Scientists have long advanced the claim that the human brain is the storage place for memories and the source of human thinking. But such claims are speech customs of scientists rather than things they have proven. There are numerous reasons for doubting such claims. One big reason is that the proteins in synapses have an average lifetime of only a few weeks, which is only a thousandth of the length of time (50 years or more) that humans can store memories. Another reason is that neurons and synapses are way too noisy (and synapses too unreliable signal transmitters) to explain very accurate human memory recall, such as when a Hamlet actor flawlessly recites 1476 lines. Another general reason can be stated as follows: the human brain is too slow to account for very fast thinking and very fast memory retrieval.
Consider the question of memory retrieval. Given a prompt such as a person's name or a very short description of a person, topic or event, humans can accurately retrieve detailed information about such a topic in one or two seconds. We see this ability constantly displayed on the long-running television series Jeopardy. On that show, contestants will be given a short prompt such as “This opera by Rossini had a disastrous premier,” and within a second after hearing that, a contestant may click a buzzer and then a second later give an answer mentioning The Barber of Seville. Similarly, you can play with a well-educated person a game you can call “Who Was I?” You just pick random names of actual people from the arts or history, and require the person to identify the person within about two seconds. Very frequently a person will succeed. We can imagine a session of such a game, occurring in only ten seconds:
John: Marconi.
Mary: Invented the radio.
John: Magellan.
Mary: First to sail around the globe.
John: Peter Falk.
Mary: A TV actor.
We can also imagine a visual version of this game, in which you identify random pictures of any of 1000 famous people. The answers would often be just as quick.
The question is: how could a brain possibly achieve retrieval and recognition so quickly? Let us suppose that the information about some person is stored in some particular group of neurons somewhere in the brain. Finding that exact tiny storage location would be like finding a needle in a haystack, or like finding just the right index card in a swimming pool full of index cards. It would also be like opening the door of some vast library with a million volumes and instantly finding the exact volume you were looking for.
There are certain design features that a system can have that will allow for very rapid retrieval of information. One of these features is an indexing system. An indexing system requires a position notation system, in which the exact position of some piece of information can be recorded. An ordinary textbook has both of these things. The position notation system is the page numbering system. The indexing system is the index at the back of the book. But the brain has neither of these features. There is nothing in the brain like a position notation system by which the exact position of some tiny group of neurons can be identified. The brain has no neuron numbers, and a brain has no coordinate system similar to street names in a city or Cartesian coordinates in a grid. Lacking any such position notation system, the brain has no indexing system (something that requires a position notation system).
So how is it that humans are able to recall things instantly? It seems that the brain has nothing like the speed features that would make such a thing possible. You can't get around such a difficulty by claiming that each memory is stored everywhere in the brain. There would be two versions of such an idea. The first would be that each memory is entirely stored in every little spot of the brain. That makes no more sense than the idea of a library in which each page contains the information in every page of every book. The second version of the idea would be that each memory is broken up and scattered across the brain. But such an idea actually worsens the problem of explaining memory retrieval, as it would only be harder to retrieve a memory if it is scattered all over your brain rather than in a single little spot of your brain.
We also cannot get around this navigation problem by imagining that when you are asked a question, your brain scans all of its stored information. That doesn't correspond to what happens in our minds. For example, if someone asks me, "Who was Teddy Roosevelt," my mind goes instantly to my memories of Teddy Roosevelt, and I don't experience little flashes of knowledge about countless other people, as if my brain were scanning all of its memories.
When we consider the issue of decoding encoded information, we have an additional strong reason for thinking that the brain is way too slow to account for instantaneous recall of learned information. In order for knowledge to be stored in a brain, it would have to be encoded or translated into some type of neural state. Then, when the memory is recalled, this information would have to be decoded: it would have to be translated from some stored neural state into a thought held in the mind. This requirement is the most gigantic difficulty for any claim that brains store memories. Although they typically maintain that memories are encoded and decoded in the brain, no neuroscientist has ever specified a detailed theory of how such encoding and decoding could work. Besides the huge difficulty that such a system of encoding and decoding would require a kind of "miracle of design" we would never expect for a brain to ever have naturally acquired (something a million times more complicated than the genetic code), there is the difficulty that the decoding would take quite a bit of time, a length of time greater than the time it takes to recall something.
