Firing rates in visual cortex show representational drift, while temporal spike sequences remain stable.

Neural firing-rate responses to sensory stimuli show progressive changes both within and across sessions, raising the question of how the brain maintains a stable code. One possibility is that other features of multi-neuron spiking patterns, e.g., the temporal structure, provide a stable coding mechanism. Here, we compared spike-rate and spike-timing codes in neural ensembles from six visual areas during natural video presentations. To quantify information in spike sequences, we used SpikeShip, a method based on the optimal transport theory that considers the relative spike-timing relations among all neurons. For large numbers of active neurons, temporal spike sequences conveyed more information than population firing-rate vectors. Firing-rate vectors exhibited substantial drift across repetitions and between blocks, in contrast to spike sequences, which were stable over time. These findings reveal a stable neural code based on relative spike-timing relations in high-dimensional neural ensembles.