Our lab deciphers how animals learn, memorize, and make decisions using state-of-the-art optical imaging techniques and a genetically tractable animal model—Drosophila melanogaster.
Animal brains exploit dynamic neural processes to encode, store and retrieve associative information for guiding proper behaviors. During these processes, vast biological changes occur in different regions throughout the brain and often act on various time scales. Although this distributed and dynamic nature poses extraordinary challenges, elucidating the mechanisms of memory processing would profoundly advance our understanding of the brain and eventually benefit the interventions/treatments for memory disorders.
Our laboratory uses fruit flies, Drosophila, as a model system to study how animals acquire and store information from past experiences and then shape their future decisions. Utilizing novel chronic voltage-imaging techniques, we can track neuronal changes underlying memory with unprecedented millisecond-scale temporal precision and over long-term durations of many weeks in behaving flies. The unique combination of quantitative behavior measurements, large-scale optical imaging, synaptic-level connectome analyses and computational approaches empower us to mechanistically address the following questions regarding memory processing in an integrative manner:
What is the function of heterogeneous dopamine teaching signals in memory encoding?
Dopamine signaling participates in various learning and memory processes in the Drosophila brain. Emerging evidence reveals that the neural activities of fly dopamine neurons (DANs) represent a wide array of learning-related variables, which is analogous to the heterogeneity of the dopamine system in mammals. Using in vivo chronic voltage imaging with learning tasks in virtual reality, we aim to delineate the DANs’ representations of sensory, motor, context, and internal-state variables and to study the causal role of the heterogeneous dopamine teaching signals in mediating memory formation.
What are the molecular and cellular bases of distributed memory storage?
A simple associative learning event can induce neural changes across multiple neural regions. Such distributed nature ensures efficient information encoding and robust memory storage. Drosophila mushroom bodies (MBs), a central structure for learning and memory, contain ~15 neuropil compartments with ~21 downstream MB output neuron types (MBONs) that serve as parallel memory units to guide experience-dependent behavior. Our research program is set to systematically identify where memories are stored in the MB circuit and what molecular/cellular mechanisms sustain memory traces that last for various durations.
How do the downstream regions read out and evaluate the distributed memory traces?
All memory systems face a similar challenge — how the downstream neural regions could reliably translate the distributed information into proper behaviors when necessary. Fly connectome provides clues about the potential downstream circuits that serve as integration sites for action selection. We use combinatory imaging and optogenetics techniques to dissect how distributed memory traces are processed and translated into behavioral decisions.
What are the fundamental neural architectures and computational principles that support memory processing?
With the connectome and large-scale physiology/behavior dataset, we aim to build biologically realistic models of the fly memory system and assess the functional roles of different neural coding properties, structure motifs, and computational mechanisms in supporting efficient and flexible memory processing.