So suppose I have some memory of who George Patton was, stored in my brain as some kind of synapse or neural states, after that information had somehow been translated into synapse or neural states using some encoding scheme. Then when someone asks, "Who was George Patton?" I would have to not only find this stored memory in my brain (like finding a needle in a haystack), but also translate these synapse or neural states back into an idea, so I could instantly answer, "The general in charge of the Third Army in World War II." The time required for the decoding of the stored information would be an additional reason why instantaneous recall could never be happening if you were reading information stored in your brain. The decoding of neurally stored memories would presumably require protein synthesis, but the synthesis of proteins requires minutes of time.
We also cannot get around this navigation problem by imagining that when you are asked a question, your brain scans all of its stored information. That doesn't correspond to what happens in our minds. For example, if someone asks me, "Who was Teddy Roosevelt," my mind goes instantly to my memories of Teddy Roosevelt, and I don't experience little flashes of knowledge about countless other people, as if my brain were scanning all of its memories.
When we consider the issue of decoding encoded information, we have an additional strong reason for thinking that the brain is way too slow to account for instantaneous recall of learned information. In order for knowledge to be stored in a brain, it would have to be encoded or translated into some type of neural state. Then, when the memory is recalled, this information would have to be decoded: it would have to be translated from some stored neural state into a thought held in the mind. This requirement is the most gigantic difficulty for any claim that brains store memories. Although they typically maintain that memories are encoded and decoded in the brain, no neuroscientist has ever specified a detailed theory of how such encoding and decoding could work. Besides the huge difficulty that such a system of encoding and decoding would require a kind of "miracle of design" we would never expect for a brain to ever have naturally acquired (something a million times more complicated than the genetic code), there is the difficulty that the decoding would take quite a bit of time, a length of time greater than the time it takes to recall something.
So suppose I have some memory of who George Patton was, stored in my brain as some kind of synapse or neural states, after that information had somehow been translated into synapse or neural states using some encoding scheme. Then when someone asks, "Who was George Patton?" I would have to not only find this stored memory in my brain (like finding a needle in a haystack), but also translate these synapse or neural states back into an idea, so I could instantly answer, "The general in charge of the Third Army in World War II." The time required for the decoding of the stored information would be an additional reason why instantaneous recall could never be happening if you were reading information stored in your brain. The decoding of neurally stored memories would presumably require protein synthesis, but the synthesis of proteins requires minutes of time.
There is another reason for doubting that the brain is fast enough to account for human mental activity. The reason is that the transmission of signals in a brain is way, way too slow to account for the very rapid speed of human thought and human memory retrieval.
Information travels about in a modern computer at a speed thousands of time faster than nerve signals travel in the human brain. If you type in "speed of brain signals" into the Google search engine, you will see in large letters the number 286 miles per hour, which is a speed of 128 meters per second. This is one of many examples of dubious information which sometimes pops up in a large font at the top of the Google search results. The particular number in question is an estimate made by an anonymous person who quotes no sources, and one who merely claims that brain signals "can" travel at such a speed, not that such a speed is the average speed of brain signals. There is a huge difference between the average speed at which some distance will be traveled and the maximum speed that part of that distance can be traveled (for example, while you may briefly drive at 40 miles per hour while traveling through Los Angeles, your average speed will be much, much less because of traffic lights).
A more common figure you will often see quoted is that nerve signals can travel in the human brain at a rate of about 100 meters per second. But that is the maximum speed at which such a nerve signal can travel, when a nerve signal is traveling across what is called a myelinated axon. Below we see a diagram of a neuron. The axons are the tube-like parts in the diagram below. The depicted axon is a myelinated axon (the faster type); but a large fraction of axons are unmyelinated (the slower type).
