The ability to associate neutral stimuli with valence information and to store these associations as memories forms the basis for decision making. To determine the underlying computational principles, we build a realistic computational model of a central decision module within the Drosophila mushroom body (MB), the fly's center for learning and memory. Our model combines the electron microscopy-based architecture of one MB output neuron (MBON-α3), the synaptic connectivity of its 948 presynaptic Kenyon cells (KCs), and its membrane properties obtained from patch-clamp recordings. We show that this neuron is electrotonically compact and that synaptic input corresponding to simulated odor input robustly drives its spiking behavior. Therefore, sparse innervation by KCs can efficiently control and modulate MBON activity in response to learning with minimal requirements on the specificity of synaptic localization. This architecture allows efficient storage of large numbers of memories using the flexible stochastic connectivity of the circuit.

No abstract

The ability to associate neutral stimuli with either positive or negative valence forms the basis for most forms of decision making. Long-term memory formation then enables manifestation of these associations to guide behavioral responses over prolonged periods of time. Despite recent advances in the understanding of the neuronal circuits and cellular mechanisms controlling memory formation, the computational principles at the level of individual information processing modules remain largely unknown. Here we use the Drosophila mushroom body (MB), the learning and memory center of the fly, as a model system to elucidate the cellular basis of memory computation. Recent studies resolved the precise synaptic connectome of the MB and identified the synaptic connections between Kenyon cells (KCs) and mushroom body output neurons (MBONs) as the sites of sensory association. We build a realistic computational model of the MBON-α3 neuron including precise synaptic connectivity to the 948 upstream KCs innervating the αβ MB lobes. To model membrane properties reflecting in vivo parameters we performed patch-clamp recording of MBON-α3. Based on the in vivo data we model synaptic input of individual cholinergic KC-MBON synapses by local conductance changes at the dendritic sections defined by the electron microscopic reconstruction. Modelling of activation of all individual synapses confirms prior results demonstrating that MBON-α3 is electrotonically compact. As a likely consequence of this compactness, activation pattern of individual KCs with identical numbers of synaptic connection but innervating different sections of the MBON-α3 dendritic tree result in highly similar depolarization voltages. Furthermore, we show that KC input patterns reflecting physiological activation by individual odors in vivo are sufficient to robustly drive MBON spiking. Our data suggest that the sparse innervation by KCs can control or modulate MBON activity in an efficient manner, with minimal requirements on the specificity of synaptic localization. This KC-MBON architecture therefore provides a suitable module to incorporate different olfactory associative memories based on stochastically encoded odor-specificity of KCs.