Abstract
Hands are regularly in sight in everyday life, and this visibility significantly impacts motor control, perception, and attention. Visual information is integrated into an internal model of sensorimotor control, which is reflected in spontaneous brain activity. This study tests whether spontaneous activity in the human somatomotor cortex is modulated by visual stimuli of hands versus non-hand stimuli. Using fMRI and multivariate pattern analysis (MVPA), we examined the similarity between resting-state activity patterns and task-evoked patterns during the presentation of natural hands, robot hands, gloves, and food. In the left somatomotor cortex, we observed a significantly stronger spatial correlation between resting-state activity and natural hand patterns compared to other stimuli. A trend analysis revealed that task-rest pattern similarity was influenced by inferred visual and motor attributes (Hand > Robot > Glove > Food). Notably, this similarity was absent in the visual cortex, suggesting that somatomotor regions maintain internal models of hand representations and their inferred use even at rest.
Introduction
For decades, neuroscientists have sought to understand "resting-state" activity—the continuous, high-energy brain activity observed even when an individual is at rest. Recent theories propose that this activity is not merely noise, but rather the maintenance of "priors," or internal models of the body and environment. These models likely help the brain predict and react to the world more efficiently.
While the somatomotor cortex is primarily associated with touch and movement, humans see their hands almost constantly. This study investigates whether the somatomotor cortex stores a visual representation of the hand. Specifically, we asked if spontaneous activity at rest "replays" the visual signature of a hand and whether this representation is sensitive to how much an object looks or acts like a real hand.
Methods
The study involved 19 healthy participants and utilized a three-phase experimental design to map brain activity and its spontaneous recurrences.
- Resting-State Scan: Participants fixated on a crosshair for eight minutes without performing any task to capture baseline spontaneous activity.
- Task-Evoked Scan: Participants viewed grayscale images from four categories: Natural Hands, Robot Hands (mechanically similar but non-living), Gloves (visually similar but lacking independent movement), and Food (control stimuli).
- Localizer Scan: A finger-tapping task identified the specific "hand notch" within each participant's somatomotor cortex.
Researchers employed Multi-Voxel Pattern Analysis (MVPA) to identify the unique neural "fingerprint" created by each stimulus category. They then quantified how often these patterns appeared spontaneously during rest using a metric called "U90" (the upper 90th percentile of correlation coefficients).
Results
The analysis revealed a clear spatial and functional specificity in how the brain represents hands at rest. The similarity between resting activity and visual stimuli was strongest in the left postcentral gyrus, specifically within the area dedicated to hand sensation and control.
A significant hierarchy of similarity emerged: Natural Hand > Robot Hand > Glove > Food ($F(3,54) = 3.469, p = .022$). This suggests the somatomotor cortex does not just respond to hand-like shapes; it integrates both visual form and inferred motor function. For instance, a robot hand, which moves like a human hand, elicited a stronger match than a static glove. This effect was lateralized to the left hemisphere, likely reflecting its dominance for motor selection and tool use in right-handed individuals.
Crucially, this task-rest similarity was not found in the early visual cortex (V1-V3). This implies that the long-term "prior" or internal model of the hand is stored specifically in the regions responsible for controlling the hand, rather than in primary visual processing areas.
Discussion and Conclusion
These findings suggest that the human brain maintains a constant "visual template" of our most essential tools for interaction—our hands—within the motor system itself. This internal model persists even in the absence of external stimuli or active movement.
While the sample size of 19 participants is standard for fMRI, future work could expand on these findings by using dynamic video stimuli to explore motor-sequence representations. Additionally, investigating how these resting templates change in individuals with limb loss or those who use advanced prosthetics could provide deep insights into neural plasticity. This study lays the groundwork for understanding how daily visual experiences shape the fundamental architecture of the motor brain, maintaining a ready-state model of the body's primary interface with the world.
Summary updated