A New Paper Reminds Us Neuroscientists Can't Get Their Story Straight About Memory Storage

There is a new scientific paper with the inappropriate title "Where is Memory Information Stored in the Brain?" This is not the question we should be asking. The question we should be asking is: "Is memory information stored in the brain?"  Although it was probably not the intention of the authors (James Tee and Desmond P. Taylor), what we get in the paper is a portrait of how neuroscientists are floundering around on this topic, like some poor shark that is left struggling in the sand after going after its prey too aggressively. 

Tee and Taylor claim this on page 5: "Based on his discovery of the synapse as the physiological basis of memory storage, Kandel was awarded the year 2000 Nobel Prize in Physiology or Medicine (Nobel Prize, 2000)." This is a misstatement about a very important topic. The Nobel Prize listing for Kandel does not mention memory. The official page listing the year 2000 Nobel Prize for physiology states only the following: "The Nobel Prize in Physiology or Medicine 2000 was awarded jointly to Arvid Carlsson, Paul Greengard and Eric R. Kandel 'for their discoveries concerning signal transduction in the nervous system.' " The Nobel committee did not make any claim that synapses had been discovered as the basis of memory. 

Before making this claim about the Nobel Prize, Tee and Taylor  state something that makes no sense. They state, "The groundbreaking work on how memory is (believed to be) stored in the human brain was performed by the research laboratory of Eric R. Kandel on the sea slug Aplysia (Kupfermann et al., 1970; Pinsker et al., 1970)." How could research on a tiny sea slug tell us how human beings store memories?  The paper in question can be read here. The paper fails to mention a testing of more than a single animal, thereby strongly violating rules of robust experimental research on animals (under which an effect should not be claimed unless at least 15 subjects were tested).  We have no reliable evidence about memory storage from this paper. If the paper somehow led to its authors getting a Nobel Prize, that may have been a careless accolade.  The Nobel Prize committee is pretty good about awarding prizes only to the well-deserved, but it may occasionally fall under the gravitational influence of scientists boasting about some "breakthrough" that was not really any such thing. 

Equally undeserving of a Nobel Prize was the next research discussed by our new paper on memory storage: research claiming a discovery of "place cells" in the hippocampus. John O' Keefe published a paper in 1976 claiming to detect "place units" in the hippocampus of rats. The paper also used the term "place cells."  The claim was that certain cells were more active when a rat was in a certain spatial position. The paper did not meet standards of good experimental science. For one thing, the study group sizes it used were way too small for a robust evidence to have produced.  One of the study group sizes consisted of only five rats, and another study group size consisted of only four rats.  15 animals per study group is the minimum for a moderately convincing result.  For another thing no blinding protocol was used. And the study was not a pre-registered study, but was apparently one of those studies in which an analyst is free to fish for whatever effect he may feel like finding after data has been collected. 

The visuals in the study compare wavy signal lines collected while a rat was in different areas of an enclosed unit. The wavy signal lines look pretty much the same no matter which area the rats were in. But O'Keefe claims to have found differences.  No one should be persuaded that the paper shows robust evidence for an important real effect.  We should suspect that the analyst has looked for stretches of wavy lines that looked different when the rat was in different areas, and chosen stretches of wavy lines that best-supported his claim that some cells were more active when the rats were in different areas.  Similar Questionable Research Practices (with similar too-small study groups such as four rats) can be seen in O'Keefe's 1978 paper here. 

Although O'Keefe's 1976 paper and 1978 paper were not at all a robust demonstration of any important effect, the myth that "place cells" had been discovered started to spread around among neuroscience professors.  O'Keefe even got a Nobel Prize. The Nobel Prize committee is normally pretty good about awarding prizes only when an important discovery has been made for which there was very good evidence. Awarding O'Keefe a Nobel Prize for his unconvincing work on supposed "place cells" seems like another flub of the normally trusty Nobel Prize committee. Even if certain cells are more active when rats are in certain positions (something we would always expect to observe from chance variations), that does nothing to show that there is anything like a map of spatial locations in the brain of rats. 

