CNS*2026 | Rich Lab | Department of Physiology and Neurobiology

Dr. Rich and Dr. Richard Gast have organized a workshop at CNS*2026 Halifax, entitled “Neuronal heterogeneity’s role in network dynamics and computation.” The workshop takes place on Tuesday, July 14 from 9:00 AM-12:30 PM local time. The schedule and abstracts for this session are below.

Schedule:

Time	Speaker	Title
9:00-9:25 AM	Dr. Scott Rich	Neuronal heterogeneity mechanistically distinguishes physiological and pathological oscillations
9:25-9:50 AM	Dr. Daniel Trotter	Structured by Disorder: Heterogeneity, Noise and the Resilience of Brain Dynamics
9:50-10:25 AM	Sanjna Kumari	Heterogeneities and dorsoventral gradients in intrinsic excitability of hippocampal granule and mossy cells in the rat hippocampus
10:25-10:40 AM	COFFEE BREAK
10:40-11:05 AM	Dr. Laura Medlock	Degeneracy in the Spinal Dorsal Horn: How Heterogeneous Network Architectures Shape Pain Processing
11:05-11:30 AM	Marco Zenari	Topological Origins of Timescale Diversity in Recurrent Neural Circuits
11:30-11:40 AM	COFFEE BREAK
11:40-12:05 PM	Dr. Richard Gast	The Critical Role of Neural Heterogeneity for Population Coding
12:05-12:30 PM	Dr. Andre Longtin	Heterogeneity enhances sequence memory and impedes AP backpropagation

Abstracts:

Dr. Scott Rich: Neuronal heterogeneity mechanistically distinguishes physiological and pathological oscillations

Oscillatory population activity arising from neuronal microcircuits, particularly at gamma (30-80 Hz) frequencies, serves a vital role in brain functions including learning, memory, and attention. Paradoxically, oscillatory activity also hallmarks epilepsy via seizure activity. However, there are key nuances distinguishing these physiological and pathological rhythms: seizures are typified by hyper-active neuronal spiking and long-lasting oscillatory activity, while physiological gamma activity consists of sparse spiking and lasts on the order of hundreds of milliseconds. From this perspective, it is notable that many computational explanations for oscillatory dynamics in neuronal networks, including the seminal Pyramidal Interneuron Network Gamma (PING) mechanism, produce dynamics more closely reproducing features of pathological oscillations. Here, we show that networks retaining the key elements of PING—namely, strong reciprocal connectivity between excitatory and inhibitory neurons—can produce more physiologically-realistic gamma activity when neuronal heterogeneity is accounted for rather than idealized away. We argue that such heterogeneities serve to promote oscillations driven by a damped oscillator rather than a stable limit cycle, accounting for key features distinguishing physiological and pathological rhythmicity.

Dr. Daniel Trotter: Structured by Disorder: Heterogeneity, Noise and the Resilience of Brain Dynamics

Diversity exists throughout biology, where it plays an important role in maintaining robustness and stability. The same is true of the brain, where recent datasets of unparalleled resolution have shown widespread heterogeneity, marking it as an unavoidable component of neural composition. While heterogeneity has been linked to stability and robustness, its loss has been shown in recent experiments to accompany pathological states, suggesting an important functional role. Despite this, how such changes in heterogeneity arise remains unknown. Often considered a static metaparameter resulting from solely genetic disposition, heterogeneity is, in fact, a highly dynamic property of biological networks arising from various sources. Using nonlinear random networks, we first show that intrinsic diversity enhances resilience by reducing network gain and simplifying the underlying dynamical landscape. Specifically, heterogeneity decreases the number of stable fixed points and reduces susceptibility to instability. Variability in neural activity, modelled as noise, similarly promotes robustness by linearizing network dynamics. Then, endowing a simple network model of excitatory neurons with a candidate mechanism for adaptation of neural excitability, we show how this adaptation may up- or down-regulate the network-level heterogeneity in response to external stimulation and network-related factors. Together, these results identify spatial and temporal diversity as complementary mechanisms that stabilize neural circuits and provide a unifying framework for understanding how diversity, modulation, and connectivity interact to preserve healthy network function.

Sanjna Kumari: Heterogeneities and dorsoventral gradients in intrinsic excitability of hippocampal granule and mossy cells in the rat hippocampus

The distinctions between the dorsal, intermediate, and ventral hippocampus are well characterized in terms of their behavioural roles, anatomical features, circuit connectivity, and gene expression profiles. However, physiological heterogeneity of neurons along the dorsoventral axis of the dentate gyrus (DG), the principal gateway to the hippocampus, remains poorly understood. Here, we used patch-clamp electrophysiology to systematically examine the intrinsic electrophysiological properties of the two excitatory DG neuron types, granule cells (GCs) and mossy cells (MCs), along the dorsoventral axis of the rat hippocampus. Both GCs and MCs exhibited pronounced dorsoventral gradients in intrinsic excitability, with dorsal neurons displaying lower input resistance, impedance amplitude, and firing rates than their ventral counterparts. Within GCs, blade-specific heterogeneity emerged where infrapyramidal blade neurons manifested higher firing rates than suprapyramidal blade neurons only in the ventral DG. GCs exhibited integrator-like dynamics characterized by low-pass impedance profiles, whereas MCs displayed sag, robust delta-frequency (0.5–4 Hz) resonance, and late firing. Pharmacological blockade of HCN channels abolished sag and resonance, while blockade of D-type potassium channels significantly reduced firing latency and enhanced firing rates, identifying key conductances mediating MC-specific intrinsic dynamics. Together, these findings reveal multi-regional intrinsic heterogeneity across DG excitatory circuits and underscore the functional complexity of hippocampus along its dorsoventral axis.

