Presentation Details
| Neural Representations and Task Design in Clinical Olfactory Testing Natalia Efimova, John P.McGann. Rutgers University, Piscataway, NJ, USA |
Abstract
Olfactory dysfunction is common and clinically significant, yet objective assessment of human smell function remains limited. Сlinical assays, such as the Threshold, Discrimination, and Identification (TDI) tests, are grounded in classical psychophysics but differ substantially in task structure, decision-making criteria, and cognitive demands. We modeled odor-concentration pairs as vectors in an odor representational space with variability and vector lengths scaling with response magnitude, inspired by neural population codes. Behavioral tasks were simulated as forced-choice decisions over noisy clouds of points corresponding to background and odor-evoked representations. Performance was quantified using signal detection theory and classification accuracy from linear discriminant analysis.
Simulations showed that improved performance with increasing stimulus intensity arises from increased separability of neural patterns relative to noise. Discrimination performance depended on task format: performance of the tasks comparing two stimuli was better when odors were highly overlapping, whereas performance of the tasks involving three stimuli became better once one odor clearly stood apart from the others. Identification efficiency was impaired when representations were highly similar, reflecting competition among multiple possible matches, but outperformed discrimination once odors were well separated. TDI tasks differed in difficulty not only due to the number of alternatives, but also because they impose distinct comparison operations and decision bounds. Our framework helps explain how differences in task structure and noise may contribute to divergent outcomes across olfactory assays and may offer guidance for the interpretation and design of clinical smell tests.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
Simulations showed that improved performance with increasing stimulus intensity arises from increased separability of neural patterns relative to noise. Discrimination performance depended on task format: performance of the tasks comparing two stimuli was better when odors were highly overlapping, whereas performance of the tasks involving three stimuli became better once one odor clearly stood apart from the others. Identification efficiency was impaired when representations were highly similar, reflecting competition among multiple possible matches, but outperformed discrimination once odors were well separated. TDI tasks differed in difficulty not only due to the number of alternatives, but also because they impose distinct comparison operations and decision bounds. Our framework helps explain how differences in task structure and noise may contribute to divergent outcomes across olfactory assays and may offer guidance for the interpretation and design of clinical smell tests.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
Content Locked. Log into a registered attendee account to access this presentation.