Adaptive Game Music: Hearing the Scene and Naming Its Cause
Adaptive game music can follow changing scenes. A small two-game study separates feeling a soundtrack fits from identifying what its music represents.

Start with the connection you want to hear
Adaptive game music responds to changes during play. Imagine a fictional adventure in which a quiet melody becomes restless when creatures enter a clearing. A listener might say the music suits the danger while being unable to name which creature prompted it. Those answers concern different kinds of understanding.
This invented scene gives a useful question for reading a soundtrack study. Does the evaluation ask whether music belongs with the encounter, whether players recognize a particular musical association, or whether they enjoy listening? A single favorable answer cannot settle all three.
Look at the two-game comparison
Released as a 2019 preprint, Hutchings and McCormack’s study recruited 34 people. Each played StarCraft II and Zelda Mystery of Solarus for 20 minutes apiece, with one original soundtrack and one adaptive score. Two groups received opposite pairings, with game order alternated.
For a hypothetical comparison of two soundtracks, consider what stays constant when each person answers. Keeping the scene identical makes it easier to attribute a difference to the music, although replay introduces familiarity. Changing the game avoids that exact repetition while adding another possible explanation.
Read the statistical comparisons separately
All four reported game-specific tests of immersion and music–event correspondence were non-significant at .05. The authors also reported a significant within-group comparison favoring adaptive music, but that comparison changed game as well.
These tests address different comparisons. A result spanning different games cannot independently isolate the soundtrack. Likewise, an undetected difference in a small comparison leaves uncertainty about its size. It does not establish identical experiences. Preserve the comparison behind each number before deciding what it supports.
Ask what a short clip identifies
Reported clip-association rates favored original scores. However, the Zelda clip paragraph reverses earlier group labels, and the two soundtrack conditions used clips representing different concepts. That limits interpreting the task.
Return to our fictional clearing. One recording might accompany the entire forest, while another accompanies a particular creature. Accepting “forest” for the first and requiring a creature’s name for the second creates different answer targets. The score then depends on the target and accepted words alongside anything listeners understood.
Give a proposed playtest separate questions
For an original prototype exercise, define one event and write down the musical change intended to accompany it. Playtesters could first describe the encounter freely, then answer whether the music fitted it. A separate clip task could ask which event the passage suggests. These are proposed checks, not tests performed here.
Keep the answer rules visible. Decide in advance whether “danger,” “battle” and the creature’s name count as the same association. Record unmatched answers too. A broad mood description may be valuable feedback even when it misses a narrow identification target. Report each outcome under its own question.
Carry the limits into the next question
Recruitment used Melbourne social-media pages. These two brief game sessions leave effects over longer play unresolved.
An evening-long adventure raises another proposed question about repetition. Does a musical association remain understandable after several encounters, or does the player start attaching it to something else? That would require a longer comparison. For now, evaluating a soundtrack means naming the connection being assessed and choosing an observation that can actually address it.
Keep exploring
Sources & further reading
Have a correction? Tell our editorial team.