Building intelligent systems using fruit fly navigational neural networks
Abstract
Understanding how the brain develops specialized neural architecture and organizes information into different cognitive modules can inspire the development of more sophisticated cognitive architectures in machine learning, which can lead to artificial intelligence with enhanced reasoning, problem-solving, and decision-making capabilities. Despite their small brains, insects are experts in performing complex behaviors, such as context-dependent decision-making, homing, and spatial and celestial navigation. Ants in the Sahara Desert, African dung beetles, Monarch butterflies, honey bees, and fruit flies (Drosophila) have attained remarkable navigation features. The neural circuits involved in these complex behaviors reside in a sophisticated brain region, the central complex. The central complex is an evolutionary conserved higher-order brain region conserved across all insect species, and over the millions of years of evolution, this brain region has attained the highest complexity, efficiency, and accuracy. We propose to use the Drosophila central complex and olfactory navigation circuit as a model to understand how developmental programs organize the structure and function of this circuit. Given the availability of a rich repertoire of genetic tools and full brain connectome, the Drosophila brain will be instrumental in identifying the genetic basis of neural circuit structure and function. We will investigate how hormonal signaling in neural stem cells governs specialized components of the olfactory navigation circuit to drive complex sensory-motor integration. We have found that all the essential components of the olfactory navigation circuit are made by specialized neural stem cells called Type II neural stem cells (T2NSCs). Our previous studies have identified over a dozen conserved genes expressed in T2NSCs, which are the prime candidate genes for directing the identity of the central complex neuron types. Interestingly, the ecdysone growth hormone receptor is expressed in the mid-aged T2NSCs and regulates early to late gene transitions. These data raise the exciting possibility that evolutionarily conserved hormone signaling might play an essential role in shaping the olfactory navigation circuits and behavior. The scientific objectives of the proposed work are to 1) Determine the role of ecdysone signaling in the olfactory navigation circuit development, 2) Identify the roles of steroid hormone signaling in the olfactory navigation system function, and implement behavior-tracking videos to train and inform machine learning, 3) Investigate how a bio-inspired reservoir computer can simulate some of the functions of the Drosophila central complex. A deeper understanding of the developmental principles regulating brain organization will help us not only understand the fundamental principles of brain development but will also inspire more efficient algorithms in machine learning. Imagine machines that can learn and make decisions with minimal data, just like a clever bug.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 06, 2025
- Source ID
- FA95502410214
Entities
People
- Mubarak Hussain Syed
Organizations
- Air Force Office of Scientific Research
- Office of the Secretary of Defense
- University of New Mexico