Date of Award

Winter 12-15-2021

Author's School

McKelvey School of Engineering

Author's Department

Biomedical Engineering

Degree Name

Doctor of Philosophy (PhD)

Degree Type



In many organisms, the sense of smell, driven by the olfactory system, serves as the primary sensory modality that guides a plethora of behaviors such as foraging for food, finding mates, and evading predators. Using an array of biological sensors, the olfactory system converts volatile chemical inputs from an organism’s environment into well-patterned neural responses that inform downstream motor neurons to drive appropriate behaviors (e.g., moving towards food or away from danger). For many external cues, the elicited neural responses are often determined by the genetic makeup of the organism, which assigns an innate preference, or valence, for these different stimuli. However, our environment is constantly in flux, and the same stimulus can be encountered in a variety of different contexts, such as following other cues or under different ambient conditions (e.g., humidity). This can modify the neural activation pattern ascribed to the stimulus and potentially alter the corresponding behavioral output. The objective of this dissertation is to understand how neural responses in the early olfactory system of locusts (Schisctocerca americana) are spatiotemporally structured to robustly represent innate valence in different scenarios to drive appropriate behaviors and how they can be altered through learning. To achieve this goal, we used a large panel of chemically diverse odorants and characterized the neural responses they elicited in the antennal lobe (at the level of ensembles of principal or projection neurons) as well as the innate appetitive behavioral response they produced. We found that neural responses generated both during (ON response) and after (OFF response) termination of the odorant contained information regarding its identity and could be used to predict the innate behavioral outcomes. Notably, predictions made using the ON and the OFF responses differed in the sets of neurons they used to generate the predictions, indicating that neural-behavioral transformations could be achieved in multiple ways. Furthermore, both these ON and OFF neural response classifiers outperformed attempts to predict behavior using chemical features of the stimuli (detected by NMR or IR spectra), indicating that the antennal lobe was transforming and encoding olfactory inputs to map them onto the innate valence associated with the sensory cue.

We found that the organization of odor-evoked neural responses that readily map onto innate preferences may also constrain learned odor-reward associations. While odorants with an innate positive behavioral preference alone could support learning odor-reward associations, the conditioned responses were not odor-specific but appeared to generalize to other odorants that evoked similar neural responses. The timing of the behavioral responses could be varied by delivering rewards during epochs when the odorant would generate either the ON or the OFF neural responses. Overall, we found that the organization of ON and OFF neural responses in the antennal lobe clustered into manifolds or subspaces that could be explained using innate behavioral preferences and suitability for reinforcement learning.

To understand the robustness of these results, we developed novel minimally invasive experimental methods to record locust neural responses while they actively sampled their surroundings. We found neural responses in this more naturalistic scenario to maintain their manifold organization, and classical conditioning enhanced the separation between neural responses evoked by innately appetitive and non-appetitive odorants. Our results also indicate that neural and behavioral responses in freely moving locusts were consistent with those observed earlier in highly compromised preparations. Finally, we exploited our newly-developed recording techniques to engineer an insect-based chemical sensor that could be used for a real-world application.


English (en)


Baranidharan Raman

Committee Members

Dennis Barbour

Included in

Neurosciences Commons