Dynamic tracking of social variables in simultaneous brain recordings of socially interacting monkeys

By recording wireless neural activity from two interacting macaques in a natural setting, researchers discovered that high-level visual cortex neurons dynamically track social variables and that the recipient of grooming, rather than the groomer, drives the social interaction through neural encoding of the partner's actions and brain activity.

The Gist

Imagine two monkeys, let's call them M1 and M2, hanging out in a room. They aren't strapped to chairs or forced to stare at a screen. They are free to move, talk (with gestures), and do what monkeys do best: groom each other.

Now, imagine both of them are wearing high-tech "brain backpacks" that record their thoughts and neural activity wirelessly, in real-time, without stopping them from being monkeys.

This paper is the story of what the scientists found when they listened to the brains of these two friends while they interacted naturally. Here is the breakdown of their discoveries, explained simply.

1. The "Social Bank Account" (Grooming Reciprocity)

Monkeys are very social. A common behavior is allogrooming (picking bugs off each other's fur). Usually, if Monkey A grooms Monkey B for 5 minutes, Monkey B feels a social "debt" and will eventually groom Monkey A back for about 5 minutes. It's like a tit-for-tat exchange.

The Discovery:
The scientists found that the monkeys' brains were keeping a running tally of this "social debt."

• The Metaphor: Think of a social bank account.
  • When M2 grooms M1, M1's "account" goes into the negative (they received a favor).
  • When M1 grooms M2 back, M1's account goes positive (they gave a favor).
• The Surprise: The part of the brain responsible for this math wasn't the "motor" part (the part that moves the hands). It was the visual cortex (the part that sees).
• Why? The scientists guess that the brain keeps this "favor score" in the visual area so it doesn't get mixed up with the actual physical act of grooming. It's like keeping your bank balance in a separate notebook so you don't accidentally spend the money while you're just thinking about it.

2. Who is Driving the Bus? (The Receiver is the Boss)

You might think the monkey doing the grooming (the groomer) is in charge, because they are the one moving their hands and making decisions.

The Discovery:
The data showed the exact opposite. The monkey receiving the grooming was actually the one driving the interaction.

• The Metaphor: Imagine a dance. You might think the person leading the dance (the groomer) is the boss. But in this case, the person being led (the receiver) is actually pulling the strings.
• How? The receiver would shift their body, turn their head, or use a hand gesture to say, "Hey, scratch my left ear, not my right one."
• The Brain Evidence: The groomer's brain activity was actually tracking the receiver's body movements more closely than their own. The groomer was essentially saying, "I am moving my hands to match where you want me to go." The receiver was the conductor; the groomer was just the orchestra following the baton.

3. The "Identity Badge" (Knowing Who is Who)

Even when the monkeys weren't looking directly at each other (for example, when one was closing their eyes to enjoy the grooming), their brains still knew exactly who was there.

The Discovery:
The visual part of the brain (IT cortex) kept a constant, strong signal identifying the partner.

• The Metaphor: It's like wearing a glowing ID badge in your mind. Even if you close your eyes, your brain still "sees" the person standing next to you. The monkeys didn't need to stare at their friend's face to know, "Oh, that's M2 grooming me." Their brains had already locked onto that identity.

4. The "Mirror Dance" (Mirroring Movements)

Scientists have long looked for "mirror neurons"—cells that fire when you do something and when you watch someone else do it.

The Discovery:
They found these neurons, but with a twist.

• The Metaphor: Imagine a mirror that sometimes flips the image.
  • Sometimes, when the monkey saw its partner move a shoulder, its own brain fired up as if it were moving that shoulder (a classic mirror neuron).
  • But often, the brain did the opposite: when the partner moved, the monkey's brain quieted down for that specific movement.
• Why? This might be a safety mechanism. If your friend moves their hand toward your face, your brain might suppress the urge to move your own hand in the same way to avoid bumping into them. It's a way of saying, "I see what you are doing, so I won't do the exact same thing right now."

