Veronika Koren

This study shows that efficient coding principles can explain key properties of excitatory-inhibitory neural networks, including their connectivity and balance of excitation and inhibition.

The paper models how spiking neural networks efficiently process dynamic sensory features across layers, showing that feature mixing in neurons reduces metabolic cost and matches empirical observations in the mouse somatosensory system.

The paper examines population coding in V1 and V4 of macaques during a visual task, finding that in V4, informative neurons show stronger synchrony and correlations within coding pools than across them.

The paper predicts upcoming behavioral choices from spiking activity in monkey V1 by analyzing the overlap between stimulus and choice information in population vectors.

The paper introduces a model to read-out neural spike trains for stimulus discrimination, showing it outperforms population PSTH in classification and reveals functional subnetworks in V1 and V4.