Interpreting neural decoding models using grouped model reliance.
Machine learning algorithms are becoming increasingly popular for decoding psychological constructs based on neural data. However, as a step towards bridging the gap between theory-driven cognitive neuroscience and data-driven decoding approaches, there is a need for methods that allow to interpret...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Public Library of Science (PLoS)
2020-01-01
|
| Series: | PLoS Computational Biology |
| Online Access: | https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1007148&type=printable |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850241326390444032 |
|---|---|
| author | Simon Valentin Maximilian Harkotte Tzvetan Popov |
| author_facet | Simon Valentin Maximilian Harkotte Tzvetan Popov |
| author_sort | Simon Valentin |
| collection | DOAJ |
| description | Machine learning algorithms are becoming increasingly popular for decoding psychological constructs based on neural data. However, as a step towards bridging the gap between theory-driven cognitive neuroscience and data-driven decoding approaches, there is a need for methods that allow to interpret trained decoding models. The present study demonstrates grouped model reliance as a model-agnostic permutation-based approach to this problem. Grouped model reliance indicates the extent to which a trained model relies on conceptually related groups of variables, such as frequency bands or regions of interest in electroencephalographic (EEG) data. As a case study to demonstrate the method, random forest and support vector machine models were trained on within-participant single-trial EEG data from a Sternberg working memory task. Participants were asked to memorize a sequence of digits (0-9), varying randomly in length between one, four and seven digits, where EEG recordings for working memory load estimation were taken from a 3-second retention interval. The present results confirm previous findings insofar as both random forest and support vector machine models relied on alpha-band activity in most subjects. However, as revealed by further analyses, patterns in frequency and topography varied considerably between individuals, pointing to more pronounced inter-individual differences than previously reported. |
| format | Article |
| id | doaj-art-1a0ca2cbc75e43aea3b42eb537b7e226 |
| institution | OA Journals |
| issn | 1553-734X 1553-7358 |
| language | English |
| publishDate | 2020-01-01 |
| publisher | Public Library of Science (PLoS) |
| record_format | Article |
| series | PLoS Computational Biology |
| spelling | doaj-art-1a0ca2cbc75e43aea3b42eb537b7e2262025-08-20T02:00:38ZengPublic Library of Science (PLoS)PLoS Computational Biology1553-734X1553-73582020-01-01161e100714810.1371/journal.pcbi.1007148Interpreting neural decoding models using grouped model reliance.Simon ValentinMaximilian HarkotteTzvetan PopovMachine learning algorithms are becoming increasingly popular for decoding psychological constructs based on neural data. However, as a step towards bridging the gap between theory-driven cognitive neuroscience and data-driven decoding approaches, there is a need for methods that allow to interpret trained decoding models. The present study demonstrates grouped model reliance as a model-agnostic permutation-based approach to this problem. Grouped model reliance indicates the extent to which a trained model relies on conceptually related groups of variables, such as frequency bands or regions of interest in electroencephalographic (EEG) data. As a case study to demonstrate the method, random forest and support vector machine models were trained on within-participant single-trial EEG data from a Sternberg working memory task. Participants were asked to memorize a sequence of digits (0-9), varying randomly in length between one, four and seven digits, where EEG recordings for working memory load estimation were taken from a 3-second retention interval. The present results confirm previous findings insofar as both random forest and support vector machine models relied on alpha-band activity in most subjects. However, as revealed by further analyses, patterns in frequency and topography varied considerably between individuals, pointing to more pronounced inter-individual differences than previously reported.https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1007148&type=printable |
| spellingShingle | Simon Valentin Maximilian Harkotte Tzvetan Popov Interpreting neural decoding models using grouped model reliance. PLoS Computational Biology |
| title | Interpreting neural decoding models using grouped model reliance. |
| title_full | Interpreting neural decoding models using grouped model reliance. |
| title_fullStr | Interpreting neural decoding models using grouped model reliance. |
| title_full_unstemmed | Interpreting neural decoding models using grouped model reliance. |
| title_short | Interpreting neural decoding models using grouped model reliance. |
| title_sort | interpreting neural decoding models using grouped model reliance |
| url | https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1007148&type=printable |
| work_keys_str_mv | AT simonvalentin interpretingneuraldecodingmodelsusinggroupedmodelreliance AT maximilianharkotte interpretingneuraldecodingmodelsusinggroupedmodelreliance AT tzvetanpopov interpretingneuraldecodingmodelsusinggroupedmodelreliance |