Student Presenter(s): Asim Ahmed
Faculty Mentor: Gonzalo Otazu Aldana
Department: Life Sciences
School/College: College of Arts and Science, Long Island

Rodents can identify target odors of interest in the presence of a strong background of odors in their natural environment. Experiments in the lab have shown that this identification ability can come after long training periods. However, it is not known whether an animal can generalize this identification ability to situations where the background is novel and there has not been a long training period. In order to test the generalization ability of mice, we presented awake—head fixed mice with target odors in the presence of a small set of background odors. We used intrinsic optical imaging to record dorsal glomerular activation patterns to these odor combinations (training set) and used those data to create a linear classifier for detecting the target odors. We then tested the learned linear classifier with odor mixtures that included a larger set of novel background odor. We found that the linear classifier performance varied between 65% and 90% accuracy for a set of 9 novel background odors. We then trained mice to perform a go/no-go task using the training set. Mice reached 90% performance after 8–10 days of training. We then tested the detection capability using a test set composed of novel odors. Mice correctly identified the target odors with the novel background odors at a rate that was consistent with the imaging data. Mice required an extra sniff to respond to the target odor and mice response times were slower in the novel odor environment compared to the response.