Current Research
Structured recurrent inhibition computes contrast across converging somatosensory signals
Koren V, Emanuel AJ, Panzeri S. Research manuscript in preparation.
We investigate how structured recurrent inhibition enables neural circuits to compute contrast across converging somatosensory inputs.
Talk — SNUFA 2024 · COSYNE 2025 poster, abstract and video · Related earlier preprint
Publications
Design principles for cost functions in optimal sensory coding
Koren V, Panzeri S. Frontiers in Computational Neuroscience 20:1924396, 2026.
We review how metabolic costs and constraints on downstream information transmission shape optimal sensory codes beyond encoding accuracy.
Paper
Efficient coding in biophysically realistic excitatory-inhibitory spiking networks
Koren V, Blanco Malerba S, Schwalger T, Panzeri S. eLife, 2025.
We show how efficient coding accounts for structured excitatory-inhibitory connectivity and biologically realistic coding, dynamical and biophysical properties of cortical networks.
Paper · Talk — SNUFA 2023 · Talk — CNS 2021 · Repository
Firing rates and representational error in efficient spiking networks are bounded by design
Urdu M, Lorenz MG, Huang CP, Panzeri S, Koren V. Artificial Neural Networks and Machine Learning (ICANN), 2025.
We demonstrate analytically that efficient spiking networks have bounded firing rates and representation errors, and that most spikes contribute to improving the neural representation.
Paper · Free PDF · Talk — ICANN 2025
Computational methods to study information processing in neural circuits
Koren V, Bondanelli G, Panzeri S. Computational and Structural Biotechnology Journal, 2023.
We review theoretical and computational methods for studying the encoding, transmission and decoding of information in neural circuits.
Paper
Biologically plausible solutions for spiking networks with efficient coding
Koren V, Panzeri S. Advances in Neural Information Processing Systems (NeurIPS), 2022.
We extend the theory of efficient coding to spiking networks with biologically plausible excitatory-inhibitory architectures, intrinsic currents, and structured recurrent connectivity.
Paper
Uncovering structured responses of neural populations recorded from macaque monkeys with linear support vector machines
Koren V. STAR Protocols, 2021.
We provide a step-by-step protocol for using linear support vector machines to study the structure of responses in simultaneously recorded neural populations.
Paper · Repository
Pairwise synchrony and correlations depend on the structure of the population code in visual cortex
Koren V, Andrei AR, Hu M, Dragoi V, Obermayer K. Cell Reports, 2020.
We show that in macaque V4, but not V1, neurons carrying similar information about behavioral choice exhibit stronger synchrony and correlations than neurons with different selectivity.
Paper · Repository · Talks — SfN 2019, Göttingen 2019, CNS 2016
Learning from invariants predicts upcoming behavioral choice from spiking activity in monkey V1
Koren V, Andrei AR, Hu M, Dragoi V, Obermayer K. bioRxiv preprint, 2020.
We develop a linear decoding approach that predicts upcoming behavioral choices from V1 population activity by exploiting representations shared across task conditions.
Preprint
Reading-out task variables as a low-dimensional reconstruction of neural spike trains in single trials
Koren V, Andrei AR, Hu M, Dragoi V, Obermayer K. PLOS ONE, 2019.
We develop a biologically motivated linear readout of spike trains that reconstructs task-related variables on single trials and reveals functional structure in neural populations.
Paper · Repository
Computational account of spontaneous activity as a signature of predictive coding
Koren V, Denève S. PLOS Computational Biology, 2017.
We show how analytically derived connectivity in efficient spiking networks generates spontaneous transitions between activity states, controlled by the metabolic cost of spiking and manifested as population bursts.
Paper