Reaching for domain-general syntax: Sentence processing and tool-use reach-to-grasp share neural patterns in the basal ganglia

This study demonstrates that the basal ganglia serve as a domain-general hub for hierarchical computations by exhibiting overlapping neural patterns during the syntactic processing of complex sentences and the reach-to-grasp phase of tool use, but not during stable object holding.

The Gist

The Big Question: Is Building a Sentence Like Building a Tower with Tools?

Imagine your brain is a busy construction site. Usually, we think of language (building sentences) and action (using tools) as two completely different construction crews working in separate buildings. One crew builds with words; the other builds with hands and hammers.

But this study asks a fascinating question: What if these two crews are actually using the same blueprint and the same foreman?

The researchers wanted to know if the part of your brain that helps you understand complex sentences is the same part that helps you figure out how to use a tool, like a pair of pliers, to grab an object.

The Experiment: The "Pliers vs. Pen" Test

To find out, the researchers put 40 people inside an MRI scanner (a giant camera that takes pictures of the brain in action). They gave the participants two jobs:

1. The Language Job: They had to read sentences. Some were simple ("The writer admires the poet"), and some were tricky "twisted" sentences where the order of words was swapped ("The writer that the poet admires...").
2. The Motor Job: They had to move a peg from one spot to another on a board. Sometimes they used their bare hand (like a pen), and sometimes they had to use a pair of pliers (a tool).

The researchers broke the "pliers" job down into three specific moments:

• The Start: Just thinking about grabbing the pliers.
• The Reach: Moving the hand with the pliers toward the peg.
• The Grab: Actually closing the pliers around the peg and moving it.

The Discovery: The "Brain Foreman"

The study found a specific area deep inside the brain called the Basal Ganglia. Think of this area as the Site Foreman or the Project Manager of the brain.

Here is what they discovered:

1. The "Twisted" Sentence Connection: When people read the tricky, complex sentences, the "Site Foreman" (Basal Ganglia) got very active.
2. The "Tool" Connection: When people used the pliers, the Foreman also got active.
3. The Perfect Match: The magic happened when they looked at when the Foreman was working.
   • The Foreman was not very busy when the person was just thinking about picking up the pliers.
   • The Foreman was not very busy when the person was just holding the peg steady.
   • BUT, the Foreman went into overdrive during the Reach-to-Grasp phase. This is the moment when the hand with the pliers is moving toward the target and has to figure out exactly how to grab it.

The Analogy:
Imagine you are building a tower of blocks.

• Simple sentences are like stacking blocks in a straight line. Easy.
• Complex sentences are like stacking blocks where one block is hidden inside another block before you can place the next one. It requires a mental "fold" or "nesting."
• Using a tool is the same. Your hand isn't touching the object; the tool is. Your brain has to mentally "fold" the tool into your hand and then figure out how that new "super-hand" will touch the object.

The study found that the exact same moment your brain is doing this mental "folding" for the tool (reaching for the target) is the exact same moment it is doing the mental "folding" for the complex sentence.

Why Does This Matter?

This suggests that our brains didn't evolve language and tool use separately. Instead, they likely evolved a single, super-powerful tool (the Basal Ganglia) that handles hierarchy.

• Hierarchy means organizing things in layers (like a Russian nesting doll).
• In language, we nest clauses inside sentences.
• In action, we nest the tool inside the hand, which then acts on the object.

The researchers found that this "hierarchy manager" is most active right when the complexity peaks: when you are figuring out how to grab something with a tool, or when you are untangling a tricky sentence.

The Takeaway

Your brain doesn't have a "Language Department" and a "Tool Department." Instead, it has a Master Architect deep in the center of your brain. This architect specializes in complex structures.

Whether you are trying to understand a sentence like "The dog that the cat chased ran away" or trying to figure out how to grab a screw with a screwdriver, your brain uses the same ancient, powerful machinery to solve the puzzle. It turns out that speaking like a poet and working like a carpenter might just be the same brain activity in disguise.

Technical Summary

1. Problem Statement and Hypothesis

The study investigates the long-standing debate regarding the evolutionary and cognitive relationship between human language (specifically syntax) and complex motor skills (specifically tool use). While previous research suggests that both domains rely on shared neural resources, particularly within the basal ganglia (BG), it remains unclear which specific components of a tool-use action share these resources with syntactic processing.

• Theoretical Framework: The authors propose that tool use involves a hierarchical structure similar to linguistic syntax. Just as a tool is "embedded" into a motor plan (forming a hand-tool unit), a relative clause is embedded into a main sentence. This "embedding" creates a non-canonical, hierarchical structure.
• The Gap: Tool use can be decomposed into distinct phases: Initiation (planning), Reach-to-Grasp (approaching and establishing contact with the target), and Object Manipulation (moving the target). It is unknown if the neural overlap with syntax occurs across all phases or is specific to the phase where the hierarchical relationship between the tool and the target is dynamically updated.
• Hypothesis: The authors predict that the reach-to-grasp phase (where the tool must be articulated with the target object) will show the strongest neural overlap with complex syntactic processing (object-relative clauses), more so than initiation or manipulation phases.

2. Methodology

Participants:

• 40 healthy French native speakers (mean age ~23) underwent fMRI scanning.

