Choosing a Direction: Neural Models of Decision Making
Abstract
Arguably, the primary goal of Artificial Intelligence (AI) is to develop machines that act in an intelligent manner. A key component to acting intelligently is the ability to take information from multiple sensory inputs, combine them together with a goal, to create and act upon a decision. Researchers have had a great deal of success in developing systems that encode human-like decision making. For example, computers can now defeat even the best human chess players, stock market trading systems manage trillions of dollars with minimal human intervention, and the thermostat in our houses can control the temperature to within a few degrees of a desired setting. However, none of these systems has reached the level of true intelligence because they simply run a program where the decision making is encoded by the developer. Building on our prior work developing neural models for locomotion in the nematode Caenorhabditis elegans, this project investigates the neural circuits that encode action selection and execution. Our effort will focus on the neurons that regulate forward-reverse movement, locomotion speed, and directional control. Using currently published results along with recent advances in optogenetics, we will explore these neurons and their interrelationships to develop a computational model that can be used to drive our existing simulated worm. This model is potentially of great value to guiding the design of future artificially intelligent systems.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 30, 2018
- Source ID
- FA95501810308
Entities
People
- Roger Mailler
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- University of Tulsa