Single Neurons Identified That Support Social Reasoning
Key findings
- "Theory of mind" is fundamental to human social interaction—the ability to form detailed mental representations of other people and understand they may hold thoughts or beliefs different from our own
- To probe the cellular mechanism of theory of mind, researchers asked 15 patients undergoing awake brain surgery to complete a standard false-belief task while they recorded the activity of single neurons in the prefrontal cortex
- Neuronal activity was able to distinguish other's beliefs from non-social physical representations and self-belief representations from other-belief representations
- The neurons provided progressively finer-grained levels of detail about others' beliefs—from whether or not another's belief was involved, to whether the beliefs were true or false, to which particular item was being considered
- This research should usher in a better understanding of human social cognition and better treatment of its disorders, such as autism and schizophrenia
Subscribe to the latest updates from Neuroscience Advances in Motion
"Theory of mind" is fundamental to human social interaction—the ability to form detailed mental representations of other people and understand they may hold thoughts or beliefs different from our own.
Functional imaging studies have revealed a network of brain areas that supports social reasoning. Now, neuroscientist Mohsen Jamali, MD, PhD, and neurosurgeon Ziv M. Williams, MD, of the Department of Neurosurgery at Massachusetts General Hospital, and colleagues have begun to investigate theory of mind at the cellular level.
As they explain in Nature, they identified single neurons that encode information about other people's beliefs across various scenarios. Moreover, these neurons can distinguish between oneself's and others' perspectives, represent the content of others' beliefs and accurately evaluate whether the beliefs are true or false.
Study Methods
The team used custom-adapted multi-electrodes to record the activity of 212 single neurons in the dorsomedial prefrontal cortex of 11 participants with motor disorders before they underwent deep brain stimulator placement. The participants heard a narrative, then answered a question verbally.
Neuronal Predictions of Another's Beliefs
Other-belief vs. Non-social Physical Trials
The researchers first compared neuronal responses to narratives that required the participant to consider another person's beliefs about reality and narratives that simply concerned a physical representation of an object. Sample narratives include:
- Other-belief trial: "You and Tom see a jar on the table. After Tom leaves, you move the jar to the cupboard. Where does Tom believe the jar is?"
- Non-social physical trial: "You take a picture of a jar on the table. After the picture, you move the jar to the cupboard. Where is the jar in the picture?"
42 neurons (20%) accurately predicted whether the participant was considering another person's beliefs (P<0.025). Decoding accuracy: 83% of the time (P<0.005), the neurons distinguished other-belief–related representations from non-social representations.
True-belief vs. False-belief Trials
To double-check whether the neuronal signals reflected the other's specific beliefs rather than simply any beliefs, the team administered trials whereanother's belief was clearly false or true. Sample narratives include:
- False-belief trial: "You and Tom see a jar on the table. After Tom leaves, you move the jar to the cupboard. Where does Tom believe the jar is?"
- True-belief trial: "You and Tom see a jar on the table. After Tom leaves, you open the jar and leave it in place. Where does Tom believe the jar is?"
49 neurons (23%) accurately predicted whether the participants were considering another's true versus false belief (P<0.025). Decoding accuracy: 78% of the time (P<0.005), the participants' neurons could encode for someone else's idea of reality.
Self's Perspective vs. Another's Perspective
Other-belief Aware vs. Unaware Trials
It was also important to test whether neuronal responses reflected variations in another person's perspective of reality independently of the participant's perspective. The researchers introduced trials in which the other person's awareness varied.
- Standard other-belief "unaware" trial: "…After Tom leaves, you move the jar to the cupboard…"
- Sample other-belief "aware" trial: "…After Tom leaves, you move the jar to the cupboard while he watches through the window…"
Neurons that accurately predicted the other's beliefs on the "unaware" trials also accurately predicted the other's beliefs on the "aware" trials (P<0.001). Decoding accuracies on these trials were similar to those decoded from the standard other-belief trials (77% vs. 78%, respectively; P=0.61) and were positively correlated on a cell-by-cell basis (r, 0.3; P=0.04).
In other words, the participant's neuronal prediction of the other's beliefs could shift to reflect the other's specific perspective.
Self-belief vs. Other-belief Trials
Four participants completed a control task in which their own belief was judged true or false. 112 neurons, different from the first 212, were recorded.
- Sample trial: "You see a jar on a table. After you leave the kitchen, the jar falls onto the floor. Where will you expect to find the jar?"
31 neurons (28%) predicted whether the participant's own imagined beliefs were true or false. These neurons were mostly distinct from neurons that predicted others' beliefs (P=0.14).
Predicting the Content of the Other's Beliefs
Using other narratives, the researchers found that neurons could accurately encode for an item's identity, location and category (objects, containers, foods, places, animals and appearances).
A Foundation for Novel Research
Mirror neurons are known to reflect information about the observable behavior of others but it has been impossible to study whether or how neurons represent another's beliefs or perspective. The neurons identified in this study may support that type of inquiry.
Furthermore, neuronal responses were largely insensitive to differences in cognitive demand, complexity or depth of reasoning required. This suggests neuronal representations of the other's beliefs are probably generalizable, a property that would be necessary for supporting theory of mind.
All told, this research should usher in a better understanding of human social cognition and better treatment of its disorders, such as autism and schizophrenia.
view original journal article Subscription may be required
Learn more about the Department of Neurosurgery
Refer a patient to the Department of Neurosurgery