On page 7 of the new paper on memory storage, we have a discussion of equally unconvincing results:

"LeDoux found that this conditioned fear resulted in LTP (strengthening of synapses) in the auditory neurons of the amygdala, to which he concluded that the LTP constituted memory of the conditioned fear. That is, memory was stored by way of strengthening the synapses, as hypothesized by Hebb."

You may understand why this is nothing like convincing evidence when you realize that synapses are constantly undergoing random changes. At any moment billions of synapses may be weakening, and billions of other synapses may be strengthening.  So finding some strengthening of synapses is no evidence of memory formation. It is merely finding what goes on constantly in the brain, with weakening of synapses occurring just as often as strengthening. The new paper on memory storage confesses this when it says on page 8 that: "synapses in the brain are constantly changing, in part due to the inevitable existence of noise." 

On pages 8-9 of the new paper, Tee and Taylor say that scientists had hopes that there would be breakthroughs in handling memory problems by studying synapses, but that "the long-awaited breakthroughs have yet to be found, raising some doubts against Hebb’s synaptic [memory storage] hypothesis and the subsequent associated experimental findings." Tee and Taylor give us on page 9 a quotation from two other scientists, one that gives a great reason for rejecting theories of synaptic memory storage:

"If we believe that memories are made of patterns of synaptic connections sculpted by experience, and if we know, behaviorally, that motor memories last a lifetime, then how can we explain the fact that individual synaptic spines are constantly turning over and that aggregate synaptic strengths are constantly fluctuating? How can the memories outlast their putative constitutive components?"

Tee and Taylor  then tell us that this problem does not just involve motor memories:

"They further pointed out that this mystery existed beyond motor neuroscience, extending to all of systems neuroscience given that many studies have found such constant turn over of synapses regardless of the cortical region. In order words, synapses are constantly changing throughout the entire brain: 'How is the permanence of memory constructed from the evanescence of synaptic spines?' (Bizzi & Ajemian, 2015, p. 92). This is perhaps the biggest challenge against the notion of synapse as the physical basis of memory."

Tee and Taylor then discuss various experiments that defy the synaptic theory of memory storage.  Most of the studies are guilty of the same Questionable Research Practices that are so extremely common in neuroscience research these days, so I need not discuss them.  We hear on page 14 about various scientists postulating theories that are alternatives to the synaptic theory of memory storage:

"The logical question to pose at this point is: if memory information is not stored in the synapse, then where is it? Glanzman suggested that memory might be stored in the nucleus of the neurons (Chen et al., 2014). On the other hand, Tonegawa proposed that memory might be stored in the connectivity pathways (circuit connections) of a network of neurons (Ryan et al., 2015). Hesslow emphasized that memory is highly unlikely to be a network property (in disagreement with Tonegawa), and further posited that the memory mechanism is intrinsic to the neuron (in agreement with Glanzman) (Johansson et al., 2014)."

You get the idea? These guys are in disarray, kind of all over the map, waffling around between different cheesy theories of memory storage. All of the ideas mentioned above have their own fatal difficulties, reasons why they cannot be true.  In particular, there is no place in a neuron where memory could be written, with the exception of DNA and RNA; and there is zero evidence that learned knowledge such as episodic memories and school lessons are stored in DNA or RNA (capable of storing only low-level chemical information).  Human DNA has been extremely well-studied by long well-funded multi-year research projects such as the Human Genome Project completed in 2003 and the ENCODE project, and no one has found a bit of evidence of anything in DNA that stores episodic memory or any information learned in school.

Tee and Taylor then give us more examples of experiments that they think may support the idea of memories stored in the bodies of neurons (rather than synapses). But they fail to actually support such an idea because the studies follow Questionable Research Practices.  For example, they cite the study here, which fails to qualify as a robust well-designed study because it uses study group sizes as small as 9, 11 and 13. To give another example, Tee and Taylor cite the Glanzman study here, which  fails to qualify as a robust well-designed study because it uses study group sizes as small as 7. Alas, the use of insufficient sample sizes is the rule rather than the exception in today's cognitive neuroscience, and Tee and Taylor seem to ignore this problem.  

