Error-backpropagation in temporally encoded networks of spiking neurons

How long does it take for the human visual system to process a complex natural image? Subjectively, recognition of familiar objects and scenes appears to be virtually instantaneous, but measuring this processing time experimentally has proved difficult. Behavioural measures such as reaction times can be used, but these include not only visual processing but also the time required for response execution. However, event-related potentials (ERPs) can sometimes reveal signs of neural processing well before the motor output. Here we use a go/no-go categorization task in which subjects have to decide whether a previously unseen photograph, flashed on for just 20 ms, contains an animal. ERP analysis revealed a frontal negativity specific to no-go trials that develops roughly 150 ms after stimulus onset. We conclude that the visual processing needed to perform this highly demanding task can be achieved in under 150 ms.

A paradox that exists in auditory and electrosensory neural systems is that they encode behaviorally relevant signals in the range of a few microseconds with neurons that are at least one order of magnitude slower. The importance of temporal coding in neural information processing is not clear yet. A central question is whether neuronal firing can be more precise than the time constants of the neuronal processes involved. Here we address this problem using the auditory system of the barn owl as an example. We present a modelling study based on computer simulations of a neuron in the laminar nucleus. Three observations explain the paradox. First, spiking of an 'integrate-and-fire' neuron driven by excitatory postsynaptic potentials with a width at half-maximum height of 250 micros, has an accuracy of 25 micros if the presynaptic signals arrive coherently. Second, the necessary degree of coherence in the signal arrival times can be attained during ontogenetic development by virtue of an unsupervised hebbian learning rule. Learning selects connections with matching delays from a broad distribution of axons with random delays. Third, the learning rule also selects the correct delays from two independent groups of inputs, for example, from the left and right ear.