How close are we to developing artifical minds?

The last few days, I’ve spent some time doing some research I’d been avoiding. It’s no secret that I am interested in astronomy, am not afraid of chemistry or physics, but I’ve rather been avoiding medical science, since I don’t feel particularly qualified to tackle it.

But since I’m developing a world in which artificial humans exist, and where people’s minds get transferred from one body to another, I thought I had better read up on the sort of stuff we can do, or are close to doing, in terms of replicating human brain functions.

And wow, I was impressed.

The human brain contains up to 33 billion neurons, long-armed, spidery cells that may have up to 10,000 synapses each. An adult human brain is thought to have up to 500 trillion synapses. The synapse, at the end of a neuron cell, possesses mechanisms to transfer tiny electrical currents to surrounding tissue, prompting this tissue to ‘do something’, like move or withdraw your hand from something that’s hot, or think or remember.

These tiny currents can be measured with various types of EEG (Electro-encephalography) techniques, which involve placing electrodes on someone’s head.

In today’s medicine, EEG is used for a variety of brain function measurements, such as the diagnosis of epilepsy, and other conditions that affect localised brain function.

But whatever constitutes a person’s ‘mind’ is still poorly understood. While brain chemistry and neurology approach understanding of the human brain from the bottom up, ‘if we understand how individual cells work, we’ll eventually get to the overall picture’, psychology makes an approach from the top, asking questions ‘how does the whole thing work?’, treating the minor workings of the brain as a black box.

But how do you go from here to re-creating an entire artificial brain?

When a current passes through a synapse, its potential changes long-term through a process called synaptic plasticity, and this is how memory and learning works. Supposedly, if you were able to record the potentials for all synapses of a human brain, you’d be able to simulate that brain at that point in time.

Five hundred trillion requires a heck of a lot of computing power, but there is a project to this extent which claims to be able to build a functional computerised brain within ten years (see for example the Blue Brain Project).

A simulation model would involve the mapping of brain processes with EEG equipment more detailed than currently available, or alternately through serial sectioning, in which very thin slices of the brain are scanned (hint: this is not good for the patient). This material would then be encoded and uploaded and then the code might develop a mind and identity. This takes a bottom-up black box approach that’s similar to throwing together a bunch of springs, gears and a case and hoping that it will somehow assemble itself into an old-fashioned working watch.

So, are artificial brains far future science fiction?

Yeah, I thought so, but then I stumbled across some research dealing with Brain-Computer Interfaces (BCI). These are electrodes of some description that connect directly from the brain to an external device. Current thoughts are that these devices may help patients retain or recover memories or body functions for diseases like dementia or Parkinson’s. One such research project concentrated on the neuron firings of just 177 cells in the thalamus region of the brain, which decodes signals from the eye. The scientists were able to reconstruct images of what the cats saw.

Others are already working on processes that take this one step further: to stimulate the body to do something with this information, for example, in order to steer a prosthetic limb (check out the awesome Wikipedia summary about this research)

The step from this to encoding and developing an entire brain is no doubt great, but no longer seems to be impossible.

2 Comments:

  1. It’s a fascinating subject. I’ve seen those images reconstructed from the brain and it sure is impressive. So, too, are the stimulation of brain cells to control limbs. So I agree an artificial brain that does those motor things (for want of a better expression) is quite close. “Intelligence”, though? The ability to paint a picture, come up with a new theory for something? Perhaps. After all, new understandings are built on old learning, which a computer can do. The ability to play chess is one we’ve achieved. Lateral thinking? Creative, radically different ideas? It’ll be interesting.

  2. But since I’m developing a world in which artificial humans exist, and where people’s minds get transferred from one body to another,

    This may be of interest.

    Five hundred trillion requires a heck of a lot of computing power, but there is a project to this extent which claims to be able to build a functional computerised brain within ten years (see for example the Blue Brain Project).

    That would require a truly astonishing leap in our understanding of the behavior of neurons, I think the BBP are rather over-optimistic about the timeframe.

Comments are closed