hide
Free keywords:
-
Abstract:
Although we know a lot about how individual neurons in the brain represent the sensory environment, we are far from understanding how neural populations represent sensory information. Because individual neurons are noisy, pooling the activity of many neurons with similar response properties seems necessary to obtain an accurate representation of the sensory environment. However, it is widely believed that shared noise (or, noise correlations) in the activity of nearby neurons renders such pooling ineffective, profoundly limiting the accuracy of any
population code and, ultimately, behavior. This belief is based on model-based extrapolations from correlations measured in individual pairs of neurons, as it has been impossible to record simultaneously from complete neuronal populations. Here, we use a novel 3D high-speed in vivo two-photon microscope to record nearly all of the hundreds of neurons in a small volume of the mouse primary visual cortex and directly measure the amount of information encoded by these local populations. In contrast to previous predictions, we find that the information in a sensory population increases approximately linearly with population size and does not saturate even for several hundred neurons. Moreover, even a decoder ignoring correlations between neurons can decode 80 of the information in the population. Our results suggest that sensory neural populations represent information in a truly distributed manner and pooling of neural activity within local circuits is much more effective than previously anticipated. Thus, the representation in early sensory areas does not appear to be impaired substantially by shared sensory noise and limitations in behavioral performance in psychophysical tasks may need to be attributed to processes downstream of the sensory population.