A burst-timing based learning rule at the retinogeniculate synapse

Daniel A. Butts, Patrick O. Kanold, Carla J. Shatz


Synaptic refinement in the mammalian lateral geniculate nucleus (LGN) requires neural activity. Learning rules that specify how combinations of pre- and postsynaptic activity lead to changes in synaptic efficacy are thought to underlie this activity-dependent development, though it is unclear how natural activity patterns drive such refinement. We address this question using an integrated experimental and theoretical approach. First, we determined what aspects of the spontaneous activity in the retina present during this development carry information that could instruct refinement, leading to the prediction of a learning rule based on the timing between presynaptic and postsynaptic bursts. Then, using perforated patch recording in slice preparations of early postnatal rat LGN, we recorded from LGN neurons and measured the size of postsynaptic currents evoked by optic tract stimulation, both before and after a burst-time-based stimulation paradigm. We found that the magnitude and direction of synaptic plasticity depends on the latency between pre- and postsynaptic bursts, suggesting how a "burst-time dependent" learning rule could use natural activity patterns to drive synaptic refinement. We then use simple simulations of the retinogeniculate system to demonstrate how the observed rules of synaptic plasticity observed in vitro at the single synapse can explain the system-level activity-dependent development observed in vivo.