Two-photon imaging in vivo has been used to monitor the activity of populations of cortical neurons. With some calcium dyes, these measurements have single-cell and single action-potential resolution. However, previous experiments looking at the spatiotemporal interactions of a large population of neurons have only been carried out in anesthetized animals. Until now, the relationship between population activity in the anesthetized animal and the awake animal has remained unclear. A new study on page 749 by Greenberg and colleagues now provides direct evidence to bridge this gap. Their findings suggest that the neuronal activity patterns of awake animals cannot be predicted from activity patterns seen under anesthesia.

The authors imaged calcium transients in layer 2/3 neurons in the visual cortex of awake, head-restrained rats, then anesthetized them and continued recording. In this way, they were able to measure the activity of the same neuronal population while the animals were awake and while they were anesthetized. The authors then used an algorithm that they had previously developed to identify action potential–evoked calcium transients based on the correspondence between the optical signal and the electrocorticogram. The algorithm was optimized using a separate set of data from simultaneous cell-attached electrophysiological and optical recordings in anesthetized animals.

Neuronal firing rates were much lower in anesthetized animals than in the awake animals, although the firing rates of individual neurons were correlated in the two states. During both anesthesia and awake periods, action potential firing in a single neuron temporarily increased action potential firing in the surrounding population; this effect was stronger for some neurons than for others and was stronger during anesthesia. Although neurons that were closer together were more likely to be influenced the same way by the onset of anesthesia, correlations between pairs of neurons in the awake state were independent of the correlations in the anesthetized state. There was a higher correlation between neurons in the anesthetized state, and population synchrony was also greater in the anesthetized condition.

These results represent the first direct comparison between neural population activity in awake and anesthetized animals. They suggest that the local patterns of activity in awake animals will probably provide stronger inputs to their targets in layer 5 than in anesthetized animals. More generally, these findings raise the important caveat that the spatiotemporal structure of action potential firing in a population of neurons recorded in an awake animal can not be directly inferred from recordings of the same neurons under the influence of anesthesia.