Prof. Alon Korngreen

Neuronal Biophysics and Computation

THE QUESTION: SINGLE-NEURON COMPUTATION

We study some of the basic yet still unresolved questions in neuroscience: How do neurons process information? What is the neuronal code at the cellular level? How does synaptic integration affect neuronal computation? How do neurons turn the activity of their molecular components into computation?

Understanding how individual neurons process and integrate information—single-neuron computation—is the overarching goal of our lab. Answering these questions demands models we can trust, yet most computational neuron models are built by compressing rich, variable measurements into a single best fit, hiding the very uncertainty that ought to be part of the scientific claim.

REFERENCE-GRADE MODELS OF NEURONAL EXCITABILITY

The approach we are currently pursuing is to generate reference-grade models of neuronal excitability: computational descriptions of voltage-gated ion channels and neurons that explicitly account for and propagate measurement uncertainty. Instead of relying on one representative model, we treat the experimental data as defining a constrained family of possibilities and ask which mechanistic claims that family genuinely supports.

METHODS: PATCH-CLAMP ELECTROPHYSIOLOGY AND COMPUTATIONAL ANALYSIS

To do this, we combine patch-clamp electrophysiology in heterologous expression systems and brain slices with computational tools: sensitivity analysis to identify the sources and effects of uncertainty, model structure and experimental protocol design guided by parameter identifiability, and covariance-aware parameter ensembles.

We have applied this framework to foundational models of neuronal electrical activity, from individual ion channel kinetics to action-potential generation.

TOWARD CERTIFIED ION CHANNEL STANDARDS

Our immediate goal is to develop a set of certified, openly documented ion channel models and a theory for combining them into models of neuronal excitability. Because ion channels interact electrically when embedded in the same membrane, their combined behavior cannot always be predicted from their individual properties alone. Establishing reliable standards for individual channels is therefore a step toward trustworthy computational models of cortical neurons.