The heavily hyped Glanzman study (guilty of Questionable Research Practices) claimed a memory transfer between aplasia animals achieved by RNA injections. Such a study can have little relevance to permanent memory storage, because RNA molecules have very short lifetimes of less than an hour. 

Finally in Tee and Taylor's paper, we have a Conclusions section, which begins with this confession which should cause us to doubt all claims of neural memory storage: "After more than 70 years of research efforts by cognitive psychologists and neuroscientists, the question of where memory information is stored in the brain remains unresolved."  This is followed by a statement that is at least true in the first part: "Although the long-held synaptic hypothesis remains as the de facto and most widely accepted dogma, there is growing evidence in support of the cell-intrinsic hypothesis."  It is correct to call the synaptic memory hypothesis a dogma (as I have done repeatedly on this blog). But Tee and Taylor commit an error in claiming "there is growing evidence in support of the cell-intrinsic hypothesis" (the hypothesis that memories are stored in the bodies of neurons rather than synapses that are part of connections between neurons).  There is no robust evidence in support of such a hypothesis, and the papers Tee and Taylor have cited as supporting such a hypothesis are unconvincing because of their Questionable Research Practices such as too-small sample sizes. 

On their last two page the authors end up in shoulder-shrugging mode, saying, "while the cell might be storing the memory information, the synapse might be required for the initial formation and the subsequent retrieval of the memory."  We are left with the impression of scientists in disarray, without any clear idea of what they are talking about, rather like some theologian speculating about exactly where the angels live in heaven, bouncing around from one idea to another.  In their last paragraph Tee and Taylor speculate about memories being inherited from one generation to another by DNA, which is obviously the wildest speculation. 

Our takeaway from Tee and Taylor's recent paper should be this: scientists are in baffled disarray on the topic of memory. They have no well-established theory of memory storage in the brain, and are waffling around between different speculations that contradict each other.  We are left with strong reasons for suspecting that scientists are getting nowhere trying to establish a theory of memory storage in the brain.  This is pretty much what we should expect if memories are not stored in brains, and cannot be stored in brains.  Always be very suspicious when someone says something along the lines of, "What scientists have been teaching for decades is not true, but they have a new theory that has finally got it right." More likely the new theory is as false as the old theory. 

If anyone is tempted to put credence in this "cell-intrinsic hypothesis" of memory storage, he should remind himself of the physical limitations of DNA.  DNA uses what is called the genetic code. The genetic code is shown below. The A, C, T and G letters at the center stand for the four types of nucleotide base pairs used by DNA:  adenine (A), cytosine (C), guanine (G), and thymine (T). Different triple combinations of these base pairs stand for different amino acids (the twenty types of chemicals shown on the outer ring of the visual below). 

So DNA is profoundly limited in what it can store. In the human body DNA can only store low-level chemical information. We know of no way in which DNA in a human body could store any such things as information learned in school or episodic memories.  Such things cannot be stored using the genetic code used by DNA.  No one has ever found any evidence that strings of characters (such as memorized text) are stored in human DNA, nor has anyone found any evidence that visual information is stored in human DNA. Moreover, if we had to write memories to DNA or read memories from DNA, it would be all-the-more impossible to explain the phenomena of instant memory formation and instant memory retrieval. 

Some have suggested that DNA methylation marks might be some mechanism for memory storage. This idea is very unbelievable. DNA methylation is the appearance of a chemical mark on different positions of DNA.  The chemical mark is almost always the same H₃C addition to the cytosine nucleotide base pair.  These chemical marks serve as transcription suppressors which prevent particular genes from being expressed. Conceptually we may think of a DNA methylation mark as an "off switch" that turns off particular genes. 

