The Mystery Is Not Infantile Amnesia, But How Anything Can Be Instantly Learned or Instantly Recalled

 A recent article at the neuroscience site The Transmitter is entitled "Infant Memories: Lost But Not Gone?" The article has quite a few groundless and incorrect claims about memory, which the authors (Paul Frankland, Sheena Josselyn and Nick Turke-Browne) attempt to substantiate by linking mostly to low-quality papers, some of which they co-authored themselves. 

Early on, the authors state, "The advent of activity-dependent engram labeling and optogenetic tools has enabled researchers to directly address the encoding versus retrieval debate." No such thing as engram labeling exists, and claims that scientists have discovered engrams (reputed places of memory storage in a brain) are without any foundation in good science. 

It is interesting that both of the links above are to papers authored by two of the three Transmitter article authors (Frankland and Josselyn). So we have self-citation. The first paper is a paper by Frankland, Josselyn and Kohler entitled "Engrams." We have the claim, "At the largest scale, engrams are thought to be composed of sparse neuronal ensembles, distributed throughout the brain." That sure does not sound like some specific thing discovered in some particular part of the brain. We read this: "Between 1950 — when Lashley published his scientific magnum opus (In Search of the Engram) — and the late 2000s, barely any scientific articles bore the term engram in their title."

Then we read in the paper about experiments done in recent years. The authors of the paper fail to discuss the low quality of these experiments, and how they were guilty of Questionable Research Practices such as the use of way-too-small study groups, and unreliable methods for measuring how well a rodent remembered. In a paragraph entitled "What is the best evidence for engrams?" we have a reference to rodent experiments using "freezing behavior" judgments (really just immobility tracking).  All such studies are examples of junk science, for reasons explained here. Trying to judge a rodent's immobility (and calling that a tracking of "freezing behavior") is a worthless technique for judging whether an animal remembered something. 

The other link in the quote above from the Transmitter article is to a review article by Frankland, Josselyn and Kohler. It is called "The neurobiological foundation of memory retrieval," but fails to explain how any such thing can occur. The article is one of those review articles in which many a low-quality study is treated as if it was good evidence.

Later in the Transmitter article the authors state, "In mice, researchers can tag engrams in infant pups, track their persistence across development and optogenetically reactivate them later in life to recover seemingly lost memories." That is not at all true. The link is to a low-quality paper "Recovery of 'Lost' Infant Memories in Mice" co-authored by Josselyn, Frankland and others. The study group sizes were way below the minimum of 15 or 20 rodents needed for a study like this to be taken seriously. We read, "Separate groups of infant and adult mice were tested either 1 (P17, N = 9; P60, N = 7), 15 (P17, N = 9; P60, N = 12), 30 (P17, N = 10; P60, N = 7), or 90 days (P17, N = 8; P60, N = 10) after training." The "N" refers to how many rodents were used in particular study groups. Averaging about nine rodents per study group, those are all way-too-small study group sizes. And the worthless method of trying to judge "freezing behavior" was used throughout the study. 

Referring to this low-quality paper as if it was something good, the Transmitter article states this:

"Animals that undergo contextual fear conditioning in infancy show no behavioral evidence of remembering when tested in adulthood (that is, they do not freeze when placed back in a context where they were  shocked as pups). However, optogenetic stimulation of the neurons that were active when the pup was first conditioned triggers expression of the memory; the mature animal now freezes."

But using "freezing behavior" judgments is a worthless technique for trying to measure how well a rodent remembered. And it is particularly invalid to apply "optogenetic stimulation" (a type of brain zapping), and to then claim that this shows that a memory was artificially reactivated because an animal exhibited "freezing behavior." It has been realized by neuroscientists that artificial stimulation of many areas of the brain will produce "freezing behavior" regardless of whether any memory is being recalled. 

Imagine you are running along, and suddenly a scientist switches on some weird thing that causes some energy to pour into your brain. This all by itself might cause you to stop, even if it didn't cause you to recall some memory that caused you to stop. What could have been going on in the mice was just a kind of pausing effect caused by a novel stimulus rather than a recalled fear effect. A science paper says that it is possible to induce freezing [i.e. immobility] in rodents by stimulating a wide variety of regions. It says, "It is possible to induce freezing by activating a variety of brain areas and projections, including the hippocampus (Liu et al., 2012), lateral, basal and central amygdala (Ciocchi et al., 2010); Johansen et al., 2010; Gore et al., 2015a), periaqueductal gray (Tovote et al., 2016), motor and primary sensory cortices (Kass et al., 2013), prefrontal projections (Rajasethupathy et al., 2015) and retrosplenial cortex (Cowansage et al., 2014).”

