🧠 neuroscience

The cerebellum supports two systems for understanding others in early childhood

This study reveals that the cerebellum supports two distinct neuroanatomical systems in early childhood: Crus II facilitates explicit Theory of Mind reasoning, while inferior lobule VIIB and the posterior vermis underpin nonverbal action prediction, thereby scaffolding different stages of social cognitive development.

Original authors: Manoli, A., Saglam, N., Valk, S., Grosse Wiesmann, C.

Published 2026-08-03

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Manoli, A., Saglam, N., Valk, S., Grosse Wiesmann, C.

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 ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed.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 as a bustling city where different neighborhoods handle different jobs. Some districts are like the city hall, managing complex plans and abstract thoughts, while others are like the traffic control center, watching the streets and reacting instantly to what's happening right now. For a long time, scientists thought the "city hall" of social interaction—our ability to figure out what other people are thinking, believing, or intending—was the only place that mattered. This skill is called "Theory of Mind." It's the mental superpower that lets you realize your friend is looking for their keys in the wrong drawer because they think they left them there, even though you know they're actually in the kitchen.

But there's a twist in the story. Even before kids can talk well enough to explain these complex thoughts, they seem to "get" what others are doing. A toddler might watch a character in a story and guess where that character will look for a toy, even if the character has the wrong idea about where the toy is. This happens without words, just by watching actions. The big question for scientists has been: Is this early, wordless guessing just a baby version of the same "city hall" skill, or is it a totally different system running on a different part of the brain? For years, the focus was mostly on the cerebral cortex (the wrinkly outer layer of the brain), but a tiny, walnut-shaped structure tucked at the back of the brain called the cerebellum has been getting a lot of attention lately. It's known for helping us balance and move smoothly, but researchers are starting to suspect it might also be a secret player in how we understand each other.

This paper dives into that mystery by looking at the brains of children aged 3 to 4 years old—a magical time when kids are just starting to master the art of explicit mental reasoning. The researchers used special brain scans to see if the size of different parts of the cerebellum matched up with how good the kids were at two different tasks: one where they had to verbally explain a character's false belief (the "city hall" style), and another where they had to silently guess where a character would look based on their actions (the "traffic control" style).

Here is what they found: The brain is indeed running two separate systems for understanding people, and the cerebellum is the boss of both, but in different neighborhoods.

First, for the kids who were really good at the explicit "Theory of Mind" task (the one where they had to reason about false beliefs), the researchers found a strong link to a specific area in the cerebellum called Crus II. Think of Crus II as the cerebellum's "abstract thinking" district. The bigger the gray matter volume in this spot, the better the child was at understanding that people can have thoughts that don't match reality. This area seems to team up with the brain's main "social reasoning" network in the cerebral cortex, helping kids build complex mental models of what others are thinking.

On the flip side, for the kids who were great at the nonverbal action prediction task (the silent guessing game), the story was totally different. Their success was linked to completely different spots in the cerebellum: lobule VIIB and the posterior vermis. These are like the cerebellum's "action observation" and "alertness" zones. They are the parts that light up when you watch someone move or notice something important happening. The study suggests these areas help kids predict what others will do by paying close attention to social cues and salient details, rather than by building abstract theories about their minds.

The most exciting part is that these two systems don't overlap. The "thinking" part of the cerebellum (Crus II) and the "watching/acting" part (VIIB and vermis) are in different places and connect to different parts of the brain. The "thinking" part connects with the brain's social reasoning network, while the "watching" part connects with areas that handle attention and action.

The paper also checked to make sure these results weren't just because some kids were smarter in general or had better language skills. Even after controlling for those factors, the link between the specific cerebellum spots and the specific social skills remained strong. While the findings for the "watching" system are based on one group of children and need more testing to be absolutely certain, the results for the "thinking" system were confirmed in a second, independent group of kids, making that finding very robust.

In short, this research suggests that the cerebellum isn't just one tool for social understanding; it's a Swiss Army knife with different blades. One blade helps us build complex theories about what others are thinking, while another blade helps us instantly read the room and predict actions based on what we see. This discovery helps us understand that our social brains are built from multiple, distinct systems that develop side-by-side, each supported by its own unique corner of the cerebellum.

