Abstract
Odorant identity is encoded by patterns of activity across olfactory receptor neurons (ORNs) and their corresponding glomeruli in the antennal lobe (AL). An odorant's valence can be modified by experience, allowing animals to associate previously neutral odors with appetitive or aversive outcomes. However, some odors appear resistant to such reassignment, and the factors determining whether an odor can acquire learned valence and whether that association generalizes to chemically related odors remain poorly understood. This thesis uses Drosophila melanogaster to address this question using a fully automated proboscis extension reflex (PER) conditioning assay. Odorants are presented via a computer-controlled olfactometer, reward is delivered through optogenetic activation of Gr5a neurons, and behavior is continuously tracked using a computer vision model (YOLO26s-OBB). Testing seven odorants at three concentrations (three independent cohorts, n = 20 flies per cohort) yielded several key findings.
Naive PER rates varied with both odor identity and concentration, with odor-specific changes across concentrations. Conditioning experiments with hexanol produced robust, odorant-specific learning (69% of trained flies responding versus 7% of unrewarded controls, p = 0.001). In contrast, apple cider vinegar acquired no conditioned response under the same protocol, establishing that trainability is a property of the stimulus rather than a limitation of the assay. The hexanol-trained flies showed a cross-learning generalization toward the two alcohols in the panel (38% versus 7% and 20% respectively; non-significant trends), while other odorants remained unchanged. Receptor-space overlap alone did not account for the pattern. One interpretation these observations motivate is that associative plasticity may engage only a subset of the glomeruli that an odor recruits, rather than the full peripheral representation, in which case generalization would depend on overlap among plasticity-competent channels rather than on peripheral similarity alone.
Committee Chair
Barani Raman
Committee Members
Yehuda Ben-Shahar Cheng Huang
Degree
Master of Engineering (ME)
Author's Department
Biomedical Engineering
Document Type
Thesis
Date of Award
Summer 8-13-2026
Language
English (en)
Author's ORCID
0009-0007-5212-6456
Recommended Citation
Hanan, Cole B., "Deciphering the Rules of Odor-Driven Learning in the Fruit Fly Olfactory System" (2026). McKelvey School of Engineering Graduate Student Theses & Dissertations. 1397.
https://openscholarship.wustl.edu/eng_etds/1397