The Real Fruit Fly Brain Told Me Where I Was Cheating

The article discusses a project that aims to simulate the brain of a fruit fly. Initially, a tiny artificial brain was used for experiments, but the author eventually replaced it with the real thing - the mushroom body, a region of the fly's brain that plays a crucial role in associative learning and memory. The article highlights three main issues the author encountered while working with the real brain.

Firstly, the simulation showed that about 80% of the Kenyon cells, brain cells involved in processing sensory information, were active, which is not how a real fly's brain behaves. Real flies only respond to an odor by lighting up a small subset of their Kenyon cells. The author corrected this issue by incorporating inhibitory signaling, which helps regulate the activity of neurons and prevents the brain from "screaming" simultaneously.

Secondly, the author realized that the sensory inputs in their simulation were not biologically accurate. They were directly injecting activity into the Kenyon cells, which is not how a fly experiences sensory information. The correct process involves sensory cues being detected by olfactory receptors, then transmitted through projection neurons before reaching the Kenyon cells. The author planned to incorporate this multi-step sensory pathway in the next version of their simulation.

Lastly, the article notes that the initial use of the real connectome already yielded valuable insights. When the simulation produced different internal states for different sensory inputs, the author confirmed that the mushroom body wiring could indeed distinguish between different inputs, providing a basis for the memory-associated representations the project aims to explore.