> more sophisticated cognitive abilities than previously known
No. More sophisticated cognitive abilities than previously modelled
using atemporal neural nets.
Anyone who has spent an afternoon being outwitted by a fly in their
study knows that even a common housefly can be cunning beyond belief.
It's not just that their reaction times out-pace humans. They seem to
remember, anticipate, plan and even deceive.
Those faculties seem inexplicable by a bottom-up network of 100,000
neurons using naive connectionist models we currently understand in
artificial networks. That's because time is missing from the
models. Biological networks are spiking, and (last I researched it)
can operate at up to 1kHz, and using integration, differentiation and
non-linear feedback can work like analogue computers (Hodgkin and
Izhikevich models). Instead of sending simple activation thresholds
around the network that's like sending network packets that can
perform remote procedure calls.
> more sophisticated cognitive abilities than previously known
No. More sophisticated cognitive abilities than previously modelled using atemporal neural nets.
Anyone who has spent an afternoon being outwitted by a fly in their study knows that even a common housefly can be cunning beyond belief. It's not just that their reaction times out-pace humans. They seem to remember, anticipate, plan and even deceive.
Those faculties seem inexplicable by a bottom-up network of 100,000 neurons using naive connectionist models we currently understand in artificial networks. That's because time is missing from the models. Biological networks are spiking, and (last I researched it) can operate at up to 1kHz, and using integration, differentiation and non-linear feedback can work like analogue computers (Hodgkin and Izhikevich models). Instead of sending simple activation thresholds around the network that's like sending network packets that can perform remote procedure calls.
And yes, they love bananas.