By Dwayne Godwin and Jorge Cham
November 17, 2012 |
Dwayne Godwin is a neuroscientist at the Wake Forest University School of Medicine. His Twitter handle is @brainyacts. Jorge Cham draws the comic strip Piled Higher and Deeper at www.phdcomics.com.
10^15 calculations per second? If that were true, why would we need computers?
I believe this is true because of all the sensory data that we receive - just moving a step causes us to recalculate the distances and angles of everything around us ~
Sorry, but our brains do not mathematically calculate numerical distances, etc. - otherwise, no one would have to learn to count their fingers, much less how to perform mathematical computations. As I understand, we generally make value judgments based on sensory information (not computations based on digital data).
why would we need computers? - because humans pay attention to variety and get bored by repetition, and computers are great at repetitive mind-numbing computations/calculations like your taxes, adding up, checking your bank balances before and after transactions, remembering what you did last time you visited, quickly and accurately doing thousands of trivial tasks we prefer to forgetthe brain is a fantastic parallel processor for making intelligent assessment of complex situations in a fraction of a second - and the first entry point is often the amygdala or such - threat or opportunity, friend or foe - if that instant judgement looks bad - the rest of the opportunity for friendly calculation goes out the window while the organism flicks into survival mode.adding up your taxes is way lower priority than surviving crossing the road or getting your next meal, and if someone is shouting in your face, it's unlikely you'll be able to accurately add up lists of numbers at that moment.
The question was conditional - IF that were true - it's not. Our brains are neither digital calculators nor computers.
What's with the epilepsy, Alzheimer's and stroke numbers?I'm thinking the poster is a poorly implemented, yet somewhat clever fund-raising tool.
BTW, my point was that "10^15 calculations per second" metric most likely represents some set of actual events such as neuron 'firings' or some other analog processes that have nothing to do with numerical calculations...
I have the same question.. like I always tell my students... numbers without units attached are useless! 4 million [WHATs] with alzheimers
10^15 calculations.So why we take too many time to learn to know from 2*2=4 up to 9*9=81? They are less than 10^2 calculations
Almost all numbers without units are useless. A very important concept for everyone to know. However, the self-evident units are in this case, the # of American patients with the disease. Approx. 700,000 people have a stroke each year in the US. Therefore, the 4 E06 # is the number of folks who have had a stoke at one point. The numbers for Alzheimer's Disease and epilepsy represent the number of American patients living with the respective disease. Thanks for the excellent poster.
Our brains are "multi-taskers", while a computer can be conditionally programmed to only compute say, mathematical calculations. Your brain, for example, is also making your heart beat and your stomach churn. It is taking in the environment around you and comparing this to relevant past experience. Even something as simple as looking at a clear blue sky may bring to mind previous situations in which you learned (by deduction or instruction) the time of day via the position of the sun, future and current weather patterns, and even what you define a 'blue'. Not to mention, your brain is moving your eyes in concurrent patterns, etc....and you thought you were just looking at the sky adding 2+2. Clearly that is simply two calculations.
The reverse question is 'why a computer' can't recognize itself, even with a camera.There are for example, 10^9 neurons in the retina (rods/cones) - the optic nerve has ~10^6 nerve fibers - so to compress the information from the retina in order to funnel it down the optic nerve requires at least one calculation per retinal neuron (there are of course more) .. so that gives 10^9 - but the brain samples at a rate of 50Hz so that implies 50^10 without applying much thought ...
Our brain constantly makes such calculations to estimate the properties of the physical world. However, we don't have conscious access to this 'machinery'.This is a huge shame, but this is standard thinking in neuroscience. The evidence is overwhelming - neurons in the brain are constantly making computations based on sensory input. We are not aware of the vast majority of the computations the brain performs and we throw away much information. Take the remarkable ability of some idiot savants to achieve perfect recall of visual stimuli. The difference between idiot savants and the majority, is not that our brain computes the world differently, the difference is that idiot savants have conscious access to this information. However, there is a cost to having this information available all the time......, our conscious apparatus can only cope with a certain amount of information. Thus the purpose of all this per-conscious 'machinery' is to provide the most pertinent information to our higher brain areas.
The sensory signals are analog - not digital data. There are no numerical calculations being performed to determine anything except what mankind hath wrought. Numbers/mathematics and letters/language were developed by out own symbolic reasoning capabilities. Our sensory processes are not very different from those of animals...That computational methods can be used to model sensory processes is not evidence that the brain represents any sensory information numerically, allowing arithmetic calculations. Similarly, the equations of general relativity very accurately represent the effects produced by gravitation but they are not used by gravitation to calculate the effects it produces. Numerical models only represent the actual processes that produce observed results.Back in the earlier days of truly analog telephones, radios and televisions, very complex processes converted audio and visual information into EM signals and back again - all without any performing any numerical computations. The specifications of analog electronic processes were certainly determined analytically by humans using complex mathematical procedures, but the actual analog signal information was not represented numerically. Discrete signal transformations performed on the signal might have been quantifiable, but not in any way comparable to numeric calculations. I suggest that the brain uses similar methods to process sensory information, even if it is more convenient for neuroscientists to evaluate and represent those processes using numerical methods.Can you provide reference to any direct neurological evidence that analog sensory signal information is actually represented numerically anywhere within any animal's brain?