Experimental Design:
The study employed a within-subjects design involving two primary tasks performed inside a 3T MRI scanner:

1. Syntactic Comprehension Task:

   • Stimuli: Sentences containing center-embedded relative clauses.
   • Conditions:
     • Subject-Relative (SR): Canonical structure (e.g., "The writer that admires the poet writes the paper").
     • Object-Relative (OR): Non-canonical, complex structure (e.g., "The writer that the poet admires writes the paper").
   • Procedure: Participants viewed sentences via RSVP and judged the truth value of a subsequent affirmation.
   • Goal: Isolate neural activity associated with processing syntactic complexity (OR > SR).

2. Motor Task (Tool Use vs. Free Hand):

   • Stimuli: Moving a peg on a pegboard.
   • Conditions:
     • Free Hand: Grasping the peg directly.
     • Tool Use: Grasping the peg using a 30-cm pair of pliers.
   • Phase Decomposition: The motor action was temporally segmented into three distinct phases based on kinematic data:
     • Initiation: From the "Go" signal to movement onset (planning).
     • Reach-to-Grasp: From movement onset to contact with the peg.
     • Object Manipulation: From grasping to moving the peg to the target location.
   • Goal: Isolate neural activity specific to tool use (Tool > Free Hand) for each phase.

Neuroimaging Analysis:

• Preprocessing: Standard pipeline (fMRIPrep) including motion correction, slice-timing correction, and co-registration to MNI space.
• Univariate Analysis (GLM):
  • Contrasted OR vs. SR sentences.
  • Contrasted Tool vs. Free Hand for each of the three motor phases.
  • Identified Regions of Interest (ROIs) in the Basal Ganglia (Caudate, Putamen) and Inferior Frontal Gyrus (IFG).
• Multivariate Analysis (Representational Similarity):
  • Computed Cosine Similarity between neural activation patterns.
  • Tested if the pattern of OR sentences was more similar to Tool Use patterns than to Free Hand patterns (and compared against SR sentences).
  • Focus was placed on the BG ROIs identified in the univariate analysis.

3. Key Results

Behavioral Data:

• Syntax: Participants were significantly slower and less accurate (lower $d'$) processing Object-Relative clauses compared to Subject-Relative clauses, confirming the increased cognitive load of the complex syntax.
• Motor: Tool use required significantly longer times for all three phases (Initiation, Reach-to-Grasp, Manipulation) compared to free-hand movements.

Univariate fMRI Results:

• Syntax: Processing Object-Relative clauses significantly activated the bilateral Caudate Nuclei (Basal Ganglia). No significant difference was found in the Left Inferior Frontal Gyrus (IFG) between the two syntactic conditions, though both activated the IFG relative to baseline.
• Tool Use Phases:
  • Initiation: Activated a left-lateralized parieto-temporal network and the Putamen.
  • Reach-to-Grasp: Activated a bilateral fronto-parietal network and the Caudate Nuclei.
  • Manipulation: Activated visual/occipital regions; no significant BG activation was found for this phase compared to free hand.

Co-localization and Overlap (The Core Finding):

• Caudate Nuclei: The bilateral Caudate clusters activated by complex syntax (OR > SR) showed significantly higher activity during the Tool Reach-to-Grasp phase compared to Free-Hand Reach-to-Grasp.
• Specificity: No significant overlap was found between syntax and the Initiation phase (Putamen) or the Manipulation phase.
• Multivariate Similarity:
  • In the Right Caudate, the neural activation patterns for Object-Relative sentences were significantly more similar to the patterns for Tool Reach-to-Grasp than to Free-Hand Reach-to-Grasp or Subject-Relative sentences.
  • This similarity was not observed for the Initiation or Manipulation phases.

4. Key Contributions

1. Phase-Specific Convergence: The study provides the first evidence that the neural overlap between language and action is not global across a tool-use movement but is phase-specific. The overlap occurs specifically during the reach-to-grasp phase, where the tool is dynamically integrated with the target object.
2. Basal Ganglia as a Domain-General Hub: The findings reinforce the role of the basal ganglia (specifically the caudate) as a core hub for domain-general hierarchical computations, rather than just a motor sequencing center.
3. Refutation of General Cognitive Load: By showing that the overlap is specific to the reach-to-grasp phase and not the initiation or manipulation phases (which also involve cognitive load), the study argues against the hypothesis that the overlap is merely due to general working memory or executive load.
4. Neural Pattern Similarity: The use of multivariate analysis demonstrated that the fine-grained spatial distribution of neural activity is shared between complex syntax and tool reach-to-grasp, suggesting a common computational mechanism rather than just co-activation.

5. Significance and Implications

• Evolutionary Link: The results support the "syntax-action" hypothesis, suggesting that the neural machinery for handling hierarchical structures (embedding) evolved to support both tool use and language. The basal ganglia may have been a pre-existing system for combining simple elements into structured sequences, which was later co-opted for language.
• Mechanism of Tool Use: The study clarifies that the critical cognitive leap in tool use occurs not just in planning (initiation) or execution (manipulation), but in the dynamic integration of the tool as an effector during the reach-to-grasp phase. This phase requires updating the body schema and managing the "double function" of the tool (as both an object held by the hand and an effector acting on the target).
• Clinical Relevance: The findings offer a potential neural explanation for why patients with basal ganglia disorders (e.g., Parkinson's) or children with Developmental Language Disorders (DLD) often exhibit deficits in both complex syntax and coordinated tool use. It suggests these deficits stem from a shared impairment in processing hierarchical interdependence.
• Theoretical Shift: The study challenges the view that the Inferior Frontal Gyrus (IFG) is the sole domain-general processor for hierarchy. Instead, it posits that the IFG may handle domain-specific aspects, while the phylogenetically older basal ganglia handle the fundamental, domain-general computations of hierarchical embedding.