Abstract
The dorsal premotor cortex (PMd) is a critical hub for somatomotor decision-making. While its general role is established, the internal laminar organization of decision-related signals has remained unclear. This study investigates the temporal dynamics and spatial distribution of neural activity across PMd layers in macaques performing a color discrimination task. We identify a functional continuum of neurons: "Increased" neurons, primarily in superficial layers, exhibit activity patterns consistent with evidence accumulation (decision variables). In contrast, "Decreased" and "Perimovement" neurons, more prevalent in deeper layers, are associated with action execution. These findings reveal a laminar hierarchy where decision-related processing in superficial layers precedes motor-related signals in deeper layers.
Introduction
When faced with a choice—such as deciding whether to brake or accelerate at a traffic light—the brain must transform sensory evidence into a motor plan. The dorsal premotor cortex (PMd) is central to this "somatomotor decision" process. However, the cortex is not a uniform block; it is organized into distinct layers (laminae) with specific input-output profiles. Understanding how decision-making labor is divided among these layers is essential for mapping the brain's computational architecture.
Methods and Experimental Design
Researchers recorded neural activity from two rhesus macaques (Monkeys T and O) using linear multi-contact electrodes (U-probes). These 16-channel probes allowed for simultaneous recording across the cortical depth of the PMd, providing a vertical slice of activity.
The monkeys performed a visual discrimination task where they viewed a checkerboard of red and green squares and were required to reach for a target matching the dominant color. The difficulty was varied by changing the "color coherence" (the ratio of red to green squares).
The team categorized neurons based on a "Visuomotor Continuum Index," which measured how firing rates correlated with reaction times (RT). They also employed Drift Diffusion Models (DDM) and Urgency Gating Models (UGM) to link behavioral metrics (speed and accuracy) to neural firing patterns.
Key Insights: The Laminar "Layer Cake"
The study revealed that PMd neurons are organized along a functional spectrum rather than being a homogeneous population.
1. The Visuomotor Spectrum
Neurons in the PMd exist on a bidirectional continuum.
- Increased Neurons (51.6%): These cells ramp up their firing rates ~150 ms after stimulus onset. Their activity levels correlate with choice, stimulus difficulty, and reaction time—the hallmarks of a "decision variable."
- Decreased Neurons (14.15%): These cells reduce their firing rates after the stimulus appears.
- Perimovement Neurons (34.25%): These cells remain quiet during the decision phase and only fire ~150 ms before the physical reach begins.
2. Anatomical Distribution
A striking finding was the physical segregation of these functional types. "Increased" neurons (the "thinkers") were significantly more concentrated in the superficial layers (Layers 2/3). Conversely, "Decreased" and "Perimovement" neurons (the "doers") were found more frequently in the deeper layers (Layers 5/6). A cubic regression showed a strong relationship ($R^2 = 0.87$) between cortical depth and the visuomotor index.
3. Temporal Flow
Information does not appear everywhere at once. Choice-related signals emerged in the superficial layers approximately 15.5 ms earlier than in the deep layers. This suggests a top-down flow of information within the PMd, where the decision is formed in superficial layers before being passed to deeper layers for execution.
Results and Behavioral Correlation
The behavioral data, comprising over 237,000 trials, confirmed that the monkeys' performance was highly sensitive to stimulus difficulty. Psychometric curves showed that accuracy improved as color coherence increased.
Neural population decoding further supported the laminar difference: the "discrimination time" (the point at which a choice can be predicted from neural activity) was consistently shorter in the superficial layers compared to the deep layers.
Discussion and Limitations
This research provides a clear anatomical framework for PMd function. It suggests that superficial layers integrate sensory evidence to reach a decision, while deeper layers manage the transition to motor output.
However, the study is primarily correlational. While the timing suggests a superficial-to-deep flow, further research using causal interventions (like optogenetics) is needed to prove that superficial activity drives deep-layer responses. Additionally, the specific role of "Decreased" neurons—perhaps acting as a "brake" on movement until the decision is finalized—remains an intriguing area for future study.
Conclusion
By mapping the functional properties of neurons to their physical location in the cortical layers, this study moves us closer to a circuit-level understanding of how the brain makes choices. The PMd is not just a motor area; it is a sophisticated, layered processor that bridges the gap between seeing and doing.
Summary updated