1 There is a fundamental gap in our understanding of the circuit mechanisms underlying even simple naturalistic 2 behaviors, such as making a cup of coffee, which proceed through a sequential execution of sub-behaviors. 3 Continued existence of this gap represents an important problem because obtaining a circuit-level understand- 4 ing of complex multi-step behaviors is a necessary step toward unlocking the mysteries of healthy brain func- 5 tion and of disorders. The overarching goal is to obtain a circuit-level understanding of such naturalistic behav- 6 ior. The research objective here is to unravel the logic of sensorimotor transformation in the context of odor- 7 modulation of locomotion in Drosophila. The central hypothesis is that, like many of our own everyday actions, 8 control of odor-modulation of locomotion is hierarchical. A fly’s locomotion is built from simpler elements called 9 locomotor primitives, each of which lasts between 1-3 seconds (or 10-30 steps). Odors, instead of acting on 10 instantaneous locomotor parameters such as speed and angular speed, act on these locomotor primitives and 11 change the probability that the fly spends performing a given locomotor primitive. This hypothesis was formu- 12 lated on the basis of our previous work and preliminary data. The rationale for the proposed research is that 13 understanding odor-guided locomotion—a complex, flexible behavior—in the context of a genetically tractable 14 system will allow a precise delineation of the steps that underlie sensorimotor transformation in the context of a 15 naturalistic behavior. The hypothesis above will be tested by characterizing the circuit basis of modulation of 16 locomotion by food odors using a combination of techniques including imaging, electrophysiology, quantitative 17 behavior and computation. The proposed research has three specific aims. 1) To extract the locomotor primi- 18 tives and test the hypothesis that odors modulate locomotion by changing the time a fly spends performing dif- 19 ferent locomotor primitives. 2) To test the hypothesis that different ORN classes modulate the time spent in 20 distinct locomotor primitives. 3) To elucidate the role of lateral horn in odor modulation of locomotion. The re- 21 search is innovative because it employs sophisticated statistical tool (Hierarchical Hidden Markov model, 22 HHMM) and cutting-edge experimental tools in the context of a genetically tractable model organism to obtain 23 insights into naturalistic behaviors. The proposed research is significant because it will vertical advance our 24 understanding of sensorimotor processes involved in naturalistic behaviors. Insights from multiple fields have 25 all come to the conclusion that behavior is organized into discrete packets or behavioral primitives. Actions un- 26 fold by a sequential recruitment of these discrete packets. A critical barrier to the study of natural behavior is 27 that in most cases there is enough variabil...