View Single Post
  #307  
Old 12-28-2017, 05:39 PM
SamuelA is offline
Guest
 
Join Date: Feb 2017
Posts: 3,693
Quote:
Originally Posted by wolfpup View Post
https://www.mitpressjournals.org/doi...n.1993.5.3.263

Stephen Kosslyn here argues that mental image processing involves the visual cortex, which is at odds with syntactic-representational model. Kosslyn is one of a number of opponents of CTM. Many others are dipshit philosophers but Kosslyn does real empirical work. Many of his conclusions are wrong, but it pains me greatly to have you on my side for the wrong reasons. Please side with Kosslyn and others like him and discredit them with your supportive idiotic bloviations. Please don't be on my side.

Seems like circular reasoning since you're making the unwarranted assumption that brain functions are digital.


Of course! Signals and Systems 101 and all of cognitive science is solved! How very SamuelA! What a total fucking dipshit moron!
I don't see answers to these questions. I see a false claim that all of cognitive science is solved but not answers that prove that you enough know enough to have a meaningful dialogue. I am not claiming to have solved it, if you had bothered to look up the answers, you'd realize there's not actually much wiggle room left for the idea that the brain is not a computational system.

An impulse with pulse edges and timing in a domain with noise can be discretized digitally with a numerical time resolution that need only be better than than SNR of the brain. (signal to noise ratio)

Perform this simple mental experiment. What if you could cut every axon and replace it with a system that digitizes the signal at the first node of ranvier and reinjects the signal, after a delay that is a discrete number of ticks from a digital system, at the last node of ranvier before the next synapse.

I am telling you there is firm, near absolute certainty mathematical proof that this experiment would produce the same outcome, so long as the clock resolution of this digital system is higher than the SNR of the system it emulates.

Similarly, if you think about it, each time a synapse receives an impulse, a certain amount of membrane charge is added or subtracted. This is an analog voltage but it has finite resolution. So you could in fact secretly replace (if you could do so, this is a thought experiment) each synapse with a digital counter, and that counter's numerical resolution need be no better than the SNR of that analog voltage.

Again, we're damn certain this is going to work.

Now, yes, there's other stuff neuroscience keeps finding. Other cells seem to be able to cleanup neurotransmitters and may be part of computation. There are concentration gradients of various hormones and modulation molecules.

There's long term changes to each synapse.

But you can trivially see, if you actually break the problem down, that you could in fact build a system that emulates a brain and responds to short term impulses in the exact same way as the original brain. It will work. Hormones and long term changes and long distance concentration gradients are all slow - you can in fact build a system that gives the same responses in the short term.

But again, just call me an idiot, whatever. I am well aware it's more complex, but also know that all analog systems can be replaced with a digital equivalent, you just have to discretize to above the SNR. This is a very well known principle in some fields...guess not yours. There are theories that maybe the brain is storing data in fragile q-bits or something, but these theories are probably wrong.

Assuming no quantum magic, the evidence is actually conclusive that you can emulate any and all systems the brain uses with digital equivalents. You can think of those digital equivalents as a very large truth table (since at a certain level they are), thus the brain is a computational system.

Last edited by SamuelA; 12-28-2017 at 05:42 PM.