Both of these approaches suppress information at frequencies higher than the slower neural acquisition rate. In other cases, rates have been matched after the experiment, either by downsampling the fast variable to match the infrequent neural measurements 5, 8, 10, 11 or by interpolating the infrequent measurements to match the high-resolution variable 12, 13, 14, 15, 16, 17. In some cases, experiments were explicitly designed to match the two sampling rates 9. In practice, when computing cross-correlations, temporal receptive fields, or peristimulus time histograms, such data has most often been analyzed by matching the effective sampling rates of the two variables. The problem we address is how to relate neural activity that is sampled infrequently to other experimental variables that are sampled with high temporal resolution. These experimental variables are thus measured with high resolution in time.
And electrical signals like membrane potentials, local field potentials, or electromyograms are often recorded at 1 kHz or higher. Behaviors can often be measured from video recordings at 30 Hz or higher. Auditory stimuli may be modulated at frequencies of hundreds of Hz. Visual stimuli can be updated at 60 Hz or faster. In contrast, other experimental variables can be sampled or presented much more frequently. Thus, imaging techniques often result in a set of neural measurements that are located precisely in time, but sampled infrequently, once per volume. two-photon and confocal), may be too slow to capture fast dynamics of neural activity. Even when acquiring two-dimensional images, some imaging methodologies based on laser scanning (e.g.
Much effort has been invested in developing specialized hardware to increase the sampling rates of optical imaging 1, 2, 3, 4, yet imaging entire volumes (z stacks) frequently remains slow, with volumes typically acquired at ~10 Hz or slower 5, 6, 7, 8. Yet, because there are so many voxels to acquire, the period between measurements of a particular neuron can be much longer, equal to the duration of each full-image acquisition. Because the individual voxels are acquired over short time intervals, each neuron is measured over a short period of time. In these techniques, images are constructed from voxels (three-dimensional pixels) that are acquired sequentially. These imaging techniques include scanning two-photon, light sheet, and laser scanning confocal microscopy.
One of the most effective ways to do this is with optical measurements of neural activity, since these record the dynamics of many neurons at once. In investigating the function of circuits, experimenters often want to measure precise relationships between neural activity and other experimental variables, such as stimuli, behavior, or the activity of other neurons.
0 kommentar(er)
