RT Journal Article
SR Electronic
T1 Feedforward inhibition allows input summation to vary in recurrent cortical networks
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 109736
DO 10.1101/109736
A1 Mark H. Histed
YR 2017
UL http://biorxiv.org/content/early/2017/02/20/109736.abstract
AB Brain computations depend on how neurons transform inputs to spike outputs. Because sensory stimuli change activity in many interconnected brain areas, it has been challenging to control neuronal inputs to measure input-output transformations in vivo. To overcome that difficulty, here we paired optogenetic stimuli with a constant sensory stimulus and measured spiking of visual cortical neurons in awake mice. We found that neurons’ average responses were surprisingly linear. We then used a recurrent cortical network model to determine if these data and past observations of sublinearity could be described by a common circuit architecture. The model showed the input-output transformation could be changed from linear to sublinear with moderate (~20%) strengthening of connections between inhibitory neurons, but this change depends on the presence of feedforward inhibition. Thus, feedforward inhibition, a common feature of cortical circuitry, enables networks to flexibly change their spiking responses via changes in recurrent connectivity.