Gamma rhythms, the 30–80 Hz brain waves tied to attention and perception, have been too clean in every computer model built to date.
A study published in Communications Biology by Scott Rich, Tarun Shriram, and Steven A. Prescott shows how to weaken the model — deliberately — and why that produces something far closer to real biology.
The problem is a four-letter acronym with a thirty-year history: PING, the Pyramidal Interneuron Network Gamma mechanism. In simulations, PING networks fire with metronomic precision. Excitatory pyramidal cells spike, inhibitory interneurons answer, inhibition decays, and the cycle repeats — a strongly attracting limit cycle, in the language of dynamical systems.
The trouble, as the paper puts it, is that this rhythm is stronger and more stable than anything measured in a living brain, where gamma appears in brief bursts driven by sparse, seemingly chaotic spiking.
Real gamma is messy. Modeled gamma is a metronome. That gap between the two has bothered computational neuroscientists for decades, because a model that is too stable isn’t just aesthetically wrong — it can’t explain how brains actually use these rhythms.
Three ways to break a metronome
The UConn and Toronto-based team, whose affiliations span the University of Connecticut, the Hospital for Sick Children, the University of Toronto, and the University of Calgary per the paper’s own listing, found that three independent biophysical tweaks each weaken the PING attractor. None is exotic. All are features of the transient gamma activity observed in vivo.
- Increased heterogeneity — real neurons aren’t clones; their thresholds and firing properties vary.
- Synapse-like noise — real synaptic input is stochastic, not a clean current injection.
- A depolarized chloride reversal potential — a shift in the GABA-A reversal potential that occurs in real tissue and changes how inhibition lands.
Each adjustment, on its own, degrades the unrealistic synchronization. Minor alterations disrupt the stereotyped PING rhythm; larger ones prevent it entirely. What replaces it is the interesting part: spontaneous, transient increases in gamma-band power, with excitatory cells only weakly entrained to the population rhythm.

The Hopf bifurcation hiding in plain sight
Dynamical-systems theory says a stable focus and a limit cycle coexist near such bifurcations, and Hopf bifurcations were already characterized in Wilson–Cowan-style models of reciprocally connected excitatory-inhibitory populations, the same architecture PING requires.
That reframing matters. The team isn’t throwing out PING; they’re keeping its core ingredient — strong reciprocal excitation-inhibition coupling — and arguing that biophysical detail biases the system toward damped oscillations rather than a self-sustaining loop. The rhythm survives. The rigidity doesn’t.
There’s also a degeneracy result here that modelers in adjacent fields should sit up for: three mechanistically different perturbations converge on the same functional output. The paper’s significance statement, carried in the abstract indexed by the OUCI database, frames this as bridging “a major divide between these traditional in silico systems and the experimental reality” by showing that including biophysical detail “can promote more realistic activity in computational systems.”
Why this matters beyond neuroscience
Gamma oscillations are implicated in a long list of functions — binding features in perception, attention, memory formation — and their disruption is a recurring signature in conditions from schizophrenia to Alzheimer’s disease.
Sources agree on core claims
Nature’s reporting and the secondary coverage agree on the core claims, and the funding is disclosed in the paper itself: CIHR Foundation Grant 167276 to Prescott and a CIHR Postdoctoral Fellowship supporting Rich. The authors declare no competing interests, and the article is open access under a CC BY-NC-ND 4.0 license.
The paper proposes a revised explanation in which biophysical phenomena bias 30–80 Hz gamma activity towards damped oscillations rather than a stable limit cycle. The abstract stops at the proposal.
Rich heads the Rich Lab at the University of Connecticut, where he is an assistant professor of physiology and neurobiology.
He earned his BS at Duke University and his PhD at the University of Michigan, and his research centers on how heterogeneity and diversity across neurons contribute to physiologically relevant brain function — the same theme of variability that runs through the new paper. He was building the lab at the time of the study’s publication.
Rich, Shriram, and Prescott published their findings on November 20, 2026, in Communications Biology, in a paper whose title says it plainly: gamma activity becomes physiologically realistic by weakening the PING attractor state.