Neural drift during rest drives walking direction and memory consolidation in Drosophila

The Hidden Purpose of Neural Drift

When animals rest, their brains do not simply go silent. In many species, specialized neurons responsible for navigation continue to fire in rhythmic, wandering patterns. For a long time, this "neural drift" was treated as mere biological noise. It was viewed as the restless static of a system waiting for the next command. But a new study in Drosophila suggests that this drift is anything but accidental.

The researchers propose that when a fly stops moving, its brain enters a structured state of "offline" processing. Rather than being useless fluctuations, these drifting signals may replay recent experiences. This process helps consolidate memories of where to go. By investigating the mechanics of this drift, the authors have uncovered a hidden dialogue between active movement and quiet rest.

The mystery of the wandering signal

The central question facing the researchers was how a brain maintains a stable sense of direction. Simultaneously, it must prepare for future movement. Specifically, they sought to understand the relationship between the head direction system and periods of inactivity. The head direction system is the neural circuitry that tracks an animal's orientation.

While animals use head direction cells to build spatial maps, the purpose of rest activity has remained elusive. Does the brain simply "reset" its compass when the animal stops? Or is there a computational reason for the neurons to keep moving? The authors aimed to see if neural drift could predict subsequent walking. They also wanted to see if these patterns change as an animal learns a specific goal.

Cracks in the "noise" hypothesis

Until now, the prevailing view was that neural activity during rest was largely unorganized. In many head direction systems, researchers expected the "activity bump" to behave predictably. An activity bump is a localized cluster of firing neurons representing a specific heading. Researchers expected this bump to either stabilize or dissipate when the animal stopped.

Prior studies on other head direction cells suggested the bump should remain fixed during short bouts of immobility. If the drift in the fly's fan-shaped body (a navigation substructure) was just noise, it should not influence future decisions. There was no theoretical reason to believe that "static" could steer an animal. This gap meant the connection between real-time navigation and memory consolidation remained a missing link.

Probing the drift with light and virtual reality

To dissect this mechanism, researchers placed tethered flies in a mechanical virtual reality (VR) arena [Figure 1a, b]. The flies walked on an air-supported ball to navigate a digital landscape. Using two-photon calcium imaging (a method to visualize neural activity), they monitored PFR neurons. These neurons are known to encode heading direction.

The investigation proceeded in three stages. First, they characterized natural drift. They discovered that PFR activity moves in a cosine-like wave across the fan-shaped body during rest [Figure 1e]. Notably, this drift moves at roughly twice the speed seen during walking [Figure 1h, i]. Second, they used optogenetics—using light to trigger specific neurons—to manipulate this activity. By stimulating PFR neurons during rest, they could bias the direction the fly would walk later .

Finally, they conducted a spatial learning assay. Flies learned to find a "cool area" in the virtual arena [Figure 3a, b]. As flies learned the goal, the PFR drift during rest changed predictably. The drift began to point away from the learned goal [Figure 3g, h].

A 180-degree reversal of intent

The findings reveal a structured relationship between rest and action. The authors report that the PFR phase distribution during rest is shifted by approximately 180° compared to walking [Figure 1l, m]. This is a consistent, mathematical inversion.

Crucially, the study shows that this drift is a vehicle for memory. During learning, the PFR drift during rest biased itself to point away from the learned goal [Figure 3h]. This suggests the drifting activity acts as a memory signature. When the fly rests, the brain may be replaying navigation history to prepare for future movement.

The researchers also identified downstream $h\text{!}$ neurons that act as a corrective mechanism. While PFR neurons undergo this 180° shift, $h\text{!}$ neurons receive input through a specific connectivity pattern [Figure 4a-e]. This pattern compensates for the flip [Figure 4f]. This ensures the final output remains coherent for steering.

The architecture of offline replay

The implications of this work extend beyond the fruit fly. If this mechanism is conserved across species, it suggests "rest" is a highly active computational state. It may be used to refresh and consolidate spatial maps.

The study provides a circuit-level explanation for how memory is held "offline." Using a ring attractor model (a mathematical framework for sustaining moving signals), the authors showed that synaptic depression drives this drift .

Synaptic depression is the temporary weakening of connections after use. This implies that navigating "exhausts" certain neural pathways. This forces the signal to drift toward unused neurons during rest.

This discovery reframes neural noise as a functional tool. It suggests the brain uses the transition between states to update its internal maps. A logical next step is to test if disrupting synaptic depression prevents memory consolidation. This would confirm if the drift is indeed the engine of learning.

Figures from the paper