I have here in front of me a picture of a human nerve cell growing on a microchip, and this picture has always intrigued me.
The microchips we make now with MOS technology require very little voltage (1.3V is the limit right now) and next to no current to operate. What can a nerve cell supply?
In this sense, consider…
While a fetus is in very early development we simply “mash” some chips (or gently place, if you prefer) into the developing brain (sterilized of course). The brain will develop around the chip…
Now, the question, for those of you not keen on the obvious.
Can the brain cells utilize this chip?
[sub]Due to the incredibly large number of brain cells I don’t know that the chip will make a noticible difference in intelligence, I am merely wondering if it is plausible.[/sub]
We’re a long way off from being able to improve on the computational density of the brain- if you tried to replace brain cells with chips now, you’d probably end up losing capability, assuming you limited yourself to the size of your head.
I suppose if you could make the chip emulate the functionality of a neuron, it would appear to be a neuron to other neurons, and be thus utilitized. Of course, the chip needed to fully emulate a neuron probably wouldn’t fit in a person’s head with current technology
If the chip didn’t act like a neuron when interacting with other neurons (via electrical activity), I’m not sure what could happen, although at best it would probably be “not much”.
Of course, I’m glossing over the issue that we don’t fully understand just how neurons interact with each other, so it couldn’t really be duplicated with a chip yet, even if we had the technology.
It is my understanding that nerve cells are a lot like transistors. They have a threshhold voltage that must be recieved before they fire, and of course they have a fan-out of sorts just like transistors as well.
A neuron can be modeled with a simple threshold and fanout, but that model doesn’t really capture the full scope of what’s going on. The biggest difference is that
a neuron has some memory of recent activity, whereas a transistor (or other ideal switch) is memoryless (a fundamental difference)
a neuron is analog, while a transistor is digital (at least when used as a switch)
Hi Opal! A neuron interfaces with thousands or tens of thousands of other neurons, while a transistor (or a logic gate) typically drives 10 or so (although some logic families have a theoretical fanout in the thousands)
A typical switching transistor is thousands of times faster than a neuron
Resting potential of an average neuron is about -70 mV. The peak of the action potential is at about 50 mV–a total swing of around 0.12 V.
My guess would be, simply: no. The chip would be highly unlikely to respond to potential changes in neurons, and as the chip you’re postulating has no power source, there’s no way for the chip to transmit a signal to the neurons.
(Aside: at least, not a signal that makes any sense. I have a small bit of surgical steel in the midst of my brain, and it’s credited with causing the seizures I had several years ago. Eddy currents and minor currents can appear in the metal of a chip due to fluctuations in the local electric fields, but they wouldn’t be useful in an information capacity.)
To call this an oversimplification would be a major understatement. Your average cortical neuron has input from around 10000 other neurons, both near it and from far away. Some of these inputs are excitatory; some are inhibitory (that is, some of the inputs decrease the cell potential.) In the simplest system, several excitatory inputs will cause an action potential, but it should be noted that the AP is only a rapid change in the local potential.
I tend to think of an average neuron as equivalent to something around ten thousand very slow transistors, because each ion channel and cascading signalling pathway is a single switch–a single chemical transistor, if you will. There is also the added bit of long-term potentiation or long-term depression (the cellular memory that Arjuna mentioned), which depends on certain patterns of firing (and could be vaguely compared to computer RAM).
We’re at least decades away from merging silicon chips with neurons for anything but limited functions. It would be easier, IMO, to build an entire brain from silicon chips, though it would, as Arjuna noted, be significantly larger than your brain.
There have already been several good posts addressing your question, but I just wanted to point out one other thing (ignoring all of the other factors, such as power supply, signal strength, etc.). That is that a microchip has a specific architecture. To “utilize” that architecture, you can’t simply feed signals into it willy-nilly. The signals must arive in a coherent sequence with specific timing. Since utilization ususally implies results, we must also do something useful with the outputs of the chip. Neurons and chips are incompatible, but on a higher level brains and computers have incompatible architectures.
You’d have better luck taking a bunch of chips, resistors, capacitors, batteries, and other electronic components, soldering them together in a totally random manner and expecting the result to do something meaningful…
Neurons aren’t electical, they are electrochemical. To transmit signals to other neurons they release neurotransmitters. To even have an effect on a microchip, the chip would need receptors for those neurotransmitters. This is possible I guess, but I have no idea how you could make an ion channel interact with an electrical circuit.
Well, the implied idea was that the chips would be placed in the brain during development, so that the cells would grow into it as part of the natural environment. It wouldn’t have “willy-nilly” inputs. Then again, you make it seem like the brain’s architechture isn’t somewhat random in itself. I seem to doubt that the brains innerconnections are somehow deliberate on a part of our DNA. In that sense, our own brain is “willy nilly,” eh?
At any rate, the voltage level is interesting. I’ll have to remember that. That’s a tiny amount of voltage! I had done some researching on my own while this thread was going on, but I never found any mention of resistance or current quantity. I know this changes with age, but I would think that info would be out there somewhere…
USCErrr…electrochemical yes… but electrical signals travel over large parts of the neurons. To my knowledge, the chemical aspect of nerve communication is due to their biological independence. Within a particular neuron, though, it is all electrical (well, not ALL of course, but not chemical in the way it is implied there).
In a neuron (and in the body in general), electrical signals are carried by charged ions, through liquid. In a chip, electrical signals are carried by electrons through wires or semiconductors. An interface between the two would be a little awkward (especially neuron-sized), but not impossible. The use of ions makes lots of complicated interactions possible, that in theory could be simulated with a chip internally, if we knew exactly what was going in a neuron.
Not much is known, but the brain’s interconnections are neither predetermined by DNA nor “willy nilly”. AFAIK, the connections are formed, strengthed, weakened, etc. in the process of brain development (especially during the first year of life), and as we learn/forget new things.
It would be relatively easy to handle the voltage levels, assuming there is a way to power the artificial neuron (I supposed you could throw in a 9-volt battery :)).
Relative to the expected inputs, the neuron-silicon interface would tend to be willy-nilly. Or said another way, it isn’t likely that the interface would be automagically compatible. Both the brain and a silicon-based processor have interface protocols. You can’t mix neurons and chips into a brain soup and expect these protocols to align.
It isn’t random. It may not be predetermined in all areas, but the architecture is fairly well understood, at a high level. Just the brain, mind you, I’m not speaking of the mind…
Many of the connections ARE hard-wired by our DNA. Consider that the frontal, occipital, parietal, and temporal lobes, cerebellum, and brainstem all tend to specialize in certain functions, fairly universally across the human population. This is at a high level, architecturally, but certainly there are some variations when you dig deeper into the mapping of individual neurons. As Arjuna points out, something in the learning mechanism “installs” new connections, reinforces old connections, and “disconnects” obsolete connections. These connections may not be well understood, but they are definitely not random…
Neurotransmission is mainly electrochemical, but it also has electrical elements. If this weren’t the case, it would be very hard for us to study the elctrical properties of the neurons–which we do here in my lab using electrodes which are not sensitive to neurotransmitters. (One note: those electrodes are, in fact, stuck into the cell, which is a little bit different).
Many neurons make surface-to-surface channel contacts which allow electrical impulses to travel directly from one to another without the transduction to chemical energy. I’ll agree this isnt’ the major method of neurotransmission, but inserting an electrode and a certain amount of current would eventually give rise to an action potential, no matter the absence of neurotransmitters. In the opposite direction, a chip/electrode with enough sensitivity to voltage would be able to sense the electrical impulses, if not the chemical impulses.
As to other stuff: JoeyBlades is pretty much correct. I also like the way he’s said it: a chip inserted wouldn’t automagically interface with the brain. It would have to be painstakingly designed, and the research to do this will take decades, if not centuries.
The brain isn’t put together willy-nilly. During learning, many many changes go on. And no, unfortunately, most of this is not very well understood. That’ll also be a matter of decades, if not centuries.
Then again, you make it seem like the brain’s architechture isn’t somewhat random in itself. I seem to doubt that the brains innerconnections are somehow deliberate on a part of our DNA. In that sense, our own brain is “willy nilly,” eh?