In reply to tom_in_edinburgh:
> Just because we don't yet know how we experience things doesn't mean we should assume it can't be done by a physical machine based on a neural network paradigm.
Indeed, but until such a reductionist machine is also demonstrated to have a self awareness and an ability to experience, it would be equally unwise to assume it can be done by such a machine.
Such a machine doesn't have to be built - what is the program for a universal Turing machine to experience feelings and self awareness? This is an equivelant problem that removes the complexities of large neural networks, and one that is also unanswered to data.
> There is also a lot of evidence that the brain is a physical machine which implements a neural network.
It's not just a spiking neural network however. There are chemically diffusive modulators and sub-cellular computational potential. It's entirely possible that each individual neuron is also Turing complete. That could have any level of relevance from zero to massive on the neural network hosted on the cells.
> We don't know how far we can go with the neural network model, we are still making steady progress both in understanding the human/animal brain and in artificial intelligence. The appropriate ongoing assumption from the point of view of making testable hypotheses for experiments is that the brain is just a machine.
> It isn't helpful to declare consciousness is impossible for a mechanistic model and invoke 'get outs' like spirits or unspecified quantum mechanical effects until we get a lot further down the path of exploring what can be done without them
I'm in two minds on this. Building bigger - and massively more connected - neural networks is definitely one promising way forwards. I anticipate that it won't work, but I'd be happy to be proved wrong,
But if such a machine develops a consciousness and self awareness, it won't necessarily explain them. We'll just be left asking what it all means about two classes of consciousness instead of one.
"The hard problem of consciousness" is rooted in metaphysics, but there is no definitive refutation of its existence; lots of polarised expert opinion.
There is nothing mystical about speculating on quantum components - the light sleeting down on our eyes is full of quantum noise, macroscopic quantum effects are used in plant chlorophyll, it looks like some birds may make use of entangled quantum states in their vision. What else may be going on?
Rather, I would say the assumption you appear to hold, that the network activity hosted on a biological neural network is independent of the sub cellular behaviours, flies in the face of biological understanding where the evolution works very differently to human design where information processing crosses functional boundaries all the time. The effects of this may be small but small effects sometimes are very important. Again, building machines as people are doing may refute all this.
Post edited at 10:31