Problem Statement
The cerebellum is increasingly recognized as a contributor to Theory of Mind (ToM), the capacity to infer others' mental states. However, its specific role in the development of social cognition remains poorly understood. While explicit ToM (verbal reasoning about false beliefs) typically emerges between ages 3 and 5, preverbal infants and young children already demonstrate the ability to predict others' actions nonverbally. It is unclear whether the cerebellum supports only mature, explicit mental-state reasoning or if it also underpins these earlier-emerging, nonverbal forms of social understanding. Furthermore, it is unknown whether these distinct social cognitive abilities rely on shared or dissociable cerebellar substrates, paralleling known dissociations in the cerebral cortex between the canonical ToM network and salience/action-observation networks.

Methodology
The study investigated structural cerebellar correlates of explicit ToM and nonverbal action prediction in children aged 3–4 years, a critical period for the emergence of explicit ToM.

Datasets: The primary analysis utilized the ToMAP dataset ( $N=37$ ), comprising T1-weighted MRI and behavioral data from 3–4-year-olds. An independent replication dataset (Richardson et al., $N=41$ ) was used for explicit ToM findings. A third dataset (Baby Connectome Project, $N=31$ ) was used to examine joint attention as a precursor.
Behavioral Tasks:
- Explicit ToM: Assessed via an out-of-scanner false belief task (predicting actions based on a character's false belief).
- Nonverbal Action Prediction: Assessed via an anticipatory-looking eye-tracking task where children predicted where an agent would search for an object based on the agent's false belief.
- Controls: Analyses controlled for biological sex, chronological age, total cerebellar volume, and broader cognitive abilities (executive function, language, general cognition). Performance on one task was controlled for when analyzing the other to ensure specificity.
Imaging Analysis:
- Cerebellar Processing: Cerebellar regions were isolated and normalized to the SUIT template using the SUIT toolbox. Voxel-based morphometry (VBM) with Jacobian modulation was performed to examine gray matter volume associations with task performance.
- Cerebral Processing: Cerebral cortex was segmented and normalized using CAT12.
- Covariance Analysis: General linear models tested for interactions between cerebellar gray matter volume (in specific regions of interest) and task performance to identify cerebello-cerebral structural covariance patterns.
- Validation: Findings were contextualized against adult cerebellar atlases (MDTB) and resting-state networks. Robustness for the non-replicated nonverbal prediction findings was assessed via leave-one-out and bootstrap resampling.

Key Results

Dissociable Cerebellar Substrates:
- Explicit ToM: Higher performance on explicit false belief tasks was associated with greater gray matter volume specifically in left Crus II. This finding was replicated in the independent Richardson dataset. These clusters overlapped significantly with adult maps of the ToM network.
- Nonverbal Action Prediction: Higher performance was associated with greater gray matter volume in distinct, non-overlapping clusters located in bilateral lobule VIIB and the posterior vermis. These regions correspond to adult maps of action observation and salience processing networks.
Dissociable Cerebello-Cerebral Covariance:
- ToM System: The interaction between Crus II volume and ToM performance revealed structural covariance with canonical cerebral ToM regions, including the right temporoparietal junction (TPJ) and bilateral superior temporal sulcus (STS).
- Nonverbal Prediction System: The interaction between lobule VIIB/posterior vermis volume and nonverbal action prediction performance revealed covariance with cerebral regions implicated in salience processing, action perception, and socio-affective functions (e.g., cingulate cortex, prefrontal cortex, premotor cortex, and somatosensory cortex). These patterns did not overlap with the ToM covariance clusters.
Specificity: All associations remained significant after controlling for general cognitive, executive, and linguistic abilities, as well as performance on the alternate social cognitive task.

Significance and Claims
The paper claims to identify neuroanatomically distinct cerebellar substrates that support two separable components of early social cognition:

A system anchored to explicit mental-state reasoning (Crus II), which appears to scaffold the emergence of mature ToM through coordinated maturation with the canonical cerebral ToM network.
A system supporting earlier-emerging nonverbal action prediction (lobule VIIB and posterior vermis), which is grounded in the processing of salient social cues and action observation, linking to salience-related cerebral networks.

The authors argue that these findings position the cerebellum as a key contributor to multiple stages of social cognitive development, rather than solely to mature reasoning. By demonstrating that distinct cerebellar systems underpin different forms of social understanding, the study suggests that the cerebellum may play a disproportionate role in shaping the maturation of connected cerebral networks during early life. The authors note that these findings may help refine neurodevelopmental models of conditions characterized by atypical social functioning, such as autism, by identifying specific cerebellar pathways relevant to these disorders. The study concludes that distinct cerebellar systems likely scaffold the emergence of mature social understanding, moving beyond the view of the cerebellum as a unitary contributor to social cognition.

More like this