Event Title

Dynamical Systems Models of Odor-Tracking Behavior

Presenter Information

Alexander Riordan, Oberlin College

Location

Science Center, Bent Corridor

Start Date

9-26-2014 12:00 PM

End Date

9-26-2014 1:20 PM

Poster Number

11

Abstract

The ability to sense and track noisy environmental chemical cues is essential for survival in a multitude of diverse species, from bacteria to mammals. Despite the biological importance of this capability, the mechanisms used by most organisms, and especially rodents, to follow chemical trails remain poorly understood. Hence, we developed precise, testable computational models of odor-tracking behavior to investigate the mechanisms that underlie this widespread behavior. We created models simulating tracking mechanisms based on spatial (internostril) and temporal (sniff-to-sniff) gradients of odor concentration. Models were simulated via numerical integration using XPP/XPPAUT, and their parameters were tuned to represent mice given an odor trail tracking task. When odor trails were represented in an idealized, noiseless form, simple spatial comparison models accurately tracked the trails, yet temporal models were only successful with additional, more sophisticated features. Models incorporating both spatial and temporal comparisons also accurately tracked the odor trail. To test how well each mechanism performed in more realistic, noisy odor-signal conditions, we quantified the behavior of each model using a broad range of initial conditions when locating and following noisy odor trails. A combination of spatial and temporal comparison strategies produced the best odor-tracking behavior tested. Notably, models relying on only spatial comparison were unsuccessful in tracking the odor trail in these noisy conditions. These investigations suggest that a combination of temporal and spatial odor-signal comparison mechanisms may be an essential physiological component of biological scent-tracking behavior.

Major

Mathematics

Project Mentor(s)

Janice Thornton, Neuroscience

Document Type

Poster

This document is currently not available here.

Share

COinS
 
Sep 26th, 12:00 PM Sep 26th, 1:20 PM

Dynamical Systems Models of Odor-Tracking Behavior

Science Center, Bent Corridor

The ability to sense and track noisy environmental chemical cues is essential for survival in a multitude of diverse species, from bacteria to mammals. Despite the biological importance of this capability, the mechanisms used by most organisms, and especially rodents, to follow chemical trails remain poorly understood. Hence, we developed precise, testable computational models of odor-tracking behavior to investigate the mechanisms that underlie this widespread behavior. We created models simulating tracking mechanisms based on spatial (internostril) and temporal (sniff-to-sniff) gradients of odor concentration. Models were simulated via numerical integration using XPP/XPPAUT, and their parameters were tuned to represent mice given an odor trail tracking task. When odor trails were represented in an idealized, noiseless form, simple spatial comparison models accurately tracked the trails, yet temporal models were only successful with additional, more sophisticated features. Models incorporating both spatial and temporal comparisons also accurately tracked the odor trail. To test how well each mechanism performed in more realistic, noisy odor-signal conditions, we quantified the behavior of each model using a broad range of initial conditions when locating and following noisy odor trails. A combination of spatial and temporal comparison strategies produced the best odor-tracking behavior tested. Notably, models relying on only spatial comparison were unsuccessful in tracking the odor trail in these noisy conditions. These investigations suggest that a combination of temporal and spatial odor-signal comparison mechanisms may be an essential physiological component of biological scent-tracking behavior.