Robust Encoding of Stimulus-Response Mapping by Neurons in Visual Cortex

Academic Background

In the field of neuroscience, the visual cortex is considered a core area for processing visual information. The traditional view holds that neurons in the visual cortex primarily encode perceptual-related information, such as the location, shape, and color of stimuli. However, increasing evidence shows that activity in the visual cortex is modulated by behavioral factors such as attention, reward expectation, and working memory. These modulations are generally thought to be related to selecting specific sensory information to achieve behavioral goals. Nevertheless, whether the visual cortex participates in encoding more abstract behavioral variables, such as stimulus-response mapping rules (stimulus-response mapping), remains controversial.

Stimulus-response mapping rules refer to the association of specific visual stimuli with corresponding behavioral responses. For example, in spatial working memory tasks, subjects need to remember the location of a visual cue and make corresponding eye movements according to task rules. The traditional view holds that this complex mapping rule is mainly encoded by higher brain areas such as the prefrontal cortex, while the visual cortex is primarily responsible for processing low-level sensory information. However, through experiments, this study demonstrates that neurons in the visual cortex not only encode the location of remembered cues during spatial working memory tasks but also significantly reflect stimulus-response mapping rules. This discovery challenges the traditional view and reveals a broader role of the visual cortex in behavioral control.

Paper Source

This paper was jointly completed by Donatas Jonikaitis, Ruobing Xia, and Tirin Moore from HHMI and the Department of Neurobiology at Stanford University School of Medicine. It was published on February 24, 2025, in the journal PNAS (Proceedings of the National Academy of Sciences), titled “Robust encoding of stimulus–response mapping by neurons in visual cortex”. This research received funding from multiple institutions, including the National Institutes of Health (NIH).

Research Process and Results

Research Design and Experimental Procedure

This study used two rhesus monkeys (Monkey AQ and Monkey HB) as experimental subjects. The experimental design included two tasks: the “Look Task” and the “Avoid Task”. In both tasks, the monkeys needed to remember the location of a visual cue and make corresponding eye movement responses according to different rules.

1. Look Task: Monkeys were required to remember the location of the cue and, after the delay period, make an eye movement response toward the target matching the cue’s location.
2. Avoid Task: Monkeys were required to remember the location of the cue and, after the delay period, make an eye movement response toward the target not matching the cue’s location.

During the experiment, researchers recorded neuronal activity in the V4 region of the monkeys’ visual cortex using linear array electrodes. In each session, the number of recorded neurons ranged from 16 to 32, totaling 1,442 neurons.

Experimental Results

1. Task-dependent neuronal activity: In the look task, V4 neurons showed significant activation during the delay period, especially when the cue location was within the neuron’s receptive field (RF). However, in the avoid task, neuronal activity during the delay period was significantly reduced, even lower than the non-RF cue condition. This indicates that the delay-period activity of V4 neurons reflects not only the remembered cue location but also depends significantly on the task rule.

2. Spatial tuning correlation: During the initial presentation of the visual cue, the spatial tuning of V4 neurons was highly correlated between the two tasks (r = 0.99, p < 0.001). However, during the delay period, the spatial tuning became significantly negatively correlated between the two tasks (r = -0.36, p < 0.001). This result further supports that V4 neurons encode task rules during the delay period rather than just the remembered cue location.

3. Decoding analysis: Researchers used a support vector machine (SVM) algorithm to decode the delay-period activity of V4 neurons. The results showed that the decoder accuracy was significantly higher in the look task compared to the avoid task (look task: median = 59.4%, avoid task: median = 57.7%, p = 0.014). Additionally, the decoder performance in cross-task tests was significantly below chance level, indicating that the activity of V4 neurons indeed depends on the task rule.

4. Role of the frontal eye field (FEF): To investigate the source of delay-period activity in V4, researchers locally inactivated the FEF. The results showed that FEF inactivation significantly reduced the selectivity of V4 neurons during the delay period. In the look task, delay-period selectivity decreased by 31% (p < 0.001), and in the avoid task, it decreased by 50% (p < 0.001). This result supports the regulatory role of the FEF on V4 neuronal activity.

Conclusion and Significance

This study experimentally demonstrated that neurons in the V4 region of the visual cortex not only encode the location of remembered cues during spatial working memory tasks but also significantly reflect stimulus-response mapping rules. This finding challenges the traditional view that the visual cortex primarily processes low-level sensory information, while more abstract behavioral rules are encoded by higher brain regions such as the prefrontal cortex. The study also revealed the regulatory role of the FEF on V4 neuronal activity, further supporting the significant influence of motor signals on visual cortical activity.

Scientific Value and Application Value

1. Scientific Value: This study provides new insights into understanding the role of the visual cortex in behavioral control, revealing that the visual cortex is not only involved in processing sensory information but also plays an important role in encoding more abstract behavioral rules. This discovery helps deepen the understanding of visual cortex functions and offers new directions for future neuroscience research.

2. Application Value: The findings of this study may have significant implications for fields such as neural engineering and artificial intelligence. For example, in developing vision control systems based on brain-computer interfaces, understanding how the visual cortex encodes stimulus-response mapping rules can help design more efficient algorithms and models.

Research Highlights

1. Key Findings: V4 neurons’ delay-period activity not only reflects the location of remembered cues but also significantly depends on task rules, showing enhanced activity in the look task and suppressed activity in the avoid task.

2. Methodological Innovation: This study used linear array electrodes for multi-channel recordings and combined SVM algorithms to decode neuronal activity, providing high-precision neural coding analysis.

3. Research Significance: This study challenges traditional views and reveals the broader role of the visual cortex in behavioral control, offering new perspectives on how the brain integrates sensory information and behavioral rules.

Other Valuable Information

This study also revealed the regulatory role of the FEF on V4 neuronal activity through FEF inactivation experiments, further supporting the significant influence of motor signals on visual cortical activity. This finding provides new evidence for understanding the functional connectivity between the prefrontal cortex and the visual cortex.

Through rigorous experimental design and in-depth data analysis, this study reveals the important role of the visual cortex in behavioral control, providing new ideas and directions for future neuroscience research.