The idea that the collection of these chemical "off switches" can serve as a system for storing memories is unbelievable. DNA is slowly read by cells in a rather sluggish process called transcription, but there is no physical mechanism in the body for specifically reading only DNA methylation marks. If there were anything in the body for reading only DNA methylation marks, it would be so slow that it could never account for instant memory recall.  We know the purpose that DNA methylation marks serve in the body: the purpose of switching off the expression of particular genes. Anyone claiming that such marks also store human memories is rather like some person claiming that his laundry detergent is a secret system for storing very complex information. 

A metric relevant to such claims is the maximum speed of DNA transcription. The reading of DNA base pairs occurs at a maximum  rate of about 20 amino acids per second, which is about 60 nucleotide pairs per second.  This is the fastest rate, with preparatory work being much slower. DNA methylation occurs only for one of the four base pairs, meaning that no more than about 15 DNA methylation marks could be read in a second (after slower preparatory work is done).  

Let us imagine (very implausibly) that DNA methylation marks serve as a kind of binary code for storing information.  Let us also imagine (very implausibly) that there is a system by which letters can be stored in the body, by means of something like the ASCII code, and by means of DNA methylation.  Such a system would have storage requirements something like this:

Letter	ASCII number equivalent	Binary equivalent
A	12	1100
B	13	1101
C	14	1110

Under such a storage system, once the exact the spot had been found for reading the right information (which would take a very long time given that the brain has no indexing system and no position coordinate system), and after some chemical preparatory work had been done to enable reading from DNA, information could be read at a rate of no more than about four characters per second. But humans can recall things  much faster than such a rate. When humans talk fast, they are speaking at a rate of more than two words per second (more than 10 characters per second).  So if you ask me to describe how the American Civil War began and started and ended, I can spit out remembered information at a rate several times faster than we can account for by a reading of DNA methylation marks, even if we completely ignore the time it would take to find the right little spot in the brain that stored exactly the right information to be recalled. 

A realistic accounting of the time needed for memory recall of information stored in binary form by DNA methylation would have to add up all of these things:

• The time needed for finding the exact spot in the brain where the correct recalled information was stored (requiring many minutes or hours or days, given no indexing and no coordinate system in the brain);
• The time needed for chemical preparatory work that would have to be done before DNA can be read (such as the time needed to get RNA molecules that can do the reading);
• Reading DNA methylation marks (encoding binary numbers) at a maximum rate of no more than four characters per second (and usually a much slower rate because of a sparse scattering of such marks);
• Translating such binary numbers into their decimal equivalent;
• Translating such decimal numbers into character equivalents;
• Translating such retrieved letters into speech.

All of this would be so slow that if memories were stored as DNA methylation marks, you would never be able to speak correct recalled information at a rate a tenth as fast as two words per second, as humans can do. Similarly, you would never be able to form new memories instantly (as humans are constantly doing) if memory storage required writing binary information as DNA methylation marks, which would be a very slow process.  Humans can form new memories at the same rate at which they can recall memories. Suppose you are leaving to go food shopping and someone in your house says, "Please buy me a loaf of whole wheat bread and some orange juice." You may form a new memory of those exact words, at a rate of two words per second.  Storing such information as DNA methylation marks would be much slower than such a rate. 

I may note that while scientists can read DNA and DNA methylation marks from neural tissue, no one has ever found the slightest speck of human learned information stored in DNA or DNA methylation marks, synapse strengths, or any other type of representation in the brain; nor has anyone found any evidence of any coding scheme by which letters or numbers or visual images are stored in human DNA or DNA methylation marks.  When brain surgeons remove half of a brain (to treat very severe seizures) or remove portions of a brain (to treat severe epilepsy or cancer), they discard the cut-out brain tissue, and do not try to retrieve memory information stored in it.  They know that attempting such a thing would be utterly futile. 

It has been claimed by the few proponents of memory stored in the DNA methylation marks that such marks are a stable medium for writing information. But search for "DNA methylation turnover" and you will find contrary claims, such as a paper entitled "Rapid turnover of DNA methylation in human cells."