[robg]

I was raised on this era - you could go down quite the rabbit hole branching out from McClelland and Geoff Hinton. They were trying to reflect the brain and so backprop was initially seen as a shortcut that could be done away with once more complex models could be supported by processing power and inputs.