It is not sound to assume that you can judge the degree of immobility of a mouse in a cage, and then assume that the higher levels of immobility are evidence of a mouse “freezing in fear" because it remembers a fear stimulus.  The "freezing behavior" method of trying to judge recall in mice works like this. A mouse will be trained to fear a shock plate that produces an electrical shock. The mouse will later be put in a cage with such a shock plate. Scientists will then attempt to judge immobility of the mouse during some arbitrary time unit that can be either 30 seconds, a minute, two minutes or three minutes. An arbitrary criteria will be used to judge immobility. For example, not moving for most of ten seconds can be counted as immobility during those seconds, or not moving for most of 20 seconds can be counted as immobility during those seconds, or not moving for  most of 30 seconds can be counted as immobility. Or, any movement at all during those seconds can be counted as movement during that time period. This immobility will be called "freezing," even though scientists have no idea whether the mouse was afraid when it was not moving.  The result of trying to track the mouse's immobility is a "freezing percentage" graph that can be produced in any of 100 different ways. With so many possible analysis choices, it will be almost certain that the experimenter will be able to produce a graph purporting to show the desired difference in "freezing behavior." There are no standards in producing these "freezing behavior" charts. 

Neuroscientists follow this bungling method of trying to measure mouse recall because it is a "see whatever you want to see" method providing the researcher with a near-certainty that the desired results can be claimed. Instead of using so deficient a technique, there are other reliable techniques that can be used. A reliable way to judge whether a a rodent trained to fear some stimulus such as a shock plate is to use some setup such as the one shown below, in which an animal remembering the fear stimulus will take the harder path towards a reward rather than the easier path.

Or, more simply, a very hungry mouse trained to fear a shock plate can be put in a cage like the one show below, and it can be recorded whether he touched the shock plate while trying to get the reward, or whether the mouse did not go get the food (as it would only do if it remembered that the shock plate produces pain). 

The Transmitter article authors then approvingly cite an appalling set of studies trying to show evidence of memory formation in infants by brain scanning them. Such studies are criticized in my post "The Reckless Foolishness of Brain-Scanning Healthy Babies in Neuroscience Experiments."  Having no value in helping to understand memory, experiments such as these run very serious risks to the infants who are needlessly put in brain scanners. There is both a risk of an accident that might harm or kill the child (a young boy once died in an MRI accident), and also a very serious risk that such brain scanning may increase the child's lifetime risk of developing cancer.  

The authors of the paper are trying in a self-serving manner to get us to believe in a socially constructed triumphal legend that researchers such as themselves have done something to substantiate the idea of an engram, a claim that has no basis in robust and well-designed experimental studies. Contrary to their insinuation that this "engram" concept has blossomed in the past 15 years, a search for references to "engram" using the Google Books Ngram viewer shows little evidence of such a blossoming. 

We may compare this result with a search for a word that seems to have skyrocketed in use since the year 2000, the word "ghosts."

Judging from the graph above, you might conclude that the observational evidence for apparitions is far stronger than the observational evidence for engrams.  

The mystery is not why a 70-year-old fails to remember the first five years of his life. The mysteries are how any human is able to remember anything at all, how people are able to instantly answer correctly by giving information they learned decades ago, and also why old people are able to remember very well what occurred to them 50 or 60 years ago. 

Nothing in the brain bears any resemblance to a device for storing learned information, and nothing in the brain bears any resemblance to a device for retrieving learned information. The brain has no known writing mechanism and no known mechanism for reading learned information. The microscopic examination of brain tissue has never detected the slightest trace of anything a human ever learned, even though the brains of many corpses of recently-deceased people have been studied by scientists, and even though much brain tissue extracted from living people has been microscopically studied. 

Neuroscientists senselessly claim that memories are stored in synapses, but the proteins that make up synapses have average lifetimes 1000 times shorter than the maximum length of time that humans remember things (average lifetimes of only a few weeks). Synapses fail to even transmit data reliably, with each transmission across a synaptic occurring with a reliability of only 50% or less. Such signal transmission unreliability should make both a brain storage of memory and a brain retrieval of memory impossible, because signals would have to pass over so very many synapses when either event occurred (there are very many synapses for every neuron). A human can remember the answers to very many thousands of questions instantly, but nothing in a brain can explain such a wonder. Humans construct things that allow the instant retrieval of information, and so we know the type of things that make possible instant information retrieval. Those things are addresses, indexes and sorting. The brain has no addresses, no indexes, and no sorting. The physical architecture of the brain makes sorting within it impossible. Humans can learn things instantly, something that cannot be explained by synapse strengthening which is a slow affair.

