Trained recurrent neural networks develop phase-locked limit cycles in a working memory task.
Neural oscillations are ubiquitously observed in many brain areas. One proposed functional role of these oscillations is that they serve as an internal clock, or 'frame of reference'. Information can be encoded by the timing of neural activity relative to the phase of such oscillations. In...
| 主要な著者: | , , |
|---|---|
| フォーマット: | 論文 |
| 言語: | English |
| 出版事項: |
Public Library of Science (PLoS)
2024-02-01
|
| シリーズ: | PLoS Computational Biology |
| オンライン･アクセス: | https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1011852&type=printable |
| _version_ | 1827013085094739968 |
|---|---|
| author | Matthijs Pals Matthijs Pals Jakob H Macke Omri Barak |
| author_facet | Matthijs Pals Matthijs Pals Jakob H Macke Omri Barak |
| author_sort | Matthijs Pals |
| collection | DOAJ |
| description | Neural oscillations are ubiquitously observed in many brain areas. One proposed functional role of these oscillations is that they serve as an internal clock, or 'frame of reference'. Information can be encoded by the timing of neural activity relative to the phase of such oscillations. In line with this hypothesis, there have been multiple empirical observations of such phase codes in the brain. Here we ask: What kind of neural dynamics support phase coding of information with neural oscillations? We tackled this question by analyzing recurrent neural networks (RNNs) that were trained on a working memory task. The networks were given access to an external reference oscillation and tasked to produce an oscillation, such that the phase difference between the reference and output oscillation maintains the identity of transient stimuli. We found that networks converged to stable oscillatory dynamics. Reverse engineering these networks revealed that each phase-coded memory corresponds to a separate limit cycle attractor. We characterized how the stability of the attractor dynamics depends on both reference oscillation amplitude and frequency, properties that can be experimentally observed. To understand the connectivity structures that underlie these dynamics, we showed that trained networks can be described as two phase-coupled oscillators. Using this insight, we condensed our trained networks to a reduced model consisting of two functional modules: One that generates an oscillation and one that implements a coupling function between the internal oscillation and external reference. In summary, by reverse engineering the dynamics and connectivity of trained RNNs, we propose a mechanism by which neural networks can harness reference oscillations for working memory. Specifically, we propose that a phase-coding network generates autonomous oscillations which it couples to an external reference oscillation in a multi-stable fashion. |
| first_indexed | 2024-03-07T22:00:25Z |
| format | Article |
| id | doaj.art-c7ca0f7c32aa41d8bb5f6dadf141010b |
| institution | Directory Open Access Journal |
| issn | 1553-734X 1553-7358 |
| language | English |
| last_indexed | 2025-02-18T13:48:18Z |
| publishDate | 2024-02-01 |
| publisher | Public Library of Science (PLoS) |
| record_format | Article |
| series | PLoS Computational Biology |
| spelling | doaj.art-c7ca0f7c32aa41d8bb5f6dadf141010b2024-10-30T05:31:07ZengPublic Library of Science (PLoS)PLoS Computational Biology1553-734X1553-73582024-02-01202e101185210.1371/journal.pcbi.1011852Trained recurrent neural networks develop phase-locked limit cycles in a working memory task.Matthijs PalsMatthijs PalsJakob H MackeOmri BarakNeural oscillations are ubiquitously observed in many brain areas. One proposed functional role of these oscillations is that they serve as an internal clock, or 'frame of reference'. Information can be encoded by the timing of neural activity relative to the phase of such oscillations. In line with this hypothesis, there have been multiple empirical observations of such phase codes in the brain. Here we ask: What kind of neural dynamics support phase coding of information with neural oscillations? We tackled this question by analyzing recurrent neural networks (RNNs) that were trained on a working memory task. The networks were given access to an external reference oscillation and tasked to produce an oscillation, such that the phase difference between the reference and output oscillation maintains the identity of transient stimuli. We found that networks converged to stable oscillatory dynamics. Reverse engineering these networks revealed that each phase-coded memory corresponds to a separate limit cycle attractor. We characterized how the stability of the attractor dynamics depends on both reference oscillation amplitude and frequency, properties that can be experimentally observed. To understand the connectivity structures that underlie these dynamics, we showed that trained networks can be described as two phase-coupled oscillators. Using this insight, we condensed our trained networks to a reduced model consisting of two functional modules: One that generates an oscillation and one that implements a coupling function between the internal oscillation and external reference. In summary, by reverse engineering the dynamics and connectivity of trained RNNs, we propose a mechanism by which neural networks can harness reference oscillations for working memory. Specifically, we propose that a phase-coding network generates autonomous oscillations which it couples to an external reference oscillation in a multi-stable fashion.https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1011852&type=printable |
| spellingShingle | Matthijs Pals Matthijs Pals Jakob H Macke Omri Barak Trained recurrent neural networks develop phase-locked limit cycles in a working memory task. PLoS Computational Biology |
| title | Trained recurrent neural networks develop phase-locked limit cycles in a working memory task. |
| title_full | Trained recurrent neural networks develop phase-locked limit cycles in a working memory task. |
| title_fullStr | Trained recurrent neural networks develop phase-locked limit cycles in a working memory task. |
| title_full_unstemmed | Trained recurrent neural networks develop phase-locked limit cycles in a working memory task. |
| title_short | Trained recurrent neural networks develop phase-locked limit cycles in a working memory task. |
| title_sort | trained recurrent neural networks develop phase locked limit cycles in a working memory task |
| url | https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1011852&type=printable |
| work_keys_str_mv | AT matthijspals trainedrecurrentneuralnetworksdevelopphaselockedlimitcyclesinaworkingmemorytask AT matthijspals trainedrecurrentneuralnetworksdevelopphaselockedlimitcyclesinaworkingmemorytask AT jakobhmacke trainedrecurrentneuralnetworksdevelopphaselockedlimitcyclesinaworkingmemorytask AT omribarak trainedrecurrentneuralnetworksdevelopphaselockedlimitcyclesinaworkingmemorytask |