Asymmetric synaptic plasticity provides an attractor-based mechanism for θ – γ coupling and pathological synchronization

Short-term synaptic plasticity (STP) continuously reshapes the balance between excitation and inhibition, yet most population models assume that excitatory and inhibitory synapses share similar short-term dynamics. Experimental studies, however, have shown that inhibitory synapses typically exhibit stronger short-term depression (STD) than recurrent excitatory connections. Here, we investigate how this cell-type-specific asymmetry influences cortical population dynamics by combining a neural mass model with large-scale spiking network simulations. We show that asymmetric inhibitory STD qualitatively changes the attractor structure of the network, generating a bistable regime that is absent under the conventional symmetric assumption. Within this regime, endogenous finite-size fluctuations drive spontaneous transitions between a stable focus and a stable limit cycle, giving rise to metastable dynamics. The asymmetric paradigm exhibits θ – γ phase–amplitude coupling of strength similar to that of the symmetric one, although it arises from fundamentally different dynamics. We further show that slowing synaptic recovery progressively eliminates this bistable regime, leading to persistent hypersynchronous oscillations and reduced phase–amplitude coupling. These results identify asymmetric inhibitory STD as a mechanism that reshapes cortical attractor dynamics and suggest how altered synaptic recovery may contribute to pathological network activity. Significance Statement Experimental studies have shown that inhibitory synapses often undergo stronger short-term depression than excitatory synapses, but the consequences of this asymmetry for cortical dynamics remain unclear. Combining a neural mass model with spiking network simulations, we show that asymmetric inhibitory depression changes the attractor structure of cortical networks, creating a bistable regime that supports fluctuation-driven metastable activity. The asymmetric paradigm shows θ – γ phase–amplitude coupling similar in strength to the symmetric one, but with a different dynamical origin. Our results suggest that cell-type-specific synaptic dynamics influence not only the strength of network activity but also its underlying dynamical organization, providing a possible mechanism for both flexible cortical processing and the emergence of pathological synchronization.