Some "interesting" comments...but none that really interest me... ha!I think the raw numbers themselves are a testament to how strong the foundation underlying the precept that our brains = our minds... which is of course, ourselves.The neuroscientific evidence for a lack of a "ghost in the machine" is overwhelming... yet some persist in invoking supernatural phenomena to explain our personas, a soul or whatever.But i digress... the primate, like other large brain mammals, is a remarkably evolved neural network... instantiating conscious thought, etc. But what is even more remarkable is the fact that our subconscious does all the heavy lifting... not to mention the neural network regulating our guts... cardiac and endocrine systems.The evolutionary architechture (common throughout the Animal Kingdom) is striking... spinal chord/medulla --> brainstem --> hind/mid brain structures --> to the space conserving specialization of our right - left hemispheres of the massive cerebral cortex... and the myriad processing centers and "mind-boggling" connections.That the brain is where evolution was taking primates is obvious... and yes, size does matter. And to get that size, one only need look at our embryology... and the depth of the human placental invasion of the uterus to sustain... brain metabolism. Humans pay a higher price "at the pump" (with pregnancy loss and complications)... but we get big honkin' neural processors! Heck... it's just the coolest organ on Planet Earth. Sure it's has it's quirks... and shortcomings... but hey, nothin's perfect.Now... where did i leave my car keys... hmmm???
When our mother's have to pay the price to make us smarter, shouldn't we pay them back with respect and understanding. What enables me to conceptionalize an idea of human mechanisms is appresheated because of my mother who didn't NEED this to be able to nurture me so I could. If beleiving in "the spirit in the sky" helps me (or her) to beleive the spirit or motivation in me, then this is the essence of self-esteem and positive "knowlege" in both. Anyone who realizes emotional and moral intellegence would know that knowing our differences and the reasons behind them is key to healthy survival.
Guys, it is 10^16 calculation per second instead of 10^15. A calculation of a neuron consists on taking data(signals/signal strengths) as inputs, sum up the input, and then apply a threshold to output the result. The threshold specifies whether the summation of the inputs was greater than a given value. so, there are at most 10^16 calculations per second happening in the brain.
At any rate, the reception of upstream signals depends on the chemical environment of each synapse - while neuronal firing may be a threshold trigger process, it does not represent a simple arithmetic calculation. Once a neuron fires, which of its connected neurons receives a signal depends on their synaptic conditions.IMO, these neuronal processes can be modeled mathematically, but they are not performing numerical calculations.The implication of the specification: "10,000,000,000,000,000 Calculations per second" is that we can each square a number, for example, 10^16 times a second. I'm not that good with math - I think that most of us require several seconds to multiply a couple of two digit numbers!
BTW, a specification of 1 MIPS (millions of instructions per second) for a computer processor is based on the average time required to execute a representative mix of various instructions (the actual time for different types and even the same instruction can vary), but that processor, generally speaking, CAN square a number about a million times per second.
@jtdwyer:If the brain doesn't calculate any mathematics, it is hard to imagine how it could perform so many tasks that require not only precise movements but also awareness of the current body position.@Raoul:The brain might actually know the solution for certain mathematical problems, but we as humans do not consist merely of a brain. I think that is why we experience "aha!" moments when learning. How could our brains work at all if they did not already contain working data-sets? Its a chicken and egg problem.
http://en.wikipedia.org/wiki/Analog_computerThis may not be the best Wikipedia entry, but it might help your imagination - no numerical calculations required during operation!Working data-sets? Your brain is not a digital computer! There are no data bytes or nibbles or even bits...
I would love to see the references for these calculations. It is troubling that no accounting is made for the mass of the tissue. If you assume a mean diameter of 5 microns/axon (a bit of a compromise), and calculate the volume of a cylinder with radius 0.0000025 meters (2.5 microns) and length 160934400 meters (100000 miles), you get a volume of 0.00316 cubic meters of axon. Sounds small, until you realize that 1 cubic meter of water weighs 1000 kilograms. If we assume a density close to the density of water, that yields a mass for axon volume alone of .00316 kg, or 6.96 lbs. The human brain weighs on the order of 3 lbs., including all of the neuronal cell bodies, dendrites, and glial cells. Thus, the 100,000 miles of axons becomes deeply suspect.
Took me much more than 30 seconds to see, read, process and comprehend!Got stuck for eons in the lower right quadrant of the graphic. Took a while to realize that those numbers were in a different category of statistics from all the rest (number of humans with those diseases, not properties of the brain).
According the the Alzhimer's Association 5.3 million people in the US have the disorder. "4 million" could be someone else's statistic, since the number depends on the diagnostic criteria.
