By François Pachet.
Computers in Entertainment, 6(3), article 31, pp. 1–20, 2008. Also published as a chapter in Mario Tokoro (ed.), Open Systems Science: From Understanding Principles to Solving Problems, pp. 133–158, IOS Press, 2010.
Publication versions
2008 journal version (PDF)
Computers in Entertainment, 6(3), article 31. ACM record and DOI.
2010 book chapter version (PDF)
In Mario Tokoro (ed.), Open Systems Science: From Understanding Principles to Solving Problems, IOS Press.
About the paper
This essay argues that digital tools can help people discover and express their own creative ideas. It develops reflexive interaction as a way for a person to explore a model that a computer has learned from their behavior.
Reflexive interaction
A reflexive system continually learns from the user and produces responses that resemble aspects of their activity. The user recognizes something of themselves in those responses, reacts, and supplies new material for the system to learn. The essay explores this cycle through musical performance, personal classification of music, and design using descriptive tags.
For example, the Continuator learns from phrases played by a musician and produces continuations using that material. The musician can recognize their own patterns while hearing new combinations. Their next phrase changes what the system learns, so the interaction develops through both the musician’s choices and the system’s responses.
Related work
The Continuator illustrates the idea in musical dialogue. Description-Based Design of Melodies develops the use of a person’s descriptive words to steer generated variations. See Flow Machines for later research on assisted composition.