When a discussion drags on, many people come away feeling drained, especially after listening for a long time. This feeling is echoed in a study published on June 11, 2025 in the journal Natural Human Behavior : researchers from the University of Osaka and NICT / CiNet show how the brain builds a conversation second by second, then minute by minute.
At the heart of their results, carried in particular by Masahiro Yamashita, Shinji Nishimoto and Rieko Kubo, a strong idea appears: speaking and listening share the same circuits at very small time scales, but separate when the dialogue becomes a story. Listening then requires more memory, attention and social cognition. In other words, our brain works more to follow others than to express ourselves.
How the brain constructs a conversation over time
The researchers invited eight right-handed native Japanese speakers to converse freely through headphones while undergoing an fMRI. Each person spoke on approximately twenty-seven topics for a total of 3 hours. The team then divided these exchanges into windows of 1, 2, 4, 8, 16 and 32 seconds to see how the brain encodes meaning at each duration.
Using a GPT language model trained to follow instructions, used as a context measurement tool, the team linked each time window to brain activity. At the shortest durations, speaking and listening mainly activate shared networks. When we approach 32 seconds, word production remains local, while comprehension mobilizes additional regions to integrate the thread of the story.
Why listening costs more for the brain than speaking
When a story unfolds, comprehension requires a large network including the angular gyrus, the posterior cingulate cortex and the medial prefrontal cortex. The study published in Natural Human Behavior describes enhanced involvement of the default mode network and regions associated with theory of mind, which help connect words to our memories and represent each other’s thoughts.
According to Psychology Today, Areas like the anterior insula, the amygdala or the temporoparietal junction become more active when the conversation is emotionally charged. The brain of the participant listening must follow the content, the tone, the implications. This combination of memory, sustained attention, and empathy explains why a long meeting can leave you feeling drained even if you haven’t spoken much.
These little “hum hum” that hold the conversation together
The work shows that even the smallest signals of agreement structure the exchange. Marks like “ok,” “uh-huh,” and “you know,” create stable patterns in the brains of both speaker and listener. It’s our “okay”, “um hum” or “you see” that indicate presence and maintain the rhythm, especially when the story stretches.
Conversation is never a simple exchange of information. It depends on this progressive construction of meaning at several times, observed here in eight people, but crucial for understanding our daily dialogues.