5. The "Brain-to-Brain Wi-Fi" (Interbrain Coupling)

Finally, the scientists looked at how the two brains talked to each other.

The Discovery:
The brains were synchronized, but the signal flowed in a specific direction.

• The Metaphor: Think of a Wi-Fi connection.
  • Usually, we think the person doing the action (the groomer) sends the signal.
  • But here, the "signal" was stronger from the receiver to the groomer.
  • The groomer's brain was essentially "tuning in" to the receiver's brain frequency. The receiver was the broadcaster; the groomer was the receiver.

The Big Picture

This study is a big deal because, for the first time, scientists listened to two monkeys' brains while they were free to be themselves, not stuck in a lab cage.

The Main Takeaway:
Social interaction isn't just about one person acting and the other reacting. It's a dynamic dance where the person being groomed (the receiver) is actually the one setting the pace, directing the groomer's hands, and keeping the "social score" in their head. The brain is constantly calculating favors, tracking identities, and syncing up with the friend, all while the monkeys are just living their best, natural lives.

Technical Summary

1. Problem Statement

Understanding the neural basis of primate social interactions has been historically limited by technical constraints. Previous studies typically recorded brain activity under highly artificial, restrained conditions (e.g., monkeys interacting with partners while tethered or in fixed positions). These constraints fail to capture the dynamic, naturalistic nature of social behaviors. Furthermore, most studies have focused on single-brain recordings, leaving the mechanisms of interbrain coupling (how two brains synchronize during interaction) largely unexplored in primates. The authors aim to overcome these limitations by recording simultaneous, wireless neural activity from two monkeys interacting freely in a naturalistic setting to decode how social variables (reciprocity, identity, movement) are encoded and how the two brains coordinate.

2. Methodology

Subjects and Setup:

• Subjects: Two adult male bonnet macaques (Macaca radiata), M1 and M2, who were socially compatible (M2 dominant, M1 subordinate).
• Implants: Each monkey was implanted with 256 microelectrodes (Floating Microelectrode Arrays) targeting three specific regions:
  • Inferior Temporal (IT) Cortex: High-level visual processing (object/face recognition).
  • Ventral Premotor Cortex (PMv): Motor planning and mirror neuron systems.
  • Ventrolateral Prefrontal Cortex (PFC): Executive control and social behavior (implanted only in M1).
• Recording: Simultaneous, wireless neural recordings were conducted while the monkeys interacted freely for ~11 minutes. The session included three phases: M2 grooming M1, M1 grooming M2, and M2 grooming M1 again.
• Behavioral Tracking: High-speed video (30 fps) from multiple angles was synchronized with neural data. Joints (head, neck, shoulders) were manually annotated to track kinematics.
• Control Tasks: To interpret neural signals, monkeys performed controlled touchscreen tasks separately:
  • Visual Task: Fixation on images of themselves, their partner, and a third monkey.
  • Motor Task: Reaching for targets with left/right hands.

Analytical Approaches:

• Social Favor Signal: A "grooming surplus" metric was calculated (Duration Giving – Duration Receiving) to track reciprocity.
• Decoding: Linear classifiers were trained on neural population activity to decode grooming states (giving, receiving, none), partner identity, and hand usage.
• Predictive Modeling: Linear regression models were built to predict a neuron's firing rate based on:
  • Self-joint positions.
  • Partner-joint positions.
  • Partner's neural population activity (Interbrain coupling).
• Spectral Analysis: Local Field Potentials (LFP) and Multi-Unit Activity (MUA) power spectra were analyzed across grooming states.

3. Key Contributions

1. First Simultaneous Wireless Recording in Natural Interaction: Demonstrated the feasibility of recording high-density neural data from two freely interacting primates without physical constraints.
2. Neural Basis of Grooming Reciprocity: Identified a specific "social favor" signal encoded in the brain that tracks the net balance of grooming given versus received.
3. Receiver-Driven Interaction: Challenged the assumption that the actor (groomer) drives the interaction. Found that the receiver of grooming drives the social dynamics, influencing both the groomer's neural activity and the partner's brain activity.
4. Interbrain Coupling Dynamics: Revealed that interbrain coupling is dynamic and directionally specific, with the groomer's brain activity being predicted by the receiver's brain activity.