The less sophisticated diagram below makes it clear that axons make up only part of the length that brain signals must travel.
Below is a depiction of these components by Google's Gemini AI:
There are two types of axons: myelinated axons and non-myelinated axons (myelinated axons having a sheath-like covering shown in blue in the diagram above). According to this article, non-myelinated axons transmit nerve signals at a slower speed of only .5-2 meters per second (roughly one meter per second). Near the end of this article is a table of measured speed of nerve signals traveling across axons in different animals; and in that table we see a variety of speeds varying between .3 meters per second (only about a foot per second) and about 100 meters per second.
But from the mere fact that nerve signals can travel across myelinated axons at a maximum speed of about 100 meters per second, we are not at all entitled to conclude that nerve signals typically travel from one region of the brain to another at 100 meters per second. For one thing, only about half of the axons in the human cortex are myelinated, and the transmission speed of the unmyelinated axons is only about a meter per second. Moreover, nerve signals must also travel across dendrites and synapses, which we can see in the diagrams above. It turns out that nerve signal transmission is much slower across dendrites and synapses than across axons. To give an analogy, the axons are like a road on which you can travel fast, and the dendrites and synapses are like traffic lights or stop signs that slow down your speed.
According to neuroscientist Nikolaos C Aggelopoulos, there is an estimate of 0.5 meters per second for the speed of nerve transmission across dendrites (see here for a similar estimate). That is a speed 200 times slower than the nerve transmission speed commonly quoted for myelinated axons. Such a speed bump seems more important when we consider a quote by UCLA neurophysicist Mayank Mehta: "Dendrites make up more than 90 percent of neural tissue." Given such a percentage, and such a conduction speed across dendrites, it would seem that the average transmission speed of a brain must be only a small fraction of the 100 meter-per-second transmission in axons.
A scientific paper from 2025 documents precise measurements of the speed of signal transmission across both axons and dendrites, in both humans and rats. The paper is entitled "Accelerated signal propagation speed in human neocortical dendrites" and can be read here.
The paper gives us a speed of nerve signals (which are called action potentials) in the axons which are the fastest parts of a brain. Using the term AP to mean an action potential or nerve signal, the paper states, "We
found no significant difference in the propagation speed of APs in the axons of rats and humans (rat:
n=8, 0.848±0.291 m/s vs. human: n=9, 0.851±0.387 m/s, two-sample t-test: p=0.282, Figure 2F)." In that quote the paper gives an axon transmission speed of about .8 meter per second, which is more than 100 times slower than the "100 meters per second" figure commonly cited in popular literature as the speed of brain signals.
For the speed of signal transmission across dendrites (which make up 90% or more of brain tissue), the paper gives us two numbers, one for what it calls "forward propagating sEPSP speed" and another it calls "back propagating AP speed." We are told that these speeds were measured:
- "The AP propagation speed was calculated for each cell from the time difference between the somatic and dendritic APs divided by the distance between the two points. We found that the propagation speed was, on average, ~1.47 fold faster in human (rat: 0.233±0.095 m/s vs. human: 0.344±0.139 m/s, Mann-Whitney test: p=6.369 × 10–6, Figure 2F, Figure 2—figure supplement 1B)". This is a speed of about one third of a meter per second, roughly ten centimeters per second, the same as about one foot per second. The "m/s" in the quote above means meter per second.
- " We found that sEPSP propagation speed was, on average, ~1.26 fold faster in human (rat: 0.074±0.018 m/s vs. human: 0.093±0.025 m/s, two-sample t-test: p=0.004; Figure 2D, Figure 2—figure supplement 1D)." This is a speed of about one tenth of a meter per second, roughly ten centimeters per second, or about four inches per second. The "m/s" in the quote above means meter per second.
In Table 2 of the paper we have five different rows marked with names such as Human1, Human2, Human3, Human4 and Human5. The last column in the table is marked "Velocity." All of the velocities listed are less than a tenth of a meter per second. The average of the five velocities is 0.085 meter second.
