Detection of components in an odor mixture is tuned by experience and changes in the antennal lobe

MARACHLIAN EMILIANO; FERNANDO LOCATELLI

Washington DC

Congreso; Society for Neuroscience 2014; 2014

Odors in nature are complex mixture in whichirrelevant components may hide the presence of odors with predictive value. Moreoverthe meaning and relevance of each odor is not always fixed but may change    depending on the animal?s experience. Therefore,mechanisms must exist that allow animals optimize detection of the relevantodors according to its own experience. In agreement with this view we present behavioralexperiments performed in honey bees showing that appetitive learning of an odoris reduced when during learning the odor is presented in a mixture with anotherodor that has been learned before. The reduction in learning is interpreted asan overshadowing-like effect, in which the presence of the learned odor hindersthe perception of the novel one. This result poses the question about wherethis plasticity does occur along the olfactory processing pathway. Recentstudies have revealed that odor representation in the antennal lobe, the firstolfactory processing center in the insect brain, changes after olfactoryexperience. However the specific role of these changes remains still elusive.In the present work we test the hypothesis that plasticity in odor coding inthe antennal lobes is related with increasing the gain and detection ofrelevant odors on top of background and informative odors. To test this hypothesiswe performed calcium imaging in projection neurons of the antennal lobe to determinethe neural activity patterns that represent the pure odors and the respectivebinary mixture in naïve and in trained honey bees. Using the patterns measuredin naïve bees we established algorithms that allow accurate prediction of thepattern for the mixture based on the patterns measured for the pure components.The prediction algorithms obtained from naïve honey bees were applied to honeybees that had been trained on appetitive conditioning using as conditioned odorone of the components of the mixture. We found that the representation of themixture in trained animals deviates from the predicted mixture. The deviationfrom the predicted mixture is in favor of the representation of the learnedcomponent and away from the representation of the novel component. The changein the representation of the mixture is evidenced by a reduction in activity ofelements that encode the novel odor and not by an increase in activity ofelements that encode the learned component. This might indicate that changes inducedby training are caused by strengthening of the inhibitory interaction from thelearned odor toward the novel odor. Similar solutions might be applicable inother sensory modalities or even in the design of biomimetic sensors.