Einstein once famously lamented: Now that the mathematicians have got hold of my theory, I myself no longer understand it. This seems to have been a genuine sentiment and not just the usual big-shot faux humility (Feynman, for example, was also famous for eschewing obfuscation, apparently forgetting his Nobel was based on infinite dimensional path integrals in Banach space). Indeed, I once read Albert's slender Relativity on spring break, the Boca motel hotties giving me odd looks and a cold shoulder as usual. I suppose the mathematician's analog would be Wheeler's 200 pound Gravitation, which is a mess of Christoffel symbols and diffeomorphism. Not really the sort of thing you flip through at poolside. Der Herr Gott doesn't play dice, and He ain't exactly keen on covariant derivatives either.
Those of us in the neurogonical arts take a bemused view of these intellectual cat fights. For when the life of the mind is shot though the harsh lens of bastard neuroscience, the whole business reduces to pattern generation. Inside your noggin there are no differential equations. No integrals, no derivatives. No Banach spaces or Sigma algebras. There is only an ongoing pattern of neural firing, a pattern tempered by arriving stimuli. Pop, pop. The quadratic equation is a pattern. E = mc^2 is a pattern. Quantum electrodynamics is a pattern. Witnessing an apple falling from a tree is a pattern. Positing the same force holds the moon in place as the apple is also a pattern, one generated in Newton's noggin but not in others (apparently this made Newton "smart" and others not-so). The version of Relativity Albert understood as well as the one he did not? Different patterns. I think you see the pattern.
What does any of this have to do with Javascript? Read on.
Another famous result is were we to take out your neurons and lay them end-to-end, you would die. A more useful trick would be arranging your neurons in a square matrix and in a manner leaving you/them still functioning. How large would the resulting matrix be?
You often hear claims of how many neurons there are in the brain nonchalantly tossed around as if sitting down and counting them is some kind of neurowonk rite of passage. Hardcore neuroanatomists (yes, these exist) get livid at the notion. It's not that the concept is difficult (there are thingys; count them) but rather that technical difficulties screw the pooch. For the brain is twisty and neurons are small and many. Estimates of neuron counts even in some rather well-respected studies may be off by several orders of magnitude, as any neuroanatomist will admit once you get him/her sufficiently liquored up.
Fortunately for today's fun, many of the brain's neurons are, well, irrelevant. Einstein may have had a buff Botzinger complex, but it had little to do with him discovering Relativity save to keep him breathing. No, we have reason to focus on the cerebral cortex, that splendid organ of imagination and creation and cognition which not only grants man dominion over the Earth (although often not over himself) but also simplifies the problem of counting neurons (although not as much as you might think; Braitenberg begins his classic text on quantitative neuroanatomy with half a chapter defending just what does and doesn't qualify as "cortex").
Still, focusing on cortex allows us to work a ballpark figure of 20 billion. Twenty billion neurons in the human cerebral cortex, a number offensive enough to cause hardcore neuroanatomist to puff out their cheeks and rock their heads back and forth as they formulate an objection but not so offensive that they will take the time to come and burn down my house.
So, then. If we want to sit back and watch the cortex do its thing like so much reality television, we can arrange these 20 billion neurons into a 2D matrix roughly 141,000 neurons on a side. To give you a feel for the magnitude of such a task, it's somewhat beyond current familiar technology, although not incorrigibly so -- approximately 20 gigapixels as our digital camera friends would say. The Internet informs me Canon has recently created a 250-megapixel camera (resolution: 19580×12600). So we would have to buy about ten of them to build a simulated cerebral cortex. If Canon was selling them. Which they aren't (the design is just a prototype). I suppose it's for the best. Explaining to your PI or SO why you put ten $5,000 SLRs on the Mastercard then disassembled them at the behest of some Internet crazy person is the sort of judgement lapse that goes down on your permanent record.
Fortunately, a simulation is much more easily achieved in software. Such a tradeoff is not uncommon -- Jenga is hard; Tetris less-so, Apollo is hard; Kerbal less-so, Dating is hard; WoW less-so. Similarly, simulating the human condition in hardware is hard, but less-so in Javascript. As such, I give you JCortex. An artist's interpretation of what you would see were you to take out your cortex and carefully smoosh it flat onto the surface of your monitor and watch its neurons popping off action potentials as you think your nefarious thoughts. A kind of animated rosary for you to ponder as you contemplate your existence.
Still, even here I have taken liberties. A 2003 study by Peter Lennie (Curr. Biology 13:493–497) concluded at most ~1% of cortical neurons can be active at one time. As far as ease of simulation goes that's better than 100%, but it still demands tracking 200,000,000 neurons which is not really feasible given the memory restrictions of contemporary web browsers, at least not without making your CPU fan get testy (if you are reading this centuries in the future, that claim may no longer be true. What can I say? I work with what I have, because I must). At the end of the day it hardly matters; the sought-after effect is achieved whether we simulate 200 neurons or 200 million. So I simulated 20.
As you can see below, the app comes with a start button (to start) and a stop button (to stop) as well as a sync button, which simulates the arrival of a stimulus acting to temporarily synchronize neural activity, as a stimulus usually does. Reloading the page will get you a different random distribution of neurons, some of which you may find more aesthetically pleasing than others.
