Optimization of Discrimination and Generalization Learning in Olfactory Scent Detection
Abstract
Optimization of Discrimination and Generalization Learning in Olfactory Scent Detection Abstract: This proposal addresses a need within the Army research officeÕs Biological and Life Sciences, ÒEukaryotic Genetics, Genomics, Epigenetics and Molecular Biology thrustÓ in the stated research area of Òbiological autonomous systems for sensing and detectingÓ. The proposed studies seek to expand the goals of the existing ARO project. The proposed studies will compare the effectiveness of the lab s current approach of scent detection learning with 2 alternative forms of categorization learning and determine which of the three approaches prepare rodents to become more successful biological detectors of explosive materials. The goal of the work is to systematically identify which components of select training procedures result in more efficient and reliable patterns of detection of olfactory explosive stimuli. This effort has important implications to the Army, based upon their repeated expressed interest in low-cost, high throughput methods for producing biological detectors to assist in explosive detection missions. The behavioral component of this work has potential benefits to the Army in areas beyond explosive detection, possibly including cadaver and biological specimen detection. The technical component involves the construction of odor presentation hardware devices with enhanced capabilities and automation software that will improve the teamÕs ability to perform these and other studies for the Army in future work. The combined proposing team of the University of Virginia and Barron Associates, a small business focused on technology development and research, has a ten-year partnership developing custom hardware, software, and improved methodologies for efficient automated training of small animals. This prior work has resulted in the acceleration of behavioral training protocols for na•ve rats, through computer-based quantification of olfactory discrimination learning and predictive modeling of animal performance from early stages of learning. The proposed effort consists of two concurrently submitted proposals, one by each organization, and has two primary components: 1) advanced behavioral training programs to improve the efficiency of rodents to detect a range of explosive odorants under control and novel conditions, with updated protocols for data quantification and 2) the development of customized hardware and control software for automated, wireless odor presentation devices that are equipped with reward delivery components to achieve the basic science objectives. In all, this work will advance fundamental science in advanced autonomous biological detection by creating more sophisticated learning programs that will expand the boundaries of animal scent detection performance not available with existing technology. To support the primary behavioral science, an essential prerequisite goal of the proposed effort is to expand the capabilities of the novel computer-controlled wireless odor presentation devices developed by the team in prior work. Based on cumulative experience over the past decade, the new proposed hardware will have the capacity to quantitatively mix and dilute the emission of up to four explosive odorants by a factor of 100:1 under computer control during training sessions. This modification will permit the team to address the presently unanswered question of odor detection threshold in rodents, the influence of distractor odors in teaching discrimination and generalization learning, and to provide an accurate analysis of the lowest concentrations of common explosives (e.g. TNT, RDX, PETN, Semtex, Ammonium Nitrate or TATP), that can be reliably detected when presented individually, or when they are given as mixed compound stimuli during active searches. Page C-
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 19, 2022
- Source ID
- W911NF2210081
Entities
People
- Cedric Williams
Organizations
- Army Contracting Command
- United States Army
- University of Virginia