Testing Serial Reccurent Neural Networks and Instance Theory models of learning using skilled musicians and non-musicians

Major

Psychology

Anticipated Graduation Year

2022

Access Type

Open Access

Abstract

Two computational models (Instance theory and Serial recurrent neural networks) make different predictions about whether lower-order knowledge representations are spared or overwritten as people acquire higher-order representations as a function of learning. We tested these predictions by using a k-means clustering algorithm to quantify the number of unique improvisational jazz performances produced by expert and novice pianists. While expert jazz piano players produced the most unique performances, expert classical piano players produced more unique performances than the novices. These findings suggest that lower-order representations are not overwritten and can be accessed when task demands are high.

Streaming Media

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License.

Share

COinS
 

Testing Serial Reccurent Neural Networks and Instance Theory models of learning using skilled musicians and non-musicians

Two computational models (Instance theory and Serial recurrent neural networks) make different predictions about whether lower-order knowledge representations are spared or overwritten as people acquire higher-order representations as a function of learning. We tested these predictions by using a k-means clustering algorithm to quantify the number of unique improvisational jazz performances produced by expert and novice pianists. While expert jazz piano players produced the most unique performances, expert classical piano players produced more unique performances than the novices. These findings suggest that lower-order representations are not overwritten and can be accessed when task demands are high.