Learning is a function of the brain. Memory is stored in synapses, the connections between neurons. They become stronger each time a connection takes place. “Neurons that fire together, wire together.”
This theory was put forward by Donald Hebb in 1949, and it is, at best, one small part of the story. It may be completely false. I will briefly reference 6 counterexamples to the neuronal theory of memory, then suggest ideas how we might prospect for a new theory of memory.
Six and a half counterexamples
- Microbes can learn. Learning in paramecia was first noted in 1911. After they have been put in a narrow capillary, they find their way out faster if they had been there before. They learn to avoid an electric shock with classical Pavlovian conditioning, and if they are trained on a platinum wire that has food plaited to it, they will subsequently approach an unplaited wire [Gelber, 1952].
- Planaria are flatworms, about a centimeter long, with unlimited regenerative potential. You can cut off a head, and it grows a new tail, or cut off a tail and it grows a new head. Planaria can also be trained to respond to a Pavlovian stimulus. If you do this and then cut off the worm’s tail, then the tail-half that grows a new head remembers the training, though not as well as the head-half that grows a new tail. [Original ref = McConnell 1959; Replication by Michael Levin’s lab, 2013]
- Caterpillars/butterflies are arthropods, with more advanced nervous systems than worms. Caterpillars can be trained to prefer or to avoid some previously neutral scent. Caterpillars liquefy their brains in the chrysalis on the way to becoming a butterfly, yet the caterpillar’s memories are retained in the butterfly. [Blackiston, 2008]
- Millions of monarch butterflies spend the winter clinging to a particular tree in Pacific Grove, CA. In the spring, they fly off in a diaspora, ranging as far as Canada. But six or seven generations later, the summer is over, and the great great grandchildren of the diaspora turn around and find the same tree where the journey started. Somehow, memory is accumulated and passed through six generations of offspring.
- Heart transplant patients sometimes experience the emotions and even the interests and habits of the heart donor. [Liester, 2020]
- Monica Gagliano is responsible for a new science of plant sociology. Plants communicate with one another, they learn and they remember. Here is her book. Needless to say, plants have no brains, and nothing like a neuron.
- ?? There is literature suggesting that memory is transferable by ingestion. Brains of trained rodents are ground up and injected (or fed) into naive rodents who pick up their conditioned responses. This, if replicated, would establish the basis for specific memories storied chemically. But a few prominent references from the 1960s [Ungar, 1967] led to skeptical rebuttal. The validity of these experiments is considered doubtful today, but anomalies remain.
There is also evidence from out of body experiences, near-death experiences, mediumship, precognition, and telepathy. The credible evidence for such effects is overwhelming, but I’m not going to say more about them here because they almost certainly require an explanation outside known physics. It may be that the six examples above also require new physics, but let’s start by looking for new biology within the confines of conventional physics, and look for new physics only if that effort fails.
What are the possibilities for storing information outside the brain?
- Epigenetics — markers on DNA (or associated histones) that control gene expression patterns. I think of this as a long shot because it’s hard to imagine how the combination of proteins expressed could code for, say, a map of the return route to California. But epigenetic memory is the only known, documented means by which acquired information is passed from parent to offspring.
- Intercellular patterns of electrical potential, as Michael Levin’s lab has been documenting for 20 years. [VIdeo, review article]
- Intracellular patterns of electric potential. (Distinct from what Levin studies which are patterns across groups of cells.)
- Physical structures within a cell, for example the shape of the endoplasmic reticulum, which functions as a highway for proteins transported within a cell. Membrane structures within a cell. There is no direct evidence for this hypothesis, but degradation of the ER is associated with Alzheimer’s dementia.
- Voltage-gated ion channels that constantly pass positive ions (Na+,K+,Ca++,Mg++) in and out of a cell.
Each of these suggests a program for deciding if we’re on the right track. Actually decoding the language of memory is a formidable challenge. But perhaps we can disrupt these systems, one-at-a-time, to see if memory is thereby disrupted, and narrow the search thereby to some candidate vehicle for memory.
The simplest case is for single-celled organisms, so perhaps the inquiry can begin with repeating experiments in paramecia. Try interfering with these candidate mechanisms to see if they affect the degree to which memory is recorded.
Molecules/pathways known to be involved in learning and memory, with homologues in ciliates.
|Molecules/pathways known to be involved in learning/memory||Ciliates with reported homologues||References|
|N-methyl-D-aspartate receptor (NMDAR)||P. primaurelia (only partial sequences)||Ramoino et al., 2014|
|Glutamate receptor||P. tetraurelia||Van Houten et al., 2000|
|Calmodulin||P. tetraurelia||Plattner and Verkhratsky, 2018|
|cAMP||P. tetraurelia||Plattner and Verkhratsky, 2018|
|cAMP-dependent protein kinase||P. tetraurelia||Plattner and Verkhratsky, 2018|
|Mitogen activated protein kinase (MAPK)||P. caudatum||Wada and Watanabe, 2007|
|Protein kinase C (PKC)||T. thermophilia||Hegyesi and Csaba, 1994|
|Calcineurin||P. tetraurelia||Plattner and Verkhratsky, 2018|
|DNA methyltransferases (DNMTs)||T. thermophilia||Gutiérrez et al., 2000|
|Histone acetyltransferases (HATs)||T. thermophilia||Vavra et al., 1982|
|Histone deacetylases (HADCs)||T. thermophilia||Wiley et al., 2005|
Epigenetics — Methyl transferases could enhance memory storage, if