Pythagorean
Oct18-09, 06:29 PM
I'll discuss the theoretical part, then I'll get to how we'd empirically test it, then I'll discuss it in the context of computer-simulated neural networks.
Theoretical
Dynamic: the system of neurons is always running. The memory is not "stored" safely in a static place, but is a dynamic state of inter-neural communication.
Static: The state can be "saved" in a static configuration and can be recalled later. The configuration of the neural connections might then be a large part of what is long-term memory.
Mixed: Long-term memory is a system involving both the static state of the system and the dynamic state of neural activity.
Empirical
Like most neurology/psychology questions, this would rely on medical results. The best example would be somebody dying to the point where they have absolutely no brain activity, and then being brought back to life. (This isn't the same as immeasurable brain activity).
Are there any cases where we can confirm that all (or at least a region) of the brain has lost all neural activity, and then the patient was brought back to life and any differences in memory were noted?
Neural Networks
Would this stand to reason that a simulated neural network can only represent long term memory correctly as long as it is running, and when we stop it and save the state, we're losing information? Or can we somehow store the dynamic state of the system and recall it as well?
Theoretical
Dynamic: the system of neurons is always running. The memory is not "stored" safely in a static place, but is a dynamic state of inter-neural communication.
Static: The state can be "saved" in a static configuration and can be recalled later. The configuration of the neural connections might then be a large part of what is long-term memory.
Mixed: Long-term memory is a system involving both the static state of the system and the dynamic state of neural activity.
Empirical
Like most neurology/psychology questions, this would rely on medical results. The best example would be somebody dying to the point where they have absolutely no brain activity, and then being brought back to life. (This isn't the same as immeasurable brain activity).
Are there any cases where we can confirm that all (or at least a region) of the brain has lost all neural activity, and then the patient was brought back to life and any differences in memory were noted?
Neural Networks
Would this stand to reason that a simulated neural network can only represent long term memory correctly as long as it is running, and when we stop it and save the state, we're losing information? Or can we somehow store the dynamic state of the system and recall it as well?