Below are some relevant quotes:

• "Synaptic transmission and axonal transfer of nerve impulses are too slow to organize coordinated activity in large areas of the central nervous system. Numerous observations confirm this view [73]. The duration of a synaptic transmission is at least 0.5 ms, thus the transmission across thousands of synapses takes about hundreds or even thousands of milliseconds. The transmission speed of action potentials varies between 0.5 m/s and 120 m/s along an axon. More than 50% of the nerves fibers in the corpus callosum are without myelin, thus their speed is reduced to 0.5 m/s. How can these low velocities (i.e. classical signals) explain the fast processing in the nervous system?" -- The paper "Emission of Mitochondrial Biophotons and their Effect on Electrical Activity of Membrane via Microtubules" by 7 scientists.
• "Neural circuits consist of many noisy, slow components, with individual neurons subject to ion channel noise, axonal propagation delays, and unreliable and slow synaptic transmission." -- Four scientists (link). 
• "Neurons communicate primarily through chemical synapses, and that communication is critical for proper brain function. However, chemical synaptic transmission appears unreliable: for most synapses, when an action potential arrives at an axon terminal, about half the time, no neurotransmitter is released and so no communication happens... Furthermore, when neurotransmitter is released at an individual synaptic release site, the size of the local postsynaptic membrane conductance change is also variable. Given the importance of synapses, the energetic cost of generating action potentials, and the evolutionary timescales over which the brain has been optimized, the high level of synaptic noise seems surprising."  -- Four scientists (link). 
• "The probability of [synaptic] vesicle release [i.e. successful synaptic transmission] is known to be generally low (0.1 to 0.4) from in vitro studies in some vertebrate and invertebrate systems (Stevens, 1994). This unreliability is further compounded by the trial-to-trial variability in the amplitude of the post-synaptic response to a vesicular release." -- Two scientists (link). 
• "The release probability, the average probability that an active zone of a presynaptic terminal releases one or more vesicles following an action potential, is tightly regulated. Measurements in cultured neurons or in slices indicate that this probability can vary greatly between synapses, but on average it is estimated to be as high as 0.5....Existing evidence thus suggests that under physiological conditions in vivo, presynaptic action potentials trigger the release of neurotransmitter much less frequently than what is observed in in vitro preparations." -- A paper by two scientists, suggesting synapses transmit signals with a reliability much less than 50% (link). 
• "On average most synapses respond to only less than half of the presynaptic spikes, and if they respond, the amplitude of the postsynaptic current varies. This high degree of unreliability has been puzzling as it impairs information transmission." -- Four scientists (link).
• "Transmission at individual synaptic contacts on CAI hippocampal pyramidal neurons has been found to be very unreliable, with greater than half of the arriving presynaptic nerve impulses failing to evoke a postsynaptlc response." -- Two scientists (link). 
• "A precise estimate of the in vivo [synaptic] release probability is difficult, but...it can be expected to be closer to 0.1 than to the previous estimates of around 0.5. " -- "The low synaptic release probability in vivo" by J. Gerard G. Borst (link). 
• "The average number of connections between areas in different hemispheres is even smaller, below 1,500 axons. While previous studies have hinted that connectivity between some areas could be sparse [1], the overall sparsity of cortical connections implied by the present study still comes as a surprise. It is as if a traffic system presumed to consist of multilane highways running between most brain areas in fact consists of just a few precarious footpaths." --  The paper "The Highways and Byways of the Brain" by two scientists (link), a paper referring to "the generally very sparse connectivity" of the brain. 
• "The scale of the vast gulf in absolute connectivity between local and long-range connections is startling."--  Paper "An estimation of the absolute number of axons indicates that human cortical areas are sparsely connected" by two scientists (link).
• "Interestingly, signal propagation speeds in various conditions are similar (~0.1 m/s). Neural spikes generated by 4-aminopyridine (4-AP) travel with a longitudinal speed of 0.09/0.03 m/s along the CA3 region (Kiblerand Durand, 2011), whereas in the presence of picrotoxin, synchronous firing events propagate longitudinally at 0.14 /0.04m/s (Miles et al., 1988). High K+-, low Mg2+-, and zero-Ca2+- triggered spikes again exhibit speeds of 0.07-0.1 m/s, 0.1– 0.15m/s, and 0.04 – 0.15 m/s, respectively (Haas and Jefferys, 1984;Quilichini et al., 2002;Liu et al., 2013). In normal tissue, theta oscillations travel with a speed of 0.08 – 0.107 m/s in the hippocampus of living rodent rats (Lubenov and Siapas, 2009),whereas carbachol-induced theta oscillations travels with a speed of 0.119 m/s along the CA1 cell layer and a 0.141 m/s along the CA3 cell layer (Cappaert et al., 2009). Together, it is clear that 0.1m/s is a common propagation speed regardless of experimental models." -- Two scientists telling us that a common speed of brain signal transmission is about 3.6 inches per second (i.e. a tenth of a meter per second), 1000 times slower than the "100 meters per second" commonly given (link).