8 ms·

Roger Penrose Discusses Consciousness (2017)

72 points by howard941 7y ago

dang 7y ago
Discussed at the time: https://news.ycombinator.com/item?id=14263842 https://news.ycombinator.com/item?id=14263842
millettjon 7y ago
Penrose was on Joe Rogan last December: https://www.youtube.com/watch?v=GEw0ePZUMHA https://www.youtube.com/watch?v=GEw0ePZUMHA
danbruc 7y ago
I share most if not all of the objections mentioned in the article. But even if one grants that quantum physics or quantum gravity or even undiscovered physics plays a role, I still fail to see how this can be of much help in explaining consciousness. Quantum mechanics can be simulated classically with an exponential slowdown. Not that I want to belittle an exponential difference but in some sense it does not bring anything fundamentally new to the table. Quantum mechanics is - setting all the still unsolved problems aside - still just a set rules dictating - or describing - how things have to behave. And even undiscovered physics would have to have a very different character than known physics in order to provide something truly new and capable of bridging the perceived gap between physics and consciousness. Personally I changed my mind quite recently and the following quote from the article now resonates very much. Minsky called consciousness a “suitcase word” that lacks the rigor of a scientific concept. “We have to replace it by ‘reflection’ and ‘decisions’ and about a dozen other things,” he said. “So instead of talking about the mystery of consciousness, let’s talk about the 20 or 30 really important mental processes that are involved. Just sit down and try to nail down what you mean when you say, you are conscious or you feel pain. Try to make precise what you think distinguishes you from a philosophical zombie. When the voice in your head says that this was not the best decision and next time you should do something else, is that really any different from one system monitoring and controlling another system? When you are experiencing pain, when you try to get away from its source, when you maybe scream, is that really any different from you car turning on the low oil pressure warning light? Of course it is! Or is it? Maybe your brain just turned the I-am-special indicator light on...
- mathgenius 7y ago
  > Quantum mechanics can be simulated classically with an exponential slowdown. This is actually quite subtle. For example, I don't see how you can "simulate" bell non-locality. Is it even possible to define the word "simulate" here? The whole status of quantum randomness is unsettled, I don't think it can be settled, but anyway it is not as straightforward as you suggest. Scott Aaronson has a nice paper where he discusses related ideas: https://arxiv.org/abs/1306.0159 https://arxiv.org/abs/1306.0159
  - danbruc 7y ago
    One of the axioms of quantum mechanics is that the wave function completely describes a physical system. Therefore all you have to do in order to simulate a quantum system is to compute the evolution of the wave function which is pretty straight forward given an initial state and the Hamiltonian of the system - you just evaluate the Schrödinger equation. It is however computationally intensive for systems of non-trivial size. Also the evolution of the wave function is unitary, i.e. it is completely deterministic forward and backward in time and there is no randomness involved at all. Quantum non-locality can not be simulated classically if you distribute your computations just as the simulated system is distributed and if you do not allow for communication between the computers. This is however not an issue if you perform the simulation in one place or allow for communication between the computers. An example may help. Take the classical experiment of producing two spins with opposite orientations in one place, then separating them by a large distance, and finally measuring them at those distant locations. You can easily simulate that with one classical computer, but you can not use two classical computers - one to simulate each spin and the measurement of that spin at the distant location - without allowing the two computers to communicate even after the time the spins have been separated [in the simulation]. This especially highlights that quantum mechanics can not be described by a local hidden variable theory, or maybe it is better to think about it the other way round, such an simulation is not possible because quantum mechanics can not be explained by a local hidden variable theory as predicted by the Bell inequality and demonstrated by Bell test experiments. You can not write an algorithm that generates some hidden state while the two spins respectively the two computers are not yet separated and can therefore still communicate with each other and that then reproduces the results of all possible measurement experiments after the separation and without further communication.

72 points by howard941 7y ago

dang 7y ago

Discussed at the time: https://news.ycombinator.com/item?id=14263842 https://news.ycombinator.com/item?id=14263842

Penrose was on Joe Rogan last December: https://www.youtube.com/watch?v=GEw0ePZUMHA https://www.youtube.com/watch?v=GEw0ePZUMHA

danbruc 7y ago

I share most if not all of the objections mentioned in the article. But even if one grants that quantum physics or quantum gravity or even undiscovered physics plays a role, I still fail to see how this can be of much help in explaining consciousness. Quantum mechanics can be simulated classically with an exponential slowdown. Not that I want to belittle an exponential difference but in some sense it does not bring anything fundamentally new to the table. Quantum mechanics is - setting all the still unsolved problems aside - still just a set rules dictating - or describing - how things have to behave. And even undiscovered physics would have to have a very different character than known physics in order to provide something truly new and capable of bridging the perceived gap between physics and consciousness. Personally I changed my mind quite recently and the following quote from the article now resonates very much. Minsky called consciousness a “suitcase word” that lacks the rigor of a scientific concept. “We have to replace it by ‘reflection’ and ‘decisions’ and about a dozen other things,” he said. “So instead of talking about the mystery of consciousness, let’s talk about the 20 or 30 really important mental processes that are involved. Just sit down and try to nail down what you mean when you say, you are conscious or you feel pain. Try to make precise what you think distinguishes you from a philosophical zombie. When the voice in your head says that this was not the best decision and next time you should do something else, is that really any different from one system monitoring and controlling another system? When you are experiencing pain, when you try to get away from its source, when you maybe scream, is that really any different from you car turning on the low oil pressure warning light? Of course it is! Or is it? Maybe your brain just turned the I-am-special indicator light on...

mathgenius 7y ago
> Quantum mechanics can be simulated classically with an exponential slowdown. This is actually quite subtle. For example, I don't see how you can "simulate" bell non-locality. Is it even possible to define the word "simulate" here? The whole status of quantum randomness is unsettled, I don't think it can be settled, but anyway it is not as straightforward as you suggest. Scott Aaronson has a nice paper where he discusses related ideas: https://arxiv.org/abs/1306.0159 https://arxiv.org/abs/1306.0159
- danbruc 7y ago
  One of the axioms of quantum mechanics is that the wave function completely describes a physical system. Therefore all you have to do in order to simulate a quantum system is to compute the evolution of the wave function which is pretty straight forward given an initial state and the Hamiltonian of the system - you just evaluate the Schrödinger equation. It is however computationally intensive for systems of non-trivial size. Also the evolution of the wave function is unitary, i.e. it is completely deterministic forward and backward in time and there is no randomness involved at all. Quantum non-locality can not be simulated classically if you distribute your computations just as the simulated system is distributed and if you do not allow for communication between the computers. This is however not an issue if you perform the simulation in one place or allow for communication between the computers. An example may help. Take the classical experiment of producing two spins with opposite orientations in one place, then separating them by a large distance, and finally measuring them at those distant locations. You can easily simulate that with one classical computer, but you can not use two classical computers - one to simulate each spin and the measurement of that spin at the distant location - without allowing the two computers to communicate even after the time the spins have been separated [in the simulation]. This especially highlights that quantum mechanics can not be described by a local hidden variable theory, or maybe it is better to think about it the other way round, such an simulation is not possible because quantum mechanics can not be explained by a local hidden variable theory as predicted by the Bell inequality and demonstrated by Bell test experiments. You can not write an algorithm that generates some hidden state while the two spins respectively the two computers are not yet separated and can therefore still communicate with each other and that then reproduces the results of all possible measurement experiments after the separation and without further communication.