Dendrites, it would appear, are sluggish bottlenecks or speed bumps (by which I mean a physical feature that slows something down). And since it is often claimed that dendrites make up 90% or more of brain tissue, what does this tell us about whether brains are fast enough to account for instant human recall? It tells us that brains are way too slow to explain humans who can think at blazing fast speeds, and give the right answers instantly when asked rarely asked questions.
Besides the slow speed of dendrites, a very important additional speed bump or bottleneck is that of synaptic delay. Synaptic delay is the fact that every time a nerve signal passes the synaptic gap of a chemical synapse, there is a delay of about .5 millisecond. In a brain with an estimated 100 trillion synapses, synaptic delay would be an enormous slowing factor. This is because nerve signals would have to travel across very many synapses, resulting in a cumulative delay that might add up to quite a few seconds or many seconds.
The diagram below shows fast, slow and no-so-fast parts of the brain. The snail symbols indicate slow parts, parts that would slow down nerve signals. The thin rabbit represents a relatively fast part. The fat rabbit represents a not-so-fast part. Here "slow" and "fast" refers to the speed at which signals could transmit through such parts, which do not themselves move.
A brain is something packed with a gazillion speed bumps or bottlenecks: the speed bumps or bottlenecks of dendrites, and the speed bumps or bottlenecks of synapses, with their delays at every synaptic junction. The brain therefore screams to us in a loud voice: "I'm way too slow to explain your instant recall."
Postscript:
There are two factors we can consider that will help clarify why a dendrite signal transmission speed of only about a third or a tenth of a meter is way too slow to account for instant human recall The first factor is the total area of the human cortex. The brain tissue in the human cortex is highly folded. This means that the total area of the human cortex is surprising large, much larger than the surface area needed to make a hat.
If you use a Google search phrase such as "total area of the human cortex," you will be told that such a surface area is between 1.5 square feet and 2.5 square feet, "roughly the size of a standard pizza." A standard pizza is about 14 inches in diameter, about the distance from a man's elbow to a ring on his finger.
When you type in "compute the average distance between two points in a circular area" in a modern browser such as Chrome, you get an AI overview answer telling you that for a circular area with a radius of r, the average distance between two points is about .9r. Using the formula above, we can (ignoring the complication of tortuosity) crudely estimate that the average distance between two random points in the cortex is about .9 times 7 inches, which is about 6 inches.
But such a number would be a significant underestimation of the average distance that a brain signal would have to travel to go from one point to another in the cortex. The reason why it would be an underestimation is a reason called tortuosity. The word "tortuosity" refers to the fact that neural pathways are not straight lines but twisty, wiggly, squiggly lines. And it takes longer for signals to travel along such twisting lines than it does for a signal to travel along a straight line.
The Google Gemini diagram below illustrates quite well the concept of the tortuosity of brain pathways.
How much does this tortuosity factor affect the length of the pathways that brain signals must travel? You can get an estimate by typing "numerical estimate of the tortuosity of brain pathways" into Google Chrome. This produces the answer that this tortuosity is estimated to be about 1.6. The scientific paper here ("Extracellular space structure revealed by diffusion analysis") says that diffusion measurements show that 1.6 is the tortuosity of brain signal pathways.
Because of the tortuosity of brain signal pathways, it seems that we should multiply the previous estimate of six inches by a factor of about 1.6. Doing that leaves you with an estimate of about 10 inches for the average distance that a brain signal would have to travel to go from one random point to another random point in the cortex.
Such a distance may seem small, but when you are dealing with a nerve signal transmission speed of about one tenth of a meter per second (about 4 inches per second), such a distance means a delay of two seconds or more. The problem is that human recall can very often occur instantly. You can ask someone his address or telephone number or the names of his family members, and he will be able to answer instantly, without this requiring a delay of two seconds. And if you ask me, "What's the New York baseball team?" I will answer instantly. There must be a thousand questions you could answer instantly, as soon as someone finished asking them.
No comments:
Post a Comment