Comments on “Quantum Mind”

 

How does quantum mechanics connect to mind or intelligence? There have been many debates on that subject lately, ranging from solid and serious to utterly florid and incoherent imaginings. (Don’t worry, I won’t name names for the latter.) Here I will try to disentangle a few basic points.

            First, of course, almost everyone agrees that we need quantum mechanics to describe the lowest level of what happens in molecules or chips. That is THE physical reality, for all practical purposes, when we try to understand the brain.

            However, there is a big difference between “quantum devices” and “quantum systems.” In electronics, it is vey common to use all kinds of advanced quantum theory in order to build a simple little box which acts like a simple, classical “AND gate.” We build up large computing systems by combining many, many simple devices, each of which is designed to perform a straightforward function that doesn’t involve any kind of quantum mechanics and quantum logic. I personally tend to believe that the human brain is also based on that kind of principle, but – please forgive me – that there is also another part of our mind, which I think of as “the soul,” which does not. (I am not advocating classical Dualism, but something a bit more complex.)

            In hard core, mainstream physics and engineering research, most informed people now agree that we can (and have!) build “quantum computers,” a new kind of computer which exploits quantum principles at a systems level to achieve new kinds of capability beyond capacity of computers built up from simple devices in a classical way. But is the human brain (or even the mouse brain?) such a computer? I tend to doubt it.

            In the community which believes most in the concepts of “quantum mind,”  Stuart Hameroff and Michael Conrad have been, perhaps, the most articulate spokesmen. Conrad, in particular, developed what I view as the most refined and plausible version of the quantum mind hypothesis, though I also recall a Japanese speaker at one of Hameroff’s conferences whom I wish I remembered more.

            Conrad argued, in effect, that the molecules inside each cell might be implementing a kind of quantum associative memory. This would enable individual neurons to perform complex associative memory tasks which a classical neuron cannot. I personally feel that the probability of this is less than 20 percent, but (especially when we are involved in funding research) we need to work to keep an open mind. To act in a proper way, I challenged the quantum mind community: can you demonstrate this empirically? Can you train a single neuron, in culture on a chip, to learn to perform a complex mapping, such as XOR or the N-bit parity problem, which an associative memory neuron could perform easily but a more clasical neuron could not? Maybe this wouldn’t be the final, definitive test of quantum mind, but it would at least give us more basis for taking the idea seriously (if the data are consistent with it), and it would start to tell us real things about real neurons and real brains, at a functional level. No one took me up on it – yet. But people like Jose Principe at the University of Florida certainly are working hard on seeing what neurons really can learn --  in a systematic, task-oriented way – in refined, controlled environments.

            But again – even if (as I expect) we find no evidence for “quantum power” in the brain, that doesn’t mean that we cannot build it in machines.

            At this writing (2006), the hard core quantum computing community is somewhat discouraged. Yes, they have built and tested real quantum computers. But will they ever be able to scale them up to be truly relevant to practical or large-scale applications of any kind? (General-purpose intelligence is, of course, a large-scale phenomenon, requiring a lot more than a handful of quantum bits or “qubits”) Can we even find general purpose “programs” that could take advantage of quantum computers, for anything but factoring numbers and breaking codes? My personal feeling is that we probably can, for two reasons. First, we can take advantage of learning to develop quantum computers which exploit their massive powers without having to be programmed. (This idea is discussed in some of my older papers, but has come much closer to reality due to efforts such as those of Elizabeth Behrman and Dvaid Meyers, who have led workshops on that topic in recent years.) Second, a new understanding of the foundations of quantum theory suggests a new way of thinking about quantum computing which helps to explain some of the frustrations of the past – and points to new ways of understanding and exploiting what is actually possible.

            Curiously enough, this leads to a view of the mind which is not so different, in the end, from the picture in Dan Simmons’ science fiction series Hyperion.

            There is a totally different stream of thinking, in physics, which asks whether the universe itself – the underlying laws of reality – might actually look like “A Great Mind” instead of a “Great Machine.” Personally, I believe that we have much more urgent and compelling research to do at the present time, in order to understand what is confronting us right now in physics – but sooner or later, it does make sense to look for clues that might justify such a different possibility. It does seem more plausible to me, ultimately, than Conrad’s hypothesis. For the moment, my current thoughts about how to explore the digital universe idea are best expressed in an old email or two. One of the key challenges would be to actually prove some of the interesting claims that Wolfram makes about being able to reproduce general relativity as an emergent consequence of a more digital model. (Unfortunately, I do not have the most relevant email on that here at hand as I type this…)