Separating cognitive and motor processes in the behaving mouse

Munib A. Hasnain, Jaclyn E Birnbaum, Juan Luis Ugarte Nunez, Emma Hartman, Chandramouli Chandrasekaran, Michael N. EconomobioRxiv 2023 · DOI

Abstract

The cognitive processes supporting complex animal behavior are closely associated with ubiquitous movements responsible for our posture, facial expressions, ability to actively sample our sensory environments, and other critical processes. These movements are strongly related to neural activity across much of the brain and are often highly correlated with ongoing cognitive processes, making it challenging to dissociate the neural dynamics that support cognitive processes from those supporting related movements. In such cases, a critical issue is whether cognitive processes are separable from related movements, or if they are driven by common neural mechanisms. Here, we demonstrate how the separability of cognitive and motor processes can be assessed, and, when separable, how the neural dynamics associated with each component can be isolated. We establish a novel two-context behavioral task in mice that involves multiple cognitive processes and show that commonly observed dynamics taken to support cognitive processes are strongly contaminated by movements. When cognitive and motor components are isolated using a novel approach for subspace decomposition, we find that they exhibit distinct dynamical trajectories. Further, properly accounting for movement revealed that largely separate populations of cells encode cognitive and motor variables, in contrast to the 'mixed selectivity' often reported. Accurately isolating the dynamics associated with particular cognitive and motor processes will be essential for developing conceptual and computational models of neural circuit function and evaluating the function of the cell types of which neural circuits are composed.

Introduction: The "Contamination" Problem in Neuroscience

A central goal in systems neuroscience is to isolate the neural dynamics underlying cognition—processes like decision-making, memory, and planning. However, animals are rarely perfectly still. Even during controlled experiments, mice exhibit "uninstructed movements"—tiny twitches of the whiskers, shifts in posture, or adjustments of the jaw.

These movements are not just noise; they are often highly correlated with task variables, such as an animal's upcoming choice. This creates a significant "contamination" problem: when a researcher observes a ramping neural signal during a delay period, is it a cognitive signal of "urgency," or is it simply the neural drive for a slow preparatory movement? Traditional trial-averaging methods often fail to resolve this, as they smear cognitive and motor signals together.

Experimental Framework: The Two-Context Task

To address this, researchers developed a novel behavioral paradigm where head-fixed mice switch between two contexts: a Delayed-Response (DR) task, requiring memory and planning, and a Water-Cued (WC) task, requiring context monitoring.

Overview of the behavioral task, uninstructed movements, and neural performance.

By recording from the Antero-Lateral Motor cortex (ALM) using high-density Neuropixels probes, the team captured the activity of hundreds of neurons simultaneously while tracking every nuance of the animal's behavior with high-speed video.

Methods: Untangling the Subspaces

The core innovation of this study is a single-trial subspace decomposition method. Rather than assuming cognitive and motor processes occur at different times, the researchers identified two distinct neural subspaces: the Potent Subspace (activity directly related to movement) and the Null Subspace (activity related to internal cognitive states).

Kinematic tracking of jaw, nose, and paw movements used to define neural subspaces.

This decomposition was performed on a moment-to-moment basis within single trials, identifying periods when the animal was truly stationary versus when it was moving. This model-free approach allowed the team to isolate cognitive signals from the "fidgets" that usually obscure them.

Key Insights and Results

The "Urgency" Signal is Embodied

One of the most striking findings concerned "ramping" activity, long considered a hallmark of cognitive timing or planning. The study revealed that this ramping was almost entirely confined to the potent (motor) subspace. When the mouse was quiet, the ramping disappeared, suggesting it represents preparatory movement rather than a pure cognitive sense of urgency.

Context is Persistent and Quiet

In contrast, the representation of "context" (knowing which task the mouse was in) remained robust in the null (cognitive) subspace, even when the animal was completely still. This suggests that context is a truly internal cognitive variable.

Challenging "Mixed Selectivity"

The study also challenges the prevailing theory of "mixed selectivity," which posits that individual neurons encode a messy mixture of many variables. By properly accounting for uninstructed movements, the researchers found that separate populations of neurons actually handle cognitive and motor variables. The proportion of units confined to the potent subspace was significantly higher than expected by chance ($p = 6 \times 10^{-9}$), suggesting the "mix" seen in previous studies was largely an artifact of unmeasured movements.

Movement Decodes Choice

Remarkably, uninstructed movements alone could predict the mouse's upcoming choice with high accuracy ($R^2 \approx 0.41–0.54$) long before the "Go" cue, further emphasizing how deeply cognitive plans are reflected in physical behavior.

Conclusion and Future Directions

This work demonstrates that cognitive and motor processes are more separable than previously thought. By isolating these dynamics, neuroscientists can build more accurate computational models of how neural circuits function. Future work will need to expand this approach to other brain regions and explore the functional role of these uninstructed movements—determining which are merely incidental and which might actually support the cognitive task at hand.

This summary is based on research involving 13 mice and 1,651 recorded units in the ALM, utilizing high-speed videography and high-density electrophysiology. MATLAB code for subspace identification is available on GitHub at economolab/subspaceID.

Summary updated

Abstract

The cognitive processes supporting complex animal behavior are closely associated with ubiquitous movements responsible for our posture, facial expressions, ability to actively sample our sensory environments, and other critical processes. These movements are strongly related to neural activity across much of the brain and are often highly correlated with ongoing cognitive processes, making it challenging to dissociate the neural dynamics that support cognitive processes from those supporting related movements. In such cases, a critical issue is whether cognitive processes are separable from related movements, or if they are driven by common neural mechanisms. Here, we demonstrate how the separability of cognitive and motor processes can be assessed, and, when separable, how the neural dynamics associated with each component can be isolated. We establish a novel two-context behavioral task in mice that involves multiple cognitive processes and show that commonly observed dynamics taken to support cognitive processes are strongly contaminated by movements. When cognitive and motor components are isolated using a novel approach for subspace decomposition, we find that they exhibit distinct dynamical trajectories. Further, properly accounting for movement revealed that largely separate populations of cells encode cognitive and motor variables, in contrast to the 'mixed selectivity' often reported. Accurately isolating the dynamics associated with particular cognitive and motor processes will be essential for developing conceptual and computational models of neural circuit function and evaluating the function of the cell types of which neural circuits are composed.
BibTeX
@misc{hasnain2023separating,
  author = {Munib A. Hasnain and Jaclyn E Birnbaum and Juan Luis Ugarte Nunez and Emma Hartman and Chandramouli Chandrasekaran and Michael N. Economo},
  title = {{Separating cognitive and motor processes in the behaving mouse}},
  year = {2023},
  doi = {10.1101/2023.08.23.554474},
  howpublished = {bioRxiv},
}
Figure