Vectorized instructive signals in cortical dendrites during a brain-computer interface task
Gardado en:
| Publicado en: | bioRxiv (Jan 13, 2025) |
|---|---|
| Autor Principal: | |
| Outros autores: | , , , |
| Publicado: |
Cold Spring Harbor Laboratory Press
|
| Materias: | |
| Acceso en liña: | Citation/Abstract Full Text - PDF Full text outside of ProQuest |
| Etiquetas: |
Sen Etiquetas, Sexa o primeiro en etiquetar este rexistro!
|
MARC
| LEADER | 00000nab a2200000uu 4500 | ||
|---|---|---|---|
| 001 | 2886154931 | ||
| 003 | UK-CbPIL | ||
| 022 | |a 2692-8205 | ||
| 024 | 7 | |a 10.1101/2023.11.03.565534 |2 doi | |
| 035 | |a 2886154931 | ||
| 045 | 0 | |b d20250113 | |
| 100 | 1 | |a Francioni, Valerio | |
| 245 | 1 | |a Vectorized instructive signals in cortical dendrites during a brain-computer interface task | |
| 260 | |b Cold Spring Harbor Laboratory Press |c Jan 13, 2025 | ||
| 513 | |a Working Paper | ||
| 520 | 3 | |a Vectorization of teaching signals is a key element of virtually all modern machine learning algorithms, including backpropagation, target propagation and reinforcement learning. Vectorization allows a scalable and computationally efficient solution to the credit assignment problem by tailoring instructive signals to individual neurons. Recent theoretical models have suggested that neural circuits could implement single phase vectorized learning at the cellular level by processing feedforward and feedback information streams in separate dendritic compartments. This presents a compelling, but untested, hypothesis for how cortical circuits could solve credit assignment in the brain. We leveraged a neurofeedback brain computer interface (BCI) task with an experimenter defined reward function to test for vectorized instructive signals in dendrites. We trained mice to modulate the activity of two spatially intermingled populations (4 or 5 neurons each) of layer 5 pyramidal neurons in the retrosplenial cortex to rotate a visual grating towards a target orientation while we recorded GCaMP activity from somas and corresponding distal apical dendrites. We observed that the relative magnitudes of somatic versus dendritic signals could be predicted using the activity of the surrounding network and contained information about task-related variables that could serve as instructive signals, including reward and error. The signs of these putative teaching signals both depended on the causal role of individual neurons in the task and predicted changes in overall activity over the course of learning. Furthermore, targeted optogenetic perturbation of these signals disrupted learning. These results provide the first biological evidence of a vectorized instructive signal in the brain, implemented via semi independent computation in cortical dendrites, unveiling a potential mechanism for solving credit assignment in the brain.Competing Interest StatementThe authors have declared no competing interest.Footnotes* 1. Causal perturbation of dendritic activity disrupts learning. We found that the disruption of independent dendritic signals via optogenetic activation of layer 1 NDNF+ interneurons impaired learning, demonstrating the instructive role of the dendritic signals we observed for behavior (Figure 5h) 2. Optogenetic activation of NDNF+ interneurons disrupted reward-related information in apical dendrites of layer 5 neurons (Figure 4h-l). 3. L1 NDNF+ interneuron activation also abolished error-related information and signal vectorization in apical dendrites (Figure 5f-g). 4. Simultaneous multi-plane imaging to verify the specificity of NDNF+ IN activation on apical dendritic activity (Figure 3f-l). 5. We directly demonstrated that acute anesthesia strongly decreases the SD residual in a new cohort of mice. These experiments are a second, independent confirmation that the SD residual is a robust metric for somato-dendritic coupling and top-down signaling (Figure 3a-e). 6. Refocused the manuscript on vectorization. We believe that the shift from backpropagation broadens the relevance and appeal of our work, increasing its influence in the field. 7. Further contextualized our findings within the existing BCI literature. In the introduction, results, and discussion we added several passages which more specifically frame our findings within published BCI work. 8. We added a new Methods section to our manuscript to clarify the selection procedure for the neurons directly controlling the BCI. We also performed a new analysis to show how these neurons compare to the rest of the network on day 1 (Supplementary Figure 6b). | |
| 653 | |a Brain | ||
| 653 | |a Implants | ||
| 653 | |a Feedback | ||
| 653 | |a Visual pathways | ||
| 653 | |a Orientation behavior | ||
| 653 | |a Computer applications | ||
| 653 | |a Pyramidal cells | ||
| 653 | |a Back propagation | ||
| 653 | |a Assignment problem | ||
| 653 | |a Dendrites | ||
| 653 | |a Neural networks | ||
| 700 | 1 | |a Tang, Vincent D | |
| 700 | 1 | |a Toloza, Enrique Hs | |
| 700 | 1 | |a Brown, Norma J | |
| 700 | 1 | |a Harnett, Mark | |
| 773 | 0 | |t bioRxiv |g (Jan 13, 2025) | |
| 786 | 0 | |d ProQuest |t Biological Science Database | |
| 856 | 4 | 1 | |3 Citation/Abstract |u https://www.proquest.com/docview/2886154931/abstract/embedded/L8HZQI7Z43R0LA5T?source=fedsrch |
| 856 | 4 | 0 | |3 Full Text - PDF |u https://www.proquest.com/docview/2886154931/fulltextPDF/embedded/L8HZQI7Z43R0LA5T?source=fedsrch |
| 856 | 4 | 0 | |3 Full text outside of ProQuest |u https://www.biorxiv.org/content/10.1101/2023.11.03.565534v2 |