Systematic errors in connectivity inferred from activity in strongly recurrent networks
© 2020, The Author(s), under exclusive licence to Springer Nature America, Inc. Understanding the mechanisms of neural computation and learning will require knowledge of the underlying circuitry. Because it is difficult to directly measure the wiring diagrams of neural circuits, there has long been...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
Springer Science and Business Media LLC
2021
|
| Online Access: | https://hdl.handle.net/1721.1/135458 |
| _version_ | 1826209170570870784 |
|---|---|
| author | Das, Abhranil Fiete, Ila R |
| author2 | Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences |
| author_facet | Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences Das, Abhranil Fiete, Ila R |
| author_sort | Das, Abhranil |
| collection | MIT |
| description | © 2020, The Author(s), under exclusive licence to Springer Nature America, Inc. Understanding the mechanisms of neural computation and learning will require knowledge of the underlying circuitry. Because it is difficult to directly measure the wiring diagrams of neural circuits, there has long been an interest in estimating them algorithmically from multicell activity recordings. We show that even sophisticated methods, applied to unlimited data from every cell in the circuit, are biased toward inferring connections between unconnected but highly correlated neurons. This failure to ‘explain away’ connections occurs when there is a mismatch between the true network dynamics and the model used for inference, which is inevitable when modeling the real world. Thus, causal inference suffers when variables are highly correlated, and activity-based estimates of connectivity should be treated with special caution in strongly connected networks. Finally, performing inference on the activity of circuits pushed far out of equilibrium by a simple low-dimensional suppressive drive might ameliorate inference bias. |
| first_indexed | 2024-09-23T14:18:19Z |
| format | Article |
| id | mit-1721.1/135458 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T14:18:19Z |
| publishDate | 2021 |
| publisher | Springer Science and Business Media LLC |
| record_format | dspace |
| spelling | mit-1721.1/1354582023-10-06T20:49:20Z Systematic errors in connectivity inferred from activity in strongly recurrent networks Das, Abhranil Fiete, Ila R Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences © 2020, The Author(s), under exclusive licence to Springer Nature America, Inc. Understanding the mechanisms of neural computation and learning will require knowledge of the underlying circuitry. Because it is difficult to directly measure the wiring diagrams of neural circuits, there has long been an interest in estimating them algorithmically from multicell activity recordings. We show that even sophisticated methods, applied to unlimited data from every cell in the circuit, are biased toward inferring connections between unconnected but highly correlated neurons. This failure to ‘explain away’ connections occurs when there is a mismatch between the true network dynamics and the model used for inference, which is inevitable when modeling the real world. Thus, causal inference suffers when variables are highly correlated, and activity-based estimates of connectivity should be treated with special caution in strongly connected networks. Finally, performing inference on the activity of circuits pushed far out of equilibrium by a simple low-dimensional suppressive drive might ameliorate inference bias. 2021-10-27T20:23:32Z 2021-10-27T20:23:32Z 2020 2021-04-02T18:27:55Z Article http://purl.org/eprint/type/JournalArticle https://hdl.handle.net/1721.1/135458 en 10.1038/S41593-020-0699-2 Nature Neuroscience Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. application/pdf Springer Science and Business Media LLC bioRxiv |
| spellingShingle | Das, Abhranil Fiete, Ila R Systematic errors in connectivity inferred from activity in strongly recurrent networks |
| title | Systematic errors in connectivity inferred from activity in strongly recurrent networks |
| title_full | Systematic errors in connectivity inferred from activity in strongly recurrent networks |
| title_fullStr | Systematic errors in connectivity inferred from activity in strongly recurrent networks |
| title_full_unstemmed | Systematic errors in connectivity inferred from activity in strongly recurrent networks |
| title_short | Systematic errors in connectivity inferred from activity in strongly recurrent networks |
| title_sort | systematic errors in connectivity inferred from activity in strongly recurrent networks |
| url | https://hdl.handle.net/1721.1/135458 |
| work_keys_str_mv | AT dasabhranil systematicerrorsinconnectivityinferredfromactivityinstronglyrecurrentnetworks AT fieteilar systematicerrorsinconnectivityinferredfromactivityinstronglyrecurrentnetworks |