🧠 neuroscience

Simultaneous neuron evidence for much higher covariation with saccadic reaction time of superior colliculus than primary visual cortex visual responses

This study demonstrates that while single-neuron visual responses in the superior colliculus (SC) predict saccadic reaction times better than those in the primary visual cortex (V1), the predictive power of SC responses improves significantly with the addition of simultaneously recorded neurons, whereas V1 responses show little improvement except for dark stimuli, suggesting that SC visual signals are functionally reformatted from V1 inputs to better drive motor behavior.

Original authors: Yu, Y., Hafed, Z. M.

Published 2026-06-29

📖 3 min read☕ Coffee break read

CC BY 4.0

Original authors: Yu, Y., Hafed, Z. M.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine your brain is a massive, high-speed newsroom trying to decide when to send a messenger (your eye) to look at something new. Two key departments are involved in this process: the Primary Visual Cortex (V1), which acts like the "Raw Data Desk" receiving the initial photos, and the Superior Colliculus (SC), which acts like the "Action Command Center" that actually tells the eyes to move.

For a long time, scientists knew that if a single worker in the Action Command Center (SC) got a very strong signal, the messenger would leave quickly. If the signal was weak, the messenger would take longer. This link was much stronger in the Command Center than in the Raw Data Desk (V1). But here was the big mystery: In the real world, these departments don't work with just one person at a time; they have entire teams working simultaneously. So, if you look at the whole team in the Raw Data Desk, does their combined effort finally help predict when the messenger will leave?

The Experiment
The researchers set up a classic "look-and-point" game. They watched 1 to 10 workers in both departments at the exact same time while the subject looked at visual targets. They tried to build a "prediction machine" using the combined strength of the signals from these workers to guess exactly how fast the eye would move.

The Findings
Here is what they discovered, using a simple analogy:

The Single Worker Test: Just like before, a single strong signal from the Command Center (SC) was a much better crystal ball for predicting the eye movement speed than a single signal from the Raw Data Desk (V1).
The Teamwork Test (The Big Surprise): When they added more workers to the mix, the results split dramatically.
- In the Command Center (SC): Adding more workers was like adding more engines to a rocket. The more workers they included, the much better the prediction became. The team worked together perfectly to forecast the action.
- In the Raw Data Desk (V1): Adding more workers was like adding more people to a choir where everyone is singing slightly different tunes. No matter how many workers they added, the prediction didn't get much better. The team's combined effort didn't help predict the eye movement any better than a single person could.

The One Exception
There was one specific scenario where the Raw Data Desk (V1) team started to show some improvement: when the target was a dark object against a gray background. In this specific case, adding more workers did help the prediction, though it still couldn't match the Command Center's performance. The researchers suggest this might be because our ancestors needed to be extra good at spotting dark shapes in the wild, so this specific type of signal gets a little extra coordination.

The Bottom Line
Even though the Raw Data Desk (V1) and the Command Center (SC) receive similar visual information, they process it differently. The Command Center doesn't just copy the Raw Data; it reformats it. It takes the raw visual signals and reorganizes them into a highly coordinated team effort that is directly wired to control our eye movements. The Raw Data Desk, on the other hand, seems to keep its signals more independent, making it harder to predict movement just by looking at the whole team's activity.

Technical Summary

Problem Statement
Recent findings have established that the trial-by-trial visual response strength in the primate superior colliculus (SC) inversely correlates with saccadic reaction time (SRT) significantly more strongly than visual response strength in the primary visual cortex (V1). However, a critical gap remains in understanding how V1 visual responses predict SRT when accounting for the fact that populations of neurons in both brain areas are concurrently activated by a single visual stimulus onset. Specifically, it is unclear whether the predictive power of V1 activity changes when multiple simultaneously recorded neurons are considered, as opposed to single-unit analyses.

Methodology
The study employed a classic visually-guided saccade task to assess the predictive quality of trial-by-trial SRT based on the visual response strengths of 1 to 10 simultaneously recorded neurons in both the SC and V1. For each recording session, the researchers modeled SRT as a weighted linear combination of the visual response strengths of $N$ simultaneously recorded neurons. This approach allowed for a direct comparison of how prediction accuracy scales with the number of neurons included in the model for each brain area.

Key Results

Single-Neuron Comparison: Consistent with prior work, the visual response strength of a single SC neuron was a superior predictor of reaction time compared to that of a single V1 neuron.
Scaling with Population Size: As the number of simultaneously recorded neurons increased, the prediction quality for SRT improved substantially in the SC. In contrast, prediction quality in V1 did not show a comparable improvement with the addition of more neurons.
Stimulus Specificity in V1: An increase in V1 prediction quality with more neurons was observed only under specific conditions: 100% contrast dark stimuli (stimuli darker in luminance than the surrounding gray background). Even in this specific condition, the improvement in V1 remained substantially weaker than the gains observed in the SC.

Significance and Claims
The paper concludes that despite qualitative similarities between the visual responses of the SC and V1, the visual responses in the SC appear to be functionally reformatted from their V1 counterparts. The authors suggest that the specific increase in V1 prediction quality for dark contrasts may reflect the ecological relevance of dark contrasts in natural scenes. Ultimately, the results indicate that the SC visual response strength carries a much stronger and more scalable relationship with saccadic reaction time than the concurrent visual response strength in V1.

Technical Summary

More like this