Abstract
Hebbian plasticity, a mechanism believed to be the substrate of learning and memory, detects and further enhances correlated neural activity. Because this constitutes an unstable positive feedback loop, it requires additional homeostatic control. Computational work suggests that in recurrent networks, the homeostatic mechanisms observed in experiments are too slow to compensate instabilities arising from Hebbian plasticity and need to be complemented by rapid compensatory processes. We suggest presynaptic inhibition as a candidate that rapidly provides stability by compensating recurrent excitation induced by Hebbian changes. Presynaptic inhibition is mediated by presynaptic GABA receptors that effectively and reversibly attenuate transmitter release. Activation of these receptors can be triggered by excess network activity, hence providing a stabilising negative feedback loop that weakens recurrent interactions on sub-second timescales. We study the stabilising effect of presynaptic inhibition in recurrent networks, in which presynaptic inhibition is implemented as a multiplicative reduction of recurrent synaptic weights in response to increasing inhibitory activity. We show that networks with presynaptic inhibition display a gradual increase of firing rates with growing excitatory weights, in contrast to traditional excitatory-inhibitory networks. This alleviates the positive feedback loop between Hebbian plasticity and network activity and thereby allows homeostasis to act on timescales similar to those observed in experiments. Our results generalise to spiking networks with a biophysically more detailed implementation of the presynaptic inhibition mechanism. In conclusion, presynaptic inhibition provides a powerful compensatory mechanism that rapidly reduces effective recurrent interactions and thereby stabilises Hebbian learning.
The Stability Gap in Neural Plasticity
In computational neuroscience, Hebbian plasticity is a double-edged sword. While it allows networks to learn associations through the "fire together, wire together" principle, it inherently creates a "vicious circle": strengthening a connection makes neurons fire together more often, which further strengthens the connection. To prevent this from leading to pathologically high firing rates, the brain employs homeostasis to maintain activity within a healthy range.
However, a significant "stability gap" exists between theory and observation. Theoretical models suggest that for a network to remain stable, homeostatic mechanisms must act nearly as fast as the plasticity itself. In contrast, experimental evidence shows that most homeostatic mechanisms, such as synaptic scaling, are slow—taking hours or even days. This paper investigates whether presynaptic inhibition—a sub-second mechanism where GABA receptors reduce neurotransmitter release—can bridge this gap and provide the necessary rapid stabilization.
Mechanism: Presynaptic Inhibition as Rapid Feedback
The authors propose that presynaptic inhibition acts as a rapid, multiplicative gain control. Unlike traditional postsynaptic inhibition, which is often "additive" (subtracting from total input), presynaptic inhibition scales the synaptic weights themselves.
As shown in the biophysical model above, excess network activity leads to GABA spillover. This GABA binds to presynaptic receptors, which effectively and reversibly attenuates transmitter release. This creates a negative feedback loop that weakens recurrent interactions on sub-second timescales, providing immediate stability before slower homeostatic processes can take effect.
Results: Extending the Stability Range
Through rate-based and spiking neuron modeling, the study demonstrates that presynaptic inhibition fundamentally changes network dynamics. In standard excitatory-inhibitory (E-I) models, increasing recurrent excitation leads to a sudden "tipping point" where the network becomes unstable and firing rates explode toward a vertical asymptote.
With presynaptic inhibition, the network activity increases gradually and eventually saturates at a finite value ($1/\beta w_I$). This "graceful degradation" of stability prevents runaway excitation. Crucially, the inclusion of this mechanism increased the critical homeostatic time constant by more than tenfold. This means a homeostatic process that would normally need to act in seconds could now take minutes or hours and still maintain network stability, aligning computational models with biological reality.
Conclusion
Presynaptic inhibition provides a powerful compensatory mechanism that rapidly reduces effective recurrent interactions. By providing immediate, short-term stability, it "buys time" for slower homeostatic mechanisms to function effectively. This mechanism is robust across different mathematical implementations and biophysically detailed spiking networks, suggesting it is a fundamental component of stable Hebbian learning in the brain.
Future work may explore the spatial specificity of this inhibition—whether it is a global "spillover" effect or more localized—and the specific roles of different interneuron types in mediating this critical control loop.
Summary updated