Pattern recognition in locust early olfactory circuits

Abstract

This dissertation investigates the computational principles underlying pattern recognition in early olfactory circuits of the locust (Schistocerca americana). The olfactory system provides a tractable model for studying how the brain encodes, transforms, and decodes sensory information. Using a combination of electrophysiology, computational modeling, and information-theoretic analysis, this work characterizes how odor identity and concentration are multiplexed through population temporal codes in the antennal lobe — the first relay in the olfactory pathway. We examine the role of synchronized oscillations generated by inhibitory local neurons, the read-out of these codes by Kenyon cells in the mushroom body, and the invariance of olfactory representations to odor concentration and volatility. The results establish principles of temporal population coding that are likely to generalize to other sensory systems.

Overview

The locust olfactory system has emerged as one of the most productive models for studying neural coding. The antennal lobe generates oscillatory, synchronized responses to odors, which are then read out by the mushroom body's Kenyon cells via a sparse, combinatorial code. This dissertation provides a comprehensive analysis of this transformation, from stimulus encoding to pattern completion and discrimination.

Key contributions include: (1) a theoretical analysis of which downstream neurons can read population temporal codes; (2) evidence that olfactory representations are invariant to concentration and volatility; (3) a characterization of gain control mechanisms in the antennal lobe; and (4) novel statistical methods for analyzing neural data from oscillatory circuits.

Chapters

1. Olfaction and the Chemical Senses: A Review
2. The olfactory system is invariant to volatility
3. Who reads temporal information in synchronized spike trains?
4. Information content of neurons in population temporal coding
5. The role of synchronization in decoding temporal information
6. Multiplexing odor identity and concentration
7. Gain control via excitation/inhibition balance
8. Odor categorization: Attractors in antennal lobe representations

Cite This Work