I thank the authors generalized picture that gathered that difficult to collect. Everything is clear and simple.Recall that the level of intelligence depends more on the number of synapses that can grow like leaves on a tree.Moreover, neuronal subst. grizea, which many in the associative field can also be more than genetically incorporated.The picture perfectly reflects morphology. Difference functions adds dissimilarity people.Ask questions. Responsible, based on the innovative experience of 25 years.
neyroreabilitologist - being curious I wondered what field of study you're referencing, but a web search for "neyroreabilitology" found no related documents. A search for "neyroreabilitologist" suggested the term "neurobiologist" instead.I guess that English is not your first language, but I cannot understand on any level the statements:"The picture perfectly reflects morphology.""Difference functions adds dissimilarity people."Can you please explain? Is this some kind of alien science?
If you hear a sound, and are like most people (i.e., have normal auditory processing) you know where it came from -- but how? Well, you know because you do a lot of complex physics calculation -- and you do them outside conscious awareness, in much less than a second, using only your brain stem (this is done before the information reaches the cortex). There are actually a lot of thing the a human, or even a mouse or frog, can do automatically and very quickly, that have turned out to *VERY* hard to program a computer to do. It not that hard for computer to generate 2d picture from a 3d model, but to percieve 3d from a 2d image, not so much -- and program speech recognition has proved challenging and yield less the optimal results so far (though it has become common in use, such software has trouble with accents, background noise, and many other things that cause humans no problem) -- the list goes on and on.So why do we use electronic computers? Simple, there are a lot of abstract mathematical formulas and opperation that are not hardwired in our brains or that we would like to have done under our conscious planning and control, and the way we do arrithmetic abstractly with symbols is not nearly so fast (though the neural representation and processing of the symbols is certain far more complex than the culculation they represent).
Do not confuse the equations of physics that can be used to mathematically model physical processes with physical reality. The brain uses analog circuits to perform its processes. Do you suppose that the Moon uses Newton's law of universal gravitation or Einstein's theory general relativity to determine its orbit around the Earth? The equations of successful physical theories can be used to describe natural phenomena and even predict events, but they are merely mathematical models devised by humans to represent actual physical processes.The brain uses analog circuits to process information. If we had to solve so many equations to determine from sensory information that the fire is hot - there'd be a lot burnt fingers! Children can learn that long before they learn to count or add. Most animals can determine that fire is hot even though they can't perform the simplest calculations...Have recent generations lost touch with reality?
BTW, most people don't like understand that computers don't simply perform mathematical functions to solve equations, like 52 - 22 = 30. They actually emulate mathematical operations by employing special circuits that perform logical bit manipulation functions to accomplish numerical calculations. Please see:http://en.wikipedia.org/wiki/Adder_(electronics)As I understand, the brain does not represent sensory information as encoded numerical data, but as analog signals that trigger cascading processes based on dynamically varying amplitude thresholds, for example.
Of course the brain doesn't act like a digital computer (or like an analog computer humans would build, for that mater). I've had a computer engineering class on digital circuit design years ago -- I couldn't tell you the design of a specific processor but I know basically how they work in principle. No, the brain doesn't use abstract numerical data, and it is going to more of an analogue operation to do what it does. However, it does compute a lot of things, and often gets surprising accurate result very quickly.Of course, sound waves don't use human designed physical models to know what to do, when to reach each ear, how much of a sound shadow to make on the other side of the head, or how to resinate in the outer-ear (i.e., how to filter various frequencies) -- they just do what they do -- but none of that's relevant. After the sound reaches the ear, has been encoded (which is interesting in itself) the brain uses data based on time lags, sound shadow, and differential resinant filtering (which varies with the angle the sound inters the ear and by frequency) to figure out where it came from. No, the brainsem doesn't contain abstract functions written in algebra and calculus; and doesn't use a binary, digital, or other abstract number system -- but it does process the data, and in effect perform calculation, even if they are "analog" in nature.
BTW, the spinal cord produce the reflexive recoil from pain -- the brain brain finds out later.
Well put - I essentially agree. Recall that the original issue was that the chart claims that the brain performs 10^16 calculations per second. I don't think that the methods used by the brain to process information can possibly be quantified as any specific number of discrete 'calculations' per second! I suggest that the chart's statement is not only baseless but meaningless.
That may be -- I do suspect that terms were expressed in a dumbed-down and sensationalized way for popular media. I basically just went with it and read it loosely since I'm used to that. I supposed misinterpreted the first response I replied to as being based on a simplistic / overly-literal argument or misunderstanding of what was meant or implied, rather than the kind of criticism you now explain in more detail.
Subscribe to the Psychology and Brain Science magazine from SA.
Special collections on popular topics drawing from 167+ years of SA archives.
Online Access + Tablet Access + 20 year archive.
Deadline: Dec 11 2013
Reward: $52,000 USD
Platform technologies – tools, techniques, and instruments that enable entirely novel approaches for scientific investigation across a b
YES! Send me a free issue of Scientific American with no obligation to continue the subscription. If I like it, I will be billed for the one-year subscription.
Your Brain by the Numbers: Scientific American Mind
Give a 1 year subscription as low as $9.99
Subscribe Now >>
Yes, please link my existing account with for quick, secure access.
No, I would like to create a new account with my profile information.
Are you sure?
It has been identified that the institution you are trying to access this article from has institutional site license access to Scientific American on nature.com. To access this article in its entirety through site license access, click below.