Music shapes listeners’ imagination

Previous research has also found that what people imagine while listening to music often forms elaborate imagined stories. These share greater similarity among listeners with a shared cultural background.

Thoughts and themes in the imagined stories are shaped by the music. For example, heroic-sounding music induces empowering themes into imagined content.

Occurrences of new events in these imagined stories also tend to be similar between listeners, and are related to the pattern of musical tension and relief.

So there is strong scientific support for the idea that music can indeed affect what is imagined. But can it specifically induce imagined social interactions?

Our latest study is the first to explicitly investigate this question.

Does music make imagination more social?

We asked 600 participants to perform the imaginary journey task, either in silence or while listening to Italian, Spanish or Swedish folk music. To understand the potential effect of vocals and the meaning of lyrics on imagined content, the music was presented with or without lyrics to the participants, half of whom were native speakers and the other half non-speakers of the respective languages.

We then used tools from natural language processing – a set of computational methods for analysing language – to find underlying topics across participants’ reports of their imagined journeys.

One topic stood out: social interaction. Not only was it the predominant topic in participants’ reports of what they imagined, but it was also much stronger while listening to music compared to silence.

This suggests music can indeed affect social thought. The effect was stable regardless of whether listeners’ understood the lyrics or whether there even were lyrics in the first place.

But we can go one step further.

We used a generative AI system which produces images from text prompts (Stable Diffusion) to visualise participants’ descriptions of their imagined journeys.

By combining the natural language processing model with the image generator, we could visualise what the language processing model had learned to be a “stereotypical” representation of content imagined during silence and music listening.

The results of the computational model were further supported with manual annotations that showed three times more social interactions in journeys imagined during music listening compared to silence.

A shared imagination of music

Finally, we showed the images created from the descriptions to another group of people.

These people were able to pick out which images showed content imagined during music listening, and which showed content imagined while in silence – but they were only able to do it when listening to the same music that inspired the image.

This shows there is a shared understanding, or “theory of mind” of what another person might imagine while listening to a piece of music.

Taken together, our results suggest music can indeed be good company.