at layer 1 (L1) of the V1 under different conditions. The LM area was injected with fluorescent protein GCaMP6s and red calcium indicator jRGECO1a to label the neuronal activities of LM and V1, and signal imaging was used to verify labeling accuracy.
The experiment employed a moving grating stimulus method to stimulate the opposite visual field of the mice. By measuring the overlap of receptive fields (RF) between LM inputs and V1 neurons, the research quantified the spatial redundancy of feedback information. The team also developed a computational model to simulate how visual experience changes the feedback organization between LM and V1 neurons by minimizing receptive field overlap.
Experimental Results
1. Visual Experience Reduced Feedback Redundancy from LM to V1
The study revealed that visual experience does not alter the overall topological structure of LM inputs, as LM inputs and V1 neurons maintain spatial matching even in mice reared in darkness. However, in normally reared mice, the spatial matching between LM feedback inputs and V1 neurons was relatively lower, implying that more LM input information transmits distant background information rather than local information, a phenomenon not evident in completely dark-reared mice. Further data analysis suggested that an increase in visual experience reduces the spatial overlap of feedback inputs, thereby achieving the deduplication of visual information.
2. Differences in Feedback Pathways at Different Levels
The study also found significant differences in spatial redundancy among feedback pathways originating from different levels (L2/3 and L5) of the LM area. Feedback inputs from L2/3 layers were more concentrated in adjacent areas of V1, while inputs from L5 layers tended to transmit distant background information. In normally reared mice, L5 layer inputs exhibited experience-dependent directional selectivity. Researchers observed that neurons in the L5 layer had an advantage in transmitting peripheral visual information over those in the L2/3 layer, suggesting that different levels of feedback pathways may play distinct functional roles in visual processing.
3. Experience-Dependent Organization of Directional Selectivity
Further analysis indicated that the LM inputs from the L5 layer are organized in an experience-dependent manner according to their directional preference for gratings. Experimental results showed that neurons in the L5 layer with a preference for vertical or horizontal orientations differ in spatial positioning, and visual experience enhances the spatial organization of this directional selectivity, leading to reduced longitudinal distribution of feedback inputs with a vertical preference. This phenomenon was not observed in the L2/3 layer, indicating that visual experience mainly affects the organization of directional selectivity in the L5 layer.
4. Computational Model Validation
The research team’s computational model successfully simulated the effects of visual experience on the LM-V1 feedback pathway. By selecting non-overlapping feedback inputs, the model achieved the deduplication of visual information, confirming that the experience-dependent organization within the LM feedback pathway can be realized by reducing receptive field overlap. This finding supports the theory of the role of cortical feedback inputs in predictive coding, indicating that feedback circuits form specific learning patterns through selective reduction in connections with downstream neurons to predict joint activities of higher and lower neurons.
Research Conclusion and Significance
The study by Dias et al. demonstrated that visual experience plays a key role in reducing the spatial redundancy of LM to V1 feedback pathways. This deduplication process may achieve experience-dependent selective organization, providing an important physiological basis for explaining how the brain integrates higher visual information through feedback mechanisms. The study not only uncovered the fine organizational structure of feedback projections in the visual cortex but also demonstrated how visual experience shapes the uniqueness of this feedback organization through specific mechanisms. The results support the theory that feedback inputs in hierarchical computation need to learn to predict lower neuron activities, providing a physiological mechanism for how the brain optimizes visual information processing by reducing spatial redundancy between neurons. This discovery promises to provide new insights for studies on the neural mechanisms of visual perception and could offer theoretical foundations for applications such as visual injury recovery and perceptual training.
Research Highlights
- Spatial Redundancy Minimization Effect of Visual Experience: Visual experience can reduce the overlap of receptive fields between LM inputs and V1 neurons, achieving deduplication of visual information.
- Differentiated Roles of Hierarchical Feedback Circuits: Different levels (L2/3 and L5) of feedback input exhibit functional differences in transmitting distant or nearby information, suggesting specific roles for different feedback circuits in visual perception.
- Experience-Dependent Organization of Directional Selectivity: LM feedback inputs in the L5 layer are organized spatially based on directional selectivity in an experience-dependent manner.
- Computational Model Validation: The model successfully simulated the experience-dependent mechanism of feedback organization, supporting the predictive coding function of visual cortex feedback circuits in information processing.
The discoveries from this research offer a new perspective on understanding the feedback circuits of the visual cortex, illustrating how visual experience shapes neuronal organization, and providing essential experimental support for the hierarchical and experience-dependent nature of neural networks in visual perception.