Logo Logo
FAQ
Contact
Switch language to German
Population-level neural coding for higher cognition
Population-level neural coding for higher cognition
Higher cognition encompasses advanced mental processes that enable complex thinking, decision-making, problem-solving, and abstract reasoning. These functions involve integrating information from multiple sensory modalities and organizing action plans based on the abstraction of past information. The neural activity underlying these functions is often complex, and the contribution of single neurons in supporting population-level representations of cognitive variables is not yet clear. In this thesis, I investigated the neural mechanisms underlying higher cognition in higher-order brain regions with single-neuron resolution in human and non-human primates performing working memory tasks. I aimed to understand how representations are arranged and how neurons contribute to the population code. In the first manuscript, I investigated the population-level neural coding for the maintenance of numbers in working memory within the parietal association cortex. By analyzing intra-operative intracranial micro-electrode array recording data, I uncovered distinct representations for numbers in both symbolic and nonsymbolic formats. In the second manuscript, I delved deeper into the neuronal organizing principles of population coding to address the ongoing debate surrounding memory maintenance mechanisms. I unveiled sparse structures in the neuronal implementation of representations and identified biologically meaningful components that can be directly communicated to downstream neurons. These components were linked to subpopulations of neurons with distinct physiological properties and temporal dynamics, enabling the active maintenance of working memory while resisting distraction. Lastly, using an artificial neural network model, I demonstrated that the sparse implementation of temporally modulated working memory representations is preferred in recurrently connected neural populations such as the prefrontal cortex. In summary, this thesis provides a comprehensive investigation of higher cognition in higher-order brain regions, focusing on working memory tasks involving numerical stimuli. By examining neural population coding and unveiling sparse structures in the neuronal implementation of representations, our findings contribute to a deeper understanding of the mechanisms underlying working memory and higher cognitive functions.
representational geometry, population coding, higher cognition, working memory, neuronal implementation
Lin, Xiaoxiong
2023
English
Universitätsbibliothek der Ludwig-Maximilians-Universität München
Lin, Xiaoxiong (2023): Population-level neural coding for higher cognition. Dissertation, LMU München: Graduate School of Systemic Neurosciences (GSN)
[thumbnail of Lin_Xiaoxiong.pdf]
Preview
PDF
Lin_Xiaoxiong.pdf

11MB

Abstract

Higher cognition encompasses advanced mental processes that enable complex thinking, decision-making, problem-solving, and abstract reasoning. These functions involve integrating information from multiple sensory modalities and organizing action plans based on the abstraction of past information. The neural activity underlying these functions is often complex, and the contribution of single neurons in supporting population-level representations of cognitive variables is not yet clear. In this thesis, I investigated the neural mechanisms underlying higher cognition in higher-order brain regions with single-neuron resolution in human and non-human primates performing working memory tasks. I aimed to understand how representations are arranged and how neurons contribute to the population code. In the first manuscript, I investigated the population-level neural coding for the maintenance of numbers in working memory within the parietal association cortex. By analyzing intra-operative intracranial micro-electrode array recording data, I uncovered distinct representations for numbers in both symbolic and nonsymbolic formats. In the second manuscript, I delved deeper into the neuronal organizing principles of population coding to address the ongoing debate surrounding memory maintenance mechanisms. I unveiled sparse structures in the neuronal implementation of representations and identified biologically meaningful components that can be directly communicated to downstream neurons. These components were linked to subpopulations of neurons with distinct physiological properties and temporal dynamics, enabling the active maintenance of working memory while resisting distraction. Lastly, using an artificial neural network model, I demonstrated that the sparse implementation of temporally modulated working memory representations is preferred in recurrently connected neural populations such as the prefrontal cortex. In summary, this thesis provides a comprehensive investigation of higher cognition in higher-order brain regions, focusing on working memory tasks involving numerical stimuli. By examining neural population coding and unveiling sparse structures in the neuronal implementation of representations, our findings contribute to a deeper understanding of the mechanisms underlying working memory and higher cognitive functions.