Inhibitory-modulatory coupling generates persistent activity during working memory

Working memory requires the stable maintenance of neural representations across temporal gaps, yet the circuit mechanisms that generate and stabilize persistent activity remain unsolved. Prevailing models emphasize recurrent excitation as the principal substrate of persistence, but how inhibitory and modulatory interactions shape the stability of temporal dynamics is unclear. Here, using trace conditioning in Drosophila, a working memory-dependent form of associative learning, we identify reciprocal inhibition as a circuit mechanism for sustaining persistent activity. In trace conditioning, a trace interval separates the conditioned and unconditioned stimuli, requiring maintenance of a neural representation across the trace interval, to support learning. Combining virtual-reality behavior, targeted neurogenetic perturbations, in vivo two-photon calcium imaging, and real-time neurotransmitter measurements, we uncover a reciprocal inhibitory microcircuit within the ellipsoid body that is selectively engaged during trace, but not delay (overlapping CS-US), conditioning. During the trace interval, ER2/4m neurons exhibit sustained activity, while reciprocally connected ER3/4d neurons show progressively strengthened suppression, forming a dynamically stabilized inhibitory loop. Disrupting GABA synthesis or reception within this circuit abolishes persistent activity and impairs trace learning, demonstrating the causal requirement for reciprocal inhibition in working memory maintenance. We further show that glutamatergic and nitric oxide signaling enhance inhibitory efficacy during the trace interval. In vivo neurotransmitter imaging reveals temporally structured dynamics in which glutamatergic signaling precedes and amplifies sustained GABAergic inhibition, consistent with modulatory stabilization of circuit persistence. Together, these findings identify reciprocal inhibition, reinforced by modulatory signaling, as a core circuit mechanism for dynamically stabilizing persistent neural representations. Our results challenge excitation-centric models of working memory and establish inhibitory-modulatory loops as a fundamental substrate for maintaining memory traces across time. ### Competing Interest Statement The authors have declared no competing interest. United States Air Force Office of Scientific Research, FA9550-23-1-0024, FA9550-25-1-0299 Kavli Institute for Brain and Mind Innovative Research Grant, 2021-1779

BI 234 Juan Gallego: The Neural Manifold Manifesto | Brain Inspired

The digital sphinx: Can a worm brain control a fly body?

Animal intelligence is not purely a product of abstract computation in the brain, but emerges from dynamic interactions between the nervous system and the body. New connectome datasets and musculoskeletal models now enable integrated, closed-loop simulations of the neural and biomechanical systems of the fruit fly Drosophila, an ideal model organism to investigate embodied intelligence. However, many biological parameters of the nervous system and the body, as well as how they interface, remain unknown. To fill such gaps, researchers are turning to deep reinforcement learning (DRL), a data-driven optimization framework, to create virtual animals that imitate the behavior of real animals. Here, we provide a cautionary tale about the interpretation of such models. We constructed a virtual chimera of two phylogenetically distant species: a connectome of the C. elegans nematode worm and a biomechanical model of the fly body. The worm connectome receives sensory information from the fly body, and an artificial neural network is trained with DRL to map worm motor neuron activations to the fly's leg actuators. The resulting digital sphinx produces highly realistic fly walking - yet it is biologically meaningless. This exercise teaches us nothing about either animal and exposes a core peril of connectome-body models: behavioral fidelity is achievable without biological fidelity, making such models easy to overinterpret. Done carefully, virtual animals can be powerful partners to biological experiments, but only if their components and interfaces are grounded in biology. ### Competing Interest Statement The authors have declared no competing interest. NIH, U01NS136507, R01NS14543