Neuromodulation-inspired gated associative memory networks: Extended memory retrieval and emergent multistability.

Activity-dependent neuromodulator-like gating in a minimal associative memory network bypasses the classical spin-glass transition, maintaining robust high-overlap retrieval far beyond the standard critical capacity.

• Why it matters: Classical autoassociative models ignore neuromodulatory agents that are known to shape memory capacity and stability, leaving attractor-structure changes under such gating unclear.
• What they did: The authors used many-body simulations and dynamical mean-field theory on a biophysically motivated network with self-adaptive, activity-dependent gating that can act as fully open or fully closed switches.
• The result: Gating stabilizes transient “ghost” remnants far above the Hopfield limit, converting one memory fixed-point attractor into a cluster of fixed points with graded dependence on initial overlap, enabling persistent input-driven computation.