Workshop: Building a 1,000-Song Naturalistic EEG Dataset: Practical Lessons from ManyMusic

2026-Sep-23, CuttingGardens 2026 — Frankfurt local program

Abstract

This CuttingGardens-2026-Frankfurt-local workshop shares the experience of building a deep phenotyping dataset of music-evoked emotions, namely “ManyMusic🎶” (https://manymusic.net/). This ongoing project involves 17 EEG sessions and 17 fMRI sessions per participant, measuring neural and behavioural responses during intense affective experiences evoked by complete musical works spanning a wide range of genres, including Western classical, jazz, rock, pop, hip-hop, and more. Across six participants, the project will collect data from more than 1,000 full-length musical pieces.

In this workshop, we will discuss the challenges of building a deep phenotyping dataset such as stimulus selection, participant screening, data quality monitoring, and on-the-fly troubleshooting. Participants will also have the opportunity to examine a subset of the collected EEG data first-hand. We will conclude by discussing the future prospects of developing foundation models from deeply sampled multimodal datasets.

Materials