Organisms face a hard problem: based on noisy sensory input, they must set
a large number of synaptic weights. However, they do not receive enough
information in their lifetime to learn the correct, or optimal weights
(i.e., the weights that ensure the circuit, system, and ultimately
organism functions as effectively as possible). Instead, the best they
could possibly do is compute a probability distribution over the optimal
weights. Based on this observation, we hypothesize that synapses represent
probability distributions over weights — in contrast to the widely held
belief that they represent point estimates. From this hypothesis, we
derive learning rules for both supervised and unsupervised learning. This
introduces a new feature: the more uncertain the brain is about the
optimal weight of a synapse, the more plastic it is. Consequently, the
learning rate of each synapse is adjusted on the fly. This framework makes
several testable predictions and, combined with the ansatz that more
uncertain synapses are more variable, it is consistent with current data.