MYCOLab Ai Logo - gold and army green

The cellular architecture of memory modules in Drosophila supports stochastic input integration

Summary

Scientists created a detailed computer model of a memory-processing neuron in the fruit fly brain to understand how memories are stored and recalled. The study found that the neuron’s design allows it to store many different memories using random connections from input neurons, similar to how a brain might encode multiple learned experiences. This research reveals that memories can be efficiently stored without requiring precise positioning of individual neural connections, suggesting the brain uses flexibility and randomness as computational strategies.

Background

The mushroom body (MB) in Drosophila melanogaster serves as the fly’s learning and memory center. Understanding how cellular and circuit architectures contribute to efficient formation and storage of multiple memories remains largely unknown at the computational level.

Objective

To determine the computational principles underlying learning and memory in the context of decision making by building a realistic computational model of a central decision module (MBON-α3) within the Drosophila mushroom body, combining electron microscopy-based architecture with physiological measurements.

Results

MBON-α3 was found to be electrotonically compact with robust spiking responses to simulated odor input. The neuron’s dendritic architecture appears finely tuned to equalize the functional weight of synapses regardless of location, enabling efficient control by sparse KC innervation. Simulations showed that changes in either KC recruitment or synaptic strength by 25% produced approximately 20% changes in membrane depolarization, supporting stochastic memory storage.

Conclusion

The cellular architecture of MBON-α3 supports efficient storage of large numbers of memories through flexible stochastic connectivity. Sparse innervation by KCs can effectively control and modulate MBON activity with minimal requirements on synaptic localization specificity, providing computational principles for decision-making based on learned associations.
  • Published in:eLife,
  • Study Type:Computational Modeling Study,
  • Source: PMID: 36916672, DOI: 10.7554/eLife.77578
Scroll to Top