Dr. Laura Medlock: Degeneracy in the Spinal Dorsal Horn: How Heterogeneous Network Architectures Shape Pain Processing

Pain-related sensory input is processed in the spinal dorsal horn (SDH) before being relayed to the brain. This processing depends on interactions between heterogeneous populations of excitatory and inhibitory interneurons, but it remains unclear how these populations collectively shape circuit output. Using a multiscale, experimentally constrained model of the rodent SDH, we found that many distinct synaptic weight configurations could reproduce the same circuit responses to mechanical stimulation under baseline conditions. This degeneracy has important implications for understanding spinal circuit function and pain processing: circuits that appear equivalent in their baseline behaviour may rely on different hidden network architectures and thus respond differently to pathological perturbation. In this talk, I will focus on how degeneracy and heterogeneity emerged from the model and what they imply for chronic pain mechanisms. Across optimized model solutions, equivalent baseline behaviour could arise from different balances of excitation and inhibition, different routes by which low-threshold input accessed nociceptive projection neurons, and compensatory trade-offs between inhibitory interneuron populations. These latent differences became visible only after pathological perturbation, such as disinhibition or interneuron ablation, where previously equivalent circuits produced divergent projection neuron output (i.e. different pain). The results suggest that heterogeneity in SDH circuit parameters may help explain why similar pathological insults or therapeutic interventions can have different effects across individuals. More broadly, this work argues that degeneracy is not simply a modelling nuisance or parameter-identifiability problem. Instead, it may be a biologically meaningful feature of spinal sensory processing, allowing robust function under normal conditions while creating circuit-specific vulnerabilities under chronic pain states.

Marco Zenari: Topological Origins of Timescale Diversity in Recurrent Neural Circuits

Structural and functional heterogeneities are defining features of cortical circuits, ranging from broad connectivity distributions in neural networks to diverse intrinsic timescales across neurons. However, the mechanisms by which heterogeneity in connectivity gives rise to functional diversity remain unclear. A theoretical framework is presented that links network topology to the emergence of heterogeneous neural dynamics. Recurrent networks with heterogeneous connectivity and partially symmetric synaptic interactions are considered and a heterogeneous dynamical mean-field theory is derived to characterize neuronal activity as a function of connectivity. The theory reveals that the interplay between degree heterogeneity and partial symmetry generates effective self-couplings whose strength increases with node degree, producing a broad distribution of neuronal timescales. Highly connected neurons are therefore predicted to exhibit slower dynamics, shaping network stability and supporting multiscale computation. Finally, by instantiating the model with the topology of the MICrONS mouse cortical connectome, it is shown that realistic connectivity structure reproduces experimentally observed relationships between neuronal connectivity and intrinsic timescales. These results provide a mechanistic link between connectome topology, neural dynamics and computation.

Dr. Richard Gast: The Critical Role of Neural Heterogeneity for Population Coding

Recent advances in transcriptomics, projection tracing, and multi-neuron recordings reveal that neural heterogeneity — the continuous diversity of cellular properties — is a pervasive feature of brains. But what is the functional relevance of this heterogeneity? In the traditional view, differences between cell types are considered as the functionally relevant aspect of neural heterogeneity, whereas the contribution of additional heterogeneity within cell types is considered negligible. In this talk, I will argue that the continuously distributed physiological properties existing within cell types play an important functional role that is complementary to the role of discrete cell types. In the first part of the talk, I will discuss results obtained in models of spiking neural populations that reveal how neural heterogeneity controls the capacity of the population to . In the second part, I will present recent, unpublished results on a new mean-field method that allows (a) to apply these concepts to empirical data, and (b) to study the interaction of neural heterogeneity and synaptic plasticity.

Dr. Andre Longtin: Heterogeneity enhances sequence memory and impedes AP backpropagation

Neurons that exhibit spike frequency adaptation can represent time intervals between events because their excitability increases in proportion to time since a last stimulus. This property has been described as a neural encoding of the Laplace transform of the input. It has been seen to operate in hippocampus and prefrontal cortex, and is a putative method for encoding time sequences in pallium in e-fish thereby supporting path integration. Previous studies have considered adaptation that resets after each event, but many such cells carry some memory of previous firings through non-resetting mechanisms. Furthermore, there is broad heterogeneity of adaptation time constant, gain, threshold and degree of reset. I will present a general theory that shows, using mutual information calculations, the benefits of heterogeneity for encoding sequences. We find that it is required for encoding more than one interval; in fact, there needs to be at least as many homogeneous subpopulations as intervals to encode. Heterogeneity in the threshold parameter also allows for a division of labor, a fact that provides an additional perspective on the usefulness of rectified linear units in machine learning. A second part of the talk will focus on the spatial heterogeneity of different types of voltage-dependent sodium channels in the axon initial segment in layer 5 pyramidal cells. Their spatial segregation varies during development, and we find that this regulates the threshold for AP backpropagation and thus for certain forms of plasticity. This is shown using biophysically detailed NEURON modeling as well as a simplified three-compartment model. Finally, I will say a few words about a novel method for estimating parameter (and more generally model) heterogeneity from neural spike train data. A novel method to compute the risk distribution allows the selection of the *best* model given the experimental uncertainty.