Contextual modulation of perceptual decisions and neuronal representations by stimulus and reward history

Khanal, Shreya (2023): Contextual modulation of perceptual decisions and neuronal representations by stimulus and reward history. Dissertation, LMU München: Graduate School of Systemic Neurosciences (GSN)

Abstract

Utilizing regularities in this complex but stable environment is important for survival. Hence, our decisions do not only depend on current sensory input but are also influenced by previous experiences. This thesis investigates such effects on behavior during visual perceptual decision-making. I further implemented a long-term neuronal recording method and provide a proof of concept on how such a method can be utilized to reveal the neuronal encoding of previous experiences. I designed behavioral setups and trained head-fixed mice to perform a lick-left / lick-right orientation discrimination task using a two-alternative forced choice paradigm. Using Logistic Regression, I estimated the contributions of task-relevant (current stimulus) and task-irrelevant (past stimuli, past response, past outcomes) factors on the animals' decisions. To investigate how mice trade-off stimulus-related and contextual information, I introduced blocks of trials with stimulus-specific imbalances of reward. I further designed a neuronal recording setup and implemented a long-term recording method using an immobile silicone probe. Mice learned the task and reached stable performance in 6--8~weeks. For the current stimulus, model weights were negligible for trials with short reaction times, suggesting that on these trials, animals failed to use relevant sensory evidence to guide decisions. Irrespective of reaction time, significant weights were assigned to past stimulus, response, and outcome, indicating that history effects influence decisions across all trials. When contextual information (i.e., reward size) was manipulated, mice showed a consistent bias towards the response side associated with larger rewards and the model weight assigned to the current stimulus decreased. However, given the performance during balanced reward condition, mice performed at an optimal bias level to maximize reward. To understand how contextual signals influence neuronal responses and sensory representations in the primary visual cortex, I performed long-term neuronal recording during learning. Stable recordings were obtained for~2 months and specific neurons were tracked for several days. Preliminary data suggest that there might be variation in responses of neurons in the primary visual cortex based on previous outcomes. Taken together, even for simple visual stimuli well above threshold, perceptual decision-making in mice is influenced not only by the current visual stimulus but also by the history of past trials and the context of reward. In addition, it is important to account for trial-to-trial variability in the speed of the response to understand behavior. Finally, influences of prior outcomes might be reflected in the primary visual cortex.

• Dokumententyp: Dissertationen (Dissertation, LMU München)
• Themengebiete: 500 Naturwissenschaften und Mathematik ; 500 Naturwissenschaften und Mathematik > 570 Biowissenschaften, Biologie
• Fakultäten: Graduate School of Systemic Neurosciences (GSN)
• Sprache der Hochschulschrift: Englisch
• Datum der mündlichen Prüfung: 21. Juni 2023
• 1. Bericht­erstatter:in: Busse, Laura