4. Key Results

A. Social Favor and Reciprocity

• Grooming Surplus Encoding: Neurons in the IT cortex (visual) of both monkeys showed a strong, continuous correlation with the "grooming surplus" signal (net time giving minus receiving).
  • Correlation: $r = 0.65$ for M1 and $r = 0.5$ for M2 in IT.
  • Specificity: This signal was strongest in IT, weaker in PMv, and absent in PFC. This suggests the visual cortex maintains a "ledger" of social debt, possibly to avoid interference with motor output regions.
• Reciprocity: The monkeys exhibited near-perfect reciprocity (M2 groomed M1 for 244s; M1 groomed M2 for 230s), suggesting the neural tracking of surplus facilitates this balance.

B. Neural Tracking of Grooming State and Identity

• State Decoding: A linear decoder could accurately classify the ongoing grooming state (Giving, Receiving, No Grooming) from neural activity in all regions (IT, PMv, PFC) with accuracy significantly above chance.
• Partner Identity: IT neurons robustly decoded the partner's identity throughout the interaction, even when the monkey was not directly looking at the partner (e.g., while receiving grooming). This indicates a stable internal representation of the social partner.

C. Self vs. Partner Movement Tracking

• Dissociation of Tracking:
  • While Giving Grooming: The groomer's neural activity was predicted better by the receiver's joint positions than by their own.
  • While Receiving Grooming: The receiver's neural activity was predicted better by their own joints.
• Interpretation: The receiver actively directs the interaction (e.g., by shifting position to indicate where to be groomed), forcing the groomer to track the receiver's movements.
• Mirror Neuron Tuning: A subset of neurons showed negative correlations between self-joint and partner-joint tuning weights. This suggests that observing a partner's movement may inhibit similar self-movements, a mechanism potentially useful for distinguishing self from other.

D. Interbrain Coupling

• Directionality: The groomer's brain activity was significantly predicted by the receiver's brain activity (PMv/PFC), but not vice-versa.
• Dynamic Coupling: Interbrain coupling fluctuated dynamically, peaking during transitions between grooming states and within bouts.
• Conclusion: The social interaction is driven by the receiver, who dictates the flow, and the groomer's brain synchronizes to the receiver's state.

E. Spectral Dynamics

• Alpha Power: Increased alpha band power (8–12 Hz) was observed in IT and PMv during receiving grooming, correlating with behaviors like closing eyes or looking up, suggesting a state of relaxation.
• Beta Power: Increased beta power was associated with active motor engagement during giving grooming.

5. Significance

• Ecological Validity: This study moves social neuroscience beyond artificial constraints, demonstrating that complex social variables (reciprocity, identity, joint tracking) are encoded in the brain during natural, unrestrained behavior.
• Mechanism of Reciprocity: It provides a concrete neural mechanism for how primates maintain social balance (reciprocity) via a "social favor" signal in the visual cortex, rather than just motor or prefrontal regions.
• Reversal of Agency: The finding that the receiver drives the interaction (both behaviorally and neurally) challenges traditional views where the actor is the primary driver. It highlights the active role of the recipient in shaping social dynamics through referential gestures and body positioning.
• Interbrain Synchrony: It establishes that interbrain coupling in primates is not a static phenomenon but a dynamic process driven by the specific roles (giver vs. receiver) within the interaction, offering a new framework for studying "two-brain" systems in natural settings.

In summary, the paper elucidates a rich, dynamic neural substrate for primate social interaction, showing that the brain continuously tracks social debts, partner identity, and kinematic cues, with the receiver of social contact acting as the primary driver of the interaction's neural and behavioral flow.