Logo Logo
Switch language to English
Nehrkorn, Johannes (2016): Olfactory learning in Drosophila: learning from models. Dissertation, LMU München: Graduate School of Systemic Neurosciences (GSN)



Animals are able to form associative memories and benefit from past experience. In classical conditioning an animal is trained to associate an initially neutral stimulus by pairing it with a stimulus that triggers an innate response. The neutral stimulus is commonly referred to as conditioned stimulus (CS) and the reinforcing stimulus as unconditioned stimulus (US). The underlying neuronal mechanisms and structures are an intensely investigated topic. The fruit fly Drosophila melanogaster is a prime model animal to investigate the mechanisms of learning. In this thesis we propose fundamental circuit motifs that explain aspects of aversive olfactory learning as it is observed in the fruit fly. Changing parameters of the learning paradigm affects the behavioral outcome in different ways. The relative timing between CS and US affects the hedonic value of the CS. Reversing the order changes the behavioral response from conditioned avoidance to conditioned approach. We propose a timing-dependent biochemical reaction cascade, which can account for this phenomenon. In addition to form odor-specific memories, flies are able to associate a specific odor intensity. In aversive olfactory conditioning they show less avoidance to lower and higher intensities of the same odor. However the layout of the first two olfactory processing layers does not support this kind of learning due to a nested representation of odor intensity. We propose a basic circuit motif that transforms the nested monotonic intensity representation to a non-monotonic representation that supports intensity specific learning. Flies are able to bridge a stimulus free interval between CS and US to form an association. It is unclear so far where the stimulus trace of the CS is represented in the fly's nervous system. We analyze recordings from the first three layers of olfactory processing with an advanced machine learning approach. We argue that third order neurons are likely to harbor the stimulus trace.