So stare long into the neural fire and reflect. In these swirling patterns is everything humans ever created or wanted to create. Ideas both gross and sublime. Relativity. Gravitation. Banach space. The germ theory of disease. Operation Barbarossa. The Shelby Cobra. The Edsel. The Pieta. The strapless evening gown.
All you are, or have been. All you'll ever be.
Neural patterns all.
Those of us in the neurogonical arts take a bemused view of these intellectual cat fights. For when the life of the mind is shot though the harsh lens of bastard neuroscience, the whole business reduces to pattern generation. Inside your noggin there are no differential equations. No integrals, no derivatives. No Banach spaces or Sigma algebras. There is only an ongoing pattern of neural firing, a pattern tempered by arriving stimuli. Pop, pop. The quadratic equation is a pattern. E = mc^2 is a pattern. Quantum electrodynamics is a pattern. Witnessing an apple falling from a tree is a pattern. Positing the same force holds the moon in place as the apple is also a pattern, one generated in Newton's noggin but not in others (apparently this made Newton "smart" and others not-so). The version of Relativity Albert understood as well as the one he did not? Different patterns. I think you see the pattern.
What does any of this have to do with Javascript? Read on.
Another famous result is were we to take out your neurons and lay them end-to-end, you would die. A more useful trick would be arranging your neurons in a square matrix and in a manner leaving you/them still functioning. How large would the resulting matrix be?
You often hear claims of how many neurons there are in the brain nonchalantly tossed around as if sitting down and counting them is some kind of neurowonk rite of passage. Hardcore neuroanatomists (yes, these exist) get livid at the notion. It's not that the concept is difficult (there are thingys; count them) but rather that technical difficulties screw the pooch. For the brain is twisty and neurons are small and many. Estimates of neuron counts even in some rather well-respected studies may be off by several orders of magnitude, as any neuroanatomist will admit once you get him/her sufficiently liquored up.
Fortunately for today's fun, many of the brain's neurons are, well, irrelevant. Einstein may have had a buff Botzinger complex, but it had little to do with him discovering Relativity save to keep him breathing. No, we have reason to focus on the cerebral cortex, that splendid organ of imagination and creation and cognition which not only grants man dominion over the Earth (although often not over himself) but also simplifies the problem of counting neurons (although not as much as you might think; Braitenberg begins his classic text on quantitative neuroanatomy with half a chapter defending just what does and doesn't qualify as "cortex").
Still, focusing on cortex allows us to work a ballpark figure of 20 billion. Twenty billion neurons in the human cerebral cortex, a number offensive enough to cause hardcore neuroanatomist to puff out their cheeks and rock their heads back and forth as they formulate an objection but not so offensive that they will take the time to come and burn down my house.
So, then. If we want to sit back and watch the cortex do its thing like so much reality television, we can arrange these 20 billion neurons into a 2D matrix roughly 141,000 neurons on a side. To give you a feel for the magnitude of such a task, it's somewhat beyond current familiar technology, although not incorrigibly so -- approximately 20 gigapixels as our digital camera friends would say. The Internet informs me Canon has recently created a 250-megapixel camera (resolution: 19580×12600). So we would have to buy about ten of them to build a simulated cerebral cortex. If Canon was selling them. Which they aren't (the design is just a prototype). I suppose it's for the best. Explaining to your PI or SO why you put ten $5,000 SLRs on the Mastercard then disassembled them at the behest of some Internet crazy person is the sort of judgement lapse that goes down on your permanent record.
Fortunately, a simulation is much more easily achieved in software. Such a tradeoff is not uncommon -- Jenga is hard; Tetris less-so, Apollo is hard; Kerbal less-so, Dating is hard; WoW less-so. Similarly, simulating the human condition in hardware is hard, but less-so in Javascript. As such, I give you JCortex. An artist's interpretation of what you would see were you to take out your cortex and carefully smoosh it flat onto the surface of your monitor and watch its neurons popping off action potentials as you think your nefarious thoughts. A kind of animated rosary for you to ponder as you contemplate your existence.
Still, even here I have taken liberties. A 2003 study by Peter Lennie (Curr. Biology 13:493–497) concluded at most ~1% of cortical neurons can be active at one time. As far as ease of simulation goes that's better than 100%, but it still demands tracking 200,000,000 neurons which is not really feasible given the memory restrictions of contemporary web browsers, at least not without making your CPU fan get testy (if you are reading this centuries in the future, that claim may no longer be true. What can I say? I work with what I have, because I must). At the end of the day it hardly matters; the sought-after effect is achieved whether we simulate 200 neurons or 200 million. So I simulated 20.
As you can see below, the app comes with a start button (to start) and a stop button (to stop) as well as a sync button, which simulates the arrival of a stimulus acting to temporarily synchronize neural activity, as a stimulus usually does. Reloading the page will get you a different random distribution of neurons, some of which you may find more aesthetically pleasing than others.
So stare long into the neural fire and reflect. In these swirling patterns is everything humans ever created or wanted to create. Ideas both gross and sublime. Relativity. Gravitation. Banach space. The germ theory of disease. Operation Barbarossa. The Shelby Cobra. The Edsel. The Pieta. The strapless evening gown.
All you are, or have been. All you'll ever be.
Neural patterns all.
Action potentials in a volume of neural tissue
No comments:
Post a Comment