Computational Models of Olfaction in Fruit Flies

[Extract]For species survival, an organism is required to obtain energy, avoid harm, and reproduce. In all these behaviors, a neural representation of the environment, through the process of “neural encoding,” is created. The main thrust of many neuroscience studies is the transformation of sensory information during initial detection, neural information processing, and generation of a percept that eventually drives specific behaviors. Survival of many organisms, in particular insects, depends heavily on olfaction (a form of chemosensation) to obtain vivid qualitative, quantitative (Keene & Waddell,  2007) and temporal (Laurent,  1999) information about the stimulus through detection of weak and fluctuating signals with large numbers of volatile chemicals (Firestein,  2001). Thanks to the striking structural and functional similarity of olfactory systems in animals and insects (Ache & Young,  2005), researchers can generalize (Olsen & Wilson, 2008) many principles of olfactory information processing (olfactory perception, discrimination, olfactory memories, and associative learning (Laurent et al., 2001)) across species. The small and manageable size of Drosophila melanogaster (briefly Drosophila or fruit fly), along with a comprehensive understanding of its olfactory system (including molecular description of olfactory receptor neurons), and recent advances in molecular, genetic, and neural activity recording make it a model organism to study olfactory information processing (Olsen & Wilson, 2008). Computational models provide valuable insights into information processing and transformation in terms of neural activity and plasticity for different odors/multi-odor mixtures. In this chapter, we first summarize the structure and function of neural substrates involved in Drosophila’s olfactory process followed by a description of information processing and associative learning. We then summarize the existing computational models of olfaction.