Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz

Gamma Oscillations in QIF based E-I Network

Poster Session D - Monday, March 9, 2026, 8:00 – 10:00 am PDT, Fairview/Kitsilano Ballroom

Qiyun Zheng^1,2 (), John Rinzel²; ¹Cold Spring Habor Laboratory, ²New York University

Gamma-frequency oscillations are essential for cortical information processing, with Pyramidal-Interneuron Network Gamma (PING) emerging through E-I interactions or Interneuron Network Gamma emerging through I-I interactions. Previous Quadratic Integrate-and-Fire (QIF) network models relied on unrealistic inhibitory reversal potentials or mixed current-injection/conductance inputs, limiting biological relevance and analytical clarity. As a result, we developed a mean-field firing rate equations (QIF-FRE) to analyze macroscopic dynamics using exclusively conductance-based external and synaptic inputs with biologically plausible reversal potentials and an all-to-all connected QIF spiking network for verification of microscopic dynamics. Network dynamic was systematically characterized by varying external conductance drive and synaptic coupling strengths. Our model successfully generated 50Hz PING oscillations with realistic parameters. As external excitatory conductance input to excitatory population gradually increased, we observed three distinct dynamical regimes: (1) robust PING oscillations, (2) transition from PING to weak ING, and (3) oscillation quenching via subcritical Hopf bifurcation. Due to intractability to mean-field solution, we also derived analytical solution to single QIF neuron model which revealed that conductance-based inputs create voltage-dependent phase space compression, predicting the critical transition point in network. Additionally, we identified parameter regimes producing atypical synaptic currents due to QIF resetting properties and demonstrated that increasing simultaneously synaptic inhibitory conductance input to excitatory and inhibitory population normalizes these currents. Our work indicates that conductance-based QIF E-I networks generate biologically plausible gamma oscillations while revealing novel dynamics including input-dependent mechanism transitions and QIF’s limitation in modeling plausible neural network dynamics.

Topic Area: OTHER