P160 Preservation of neural dynamics across individuals during cognitive tasks
Ioana Lazar*1, Mostafa Safaie1, Juan Alvaro Gallego1
1Department of Bioengineering, Imperial College London, London, UK
*Email: ioana.lazar20@imperial.ac.uk
Introduction
Different individuals from the same species have brains that have similar organisation but differ in the details of their cellular architecture. Yet, despite these idiosyncrasies, the way in which neurons from the same region co-modulate their activity during a given motor task is remarkably preserved across individuals [1]. Such preserved neural population “latent dynamics” likely arise from the behavioural similarity as well as species-specific constraints on network connectivity. Here we asked whether cognitive tasks that can be solved using different covert strategies could lead to more individual-specific latent dynamics.
Methods
We investigated the preservation of latent dynamics in the prefrontal cortex across macaque monkeys performing an associative memory task in which they had to select the target associated with an initial cue following a “working memory period” in which no information was presented [2]. We computed session-specific latent dynamics using principal component analysis and tested their preservation across individuals using both canonical correlation analysis, which tests for similarity in the geometrical properties of neural population activity, and dynamical systems approaches. We interpreted the differences in the preservation of latent dynamics based on the differences in decoding accuracy of task variables.
Results
Prefrontal cortex latent dynamics were less preserved across individuals than inpreviousstudies of the motor system, especially during the working memory period, in which correlations were lower than during cue presentation and target selection. The level of preservation was strongly associated with how well the upcoming target's identity could be decoded, which varied across animals, hinting at potential different cognitive strategies as the cause for the lower preservation. Finally, monkeys developed idiosyncratic fidgets that reflected their cognitive processes: removing components of the latent dynamics related to movement decreased both within-monkey decoding of task variables and the preservation of latent dynamics across monkeys.
Discussion
This study builds on previous work on the motor system to show that different individuals from the same species also produce preserved latent dynamics when engaged in the same cognitive task. When the decoding analysis suggested that monkeys were employing different cognitive strategies to solve the task—relying more on retrospective or prospective memory—,the preservation of latent dynamics decreased, as it would be expected if the latent dynamics reflected the underlying computations. Neural population latent dynamics can thus capture fundamental differences and similarities in neural computation across individuals during both sensorimotor and cognitive processes.
Acknowledgements
References
1. Safaie, M., Chang, J., Park, J., Miller, L. E., Dudman, J. T., Perich, M. G., & Gallego, J. A. (2023). Preserved neural dynamics across animals performing similar behaviour.Nature, 623, 765–771. https://doi.org/10.1038/s41586-023-06714-0
2. Tremblay, S., Testard, C., DiTullio, R. W., Inchauspé, J., & Petrides, M. (2022). Neural cognitive signals during spontaneous movements in the macaque.Nature Neuroscience, 26, 295–305. https://doi.org/10.1038/s41593-022-01220-4
