Distributed correlates of visually-guided behavior across the mouse brain

Nicholas A. Steinmetz, Peter Zatka-Haas, Matteo Carandini, Kenneth D. HarrisbioRxiv 2018 · DOI

TL;DR

This study used high-density neural recordings in mice to map how brain-wide activity supports visually-guided behavior. They found that while visual stimuli activate a broad network of neurons, the specific signals for movement and choice are encoded by more specialized, yet distributed, neural populations, with a competitive midbrain circuit playing a key role in adjudicating decisions.

What Problem Does This Paper Address?

Understanding how the brain translates sensory input into complex behaviors like decision-making and action is a fundamental challenge. While we know that neuronal activity underlies behavior, it's unclear which neurons across the brain are involved, how their activity is distributed, and how it unfolds over time. Previous research often focused on specific brain regions, but the interconnected nature of the brain suggests that goal-directed behaviors likely involve widespread neural coordination. This paper aims to systematically survey neural activity across a large portion of the mouse brain during a visual decision-making task to map these distributed correlates.

Methods

The researchers employed a sophisticated experimental setup:

  • High-Density Neural Recordings: They used Neuropixels probes, which allow simultaneous recording from thousands of neurons across multiple brain regions. This enabled them to capture a broad picture of neural activity.
  • Two-Alternative Unforced Choice (2AUC) Task: Mice were trained to discriminate the contrast of visual stimuli presented on a screen and indicate their choice by turning a wheel. This task allowed for controlled presentation of visual stimuli and measurement of behavioral output.
  • Kernel Regression: To disentangle the contributions of different task components (visual stimulus, movement, choice) to neuronal activity, they used a reduced-rank kernel regression model. This statistical approach allowed them to model how different temporal aspects of the task influenced neural firing rates.
  • Brain-Wide Anatomical Mapping: After recordings, they meticulously mapped the precise anatomical locations of their recorded neurons within the mouse brain, correlating electrophysiological data with histological reconstructions.

Datasets

The study utilized data collected from mice performing a visual contrast discrimination task. No new benchmark datasets were introduced; rather, the paper presents novel data collected using the described experimental setup.

Key Insights and Ideas

  • Widespread Task Engagement: Performing the visual task, even when passive, engaged a significant portion (~60%) of neurons across nearly all 42 recorded brain regions, far exceeding the activation seen during passive visual stimulation. This highlights a global brain state shift associated with task engagement.
  • Distributed Movement Representation: Signals related to upcoming movement were found in nearly all brain regions, suggesting a widespread neural representation of motor preparation.
  • Rare but Distributed Choice Encoding: Neurons specifically encoding the upcoming choice (left vs. right) were surprisingly rare but found in a distinct set of distributed regions, including frontal cortex, striatum, and midbrain.
  • Competitive Midbrain Circuit for Decision Adjudication: Midbrain neurons showed a unique pattern of encoding choices: they were strongly activated before contralateral choices and suppressed before ipsilateral choices. This suggests a competitive, inhibitory circuit within the midbrain that may be crucial for resolving the decision.
  • Task Engagement Modulates Neural Excitability: The brain-wide state shift observed during task engagement, characterized by increased activity in subcortical regions like the basal ganglia and midbrain, might reflect enhanced neural excitability that amplifies stimulus-driven activity and facilitates decision-making.

Key Results

  • Approximately 60% of recorded neurons (out of ~30,000) were modulated during the task.
  • Nearly all 42 recorded brain regions showed neuronal activity related to the task.
  • Neurons selective for choice (left vs. right) were rare, found primarily in midbrain, striatum, and frontal cortex.
  • Midbrain choice-selective neurons exhibited directionally-opposed activity, increasing for one choice and decreasing for the other, suggesting competitive inhibition.
  • Movement-related activity was found in nearly all brain regions.
  • Passive stimulus presentation activated a more restricted set of regions (visual cortex, thalamus, striatum, superior colliculus) compared to active task performance.

Limitations and Future Work

  • Observational Nature: The study is observational, meaning it identifies correlations between neural activity and behavior. While it provides strong evidence for the involvement of specific circuits, it cannot definitively prove causal relationships without further manipulation (e.g., optogenetics).
  • Mouse Model: While mice are excellent models for neuroscience, there are always questions about the direct translatability of findings to other species, particularly humans.
  • Future Directions: The authors suggest that future work could explore the causal roles of these identified circuits using techniques like optogenetic manipulation. Further investigation into the specific mechanisms of competitive coding in the midbrain and the nature of the brain-wide state shift associated with task engagement are also promising avenues. The study's findings motivate a shift in focus from isolated brain regions to understanding coordinated neural circuits for behavior.

Summary updated

Abstract

Behavior arises from neuronal activity, but it is not known how the active neurons are distributed across brain regions and how their activity unfolds in time. Here, we used high-density Neuropixels probes to record from ~30,000 neurons in mice performing a visual contrast discrimination task. The task activated 60% of the neurons, involving nearly all 42 recorded brain regions, well beyond the regions activated by passive visual stimulation. However, neurons selective for choice (left vs. right) were rare, and found mostly in midbrain, striatum, and frontal cortex. Those in midbrain were typically activated prior to contralateral choices and suppressed prior to ipsilateral choices, consistent with a competitive midbrain circuit for adjudicating the subject’s choice. A brain-wide state shift distinguished trials in which visual stimuli led to movement. These results reveal concurrent representations of movement and choice in neurons widely distributed across the brain.
BibTeX
@misc{steinmetz2018distributed,
  author = {Nicholas A. Steinmetz and Peter Zatka-Haas and Matteo Carandini and Kenneth D. Harris},
  title = {{Distributed correlates of visually-guided behavior across the mouse brain}},
  year = {2018},
  doi = {10.1101/474437},
  howpublished = {bioRxiv},
}
Figure