A pleasant conversation rarely moves in a straight line: people do not agree constantly and do not jump chaotically from topic to topic. In a new paper published in Science Advances, more successful conversations are associated with an enhanced version of the mathematical pattern "pink noise." In it, interlocutors converge on a topic, briefly diverge by introducing fresh ideas, and come together again.
The authors, led by Katie O'Neil of Dartmouth College, studied the semantic closeness of utterances. Each utterance was converted into a set of numerical coordinates using a computer program, and then they measured how the distance between the meanings of neighboring statements changes. This produced a timeline of convergences and divergences.
The scientists drew on the concept of noise color from physics. White noise means that fast and slow changes are balanced, and each moment is unpredictable. Red noise, or Brownian drift, means the predominance of slow trends. Pink noise occupies the middle: slow shifts weigh more than fast ones, and small cycles are nested within large ones.
The first analysis used an open dataset of 1,656 online video chats in English between strangers. Most conversations exceeded the white-noise mark, and the average score was 0.62 on a scale where 0.5 corresponds to white noise and 1.0 to pure pink. Shuffling the order of utterances weakened the signal, meaning the sequence itself matters, not individual words.
Additional checks showed that the pattern reflects joint coordination rather than one speaker's habit. If only one participant's utterances are analyzed, the signal shifts toward white noise. Among people who took part in several chats, no similarity on this measure was found.
The pink-noise scores were then compared with answers to questions about enjoyment, connection with the partner, engagement and memory. After correction for multiple tests, positive associations were found with enjoyment, the desire to continue the conversation, self-disclosure and perceived disclosure by the partner. Associations with memory, demographics and personality traits were not significant.
In conversations among friends, the association with enjoyment was moderate, while the association with a sense of connection was not significant. In a small sample of phone dates, pairs who agreed to a second date showed stronger pink noise, but the authors call this preliminary evidence. In podcasts, the measure strengthened as the hosts worked together.
The limitations are important: all conversations were observed rather than experimentally controlled, so it cannot be said that pink noise causes enjoyment. It is possible that people who already feel good together naturally fall into this rhythm. The associations are small, and some samples are small.
The authors' practical advice is simple: do not rush in a conversation, ask clarifying questions and explore a topic more deeply before moving to the next one. Future experiments should actively guide the pace of conversation to test causality.