Loading...

Modeling Brain s Surprise Response in Learning of Patterned Sequences and Its Application to Modeling of Neurodegenerative Diseases

Kiani, Mohammad Mahdi | 2024

0 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 58452 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Karbalaei Aghajan, Hamid
  7. Abstract:
  8. Through evolution, the brain has developed to constantly predict incoming inputs, enabling timely reactions to unexpected events. It forms and updates internal models based on new information. A key question is how the brain perceives and responds to novelty. The oddball paradigm is a common test to explore this predictive behavior, often modeling the brain as a near-ideal observer. In this thesis, we analyze brain responses to binary and ternary oddball stimuli using a known Markov-based model. Several mathematical definitions of surprise are tested against EEG data using a linear decoder. Among them, Shannon surprise consistently yields the best predictive performance. We then investigate cognitive brain activity during the oddball task using phase-amplitude coupling (PAC), measured in short time windows. This approach allows us to analyze cogni-tive responses beyond basic sensory perception. Using EEG data from healthy controls and Parkinson’s patients (with and without levodopa), we extract PAC dynamics and identify distinct temporal patterns of surprise-related processing. Although levodopa effectively alleviates motor symptoms in Parkinson’s disease (PD), its im-pact on cognitive impairment is unclear. Our analysis reveals that both cognitive impairments and levodopa side effects may be linked to altered PAC dynamics, suggesting disruptions in dopamine-modulated neural pathways connecting posterior cortex, hippocampus, and pre-frontal areas via the striatum.
  9. Keywords:
  10. Oddball Paradaigm ; Parkinson Disease ; Markov Chain ; Phase-Amplitude Coupling (PAC) ; Bayesian Brain

 Digital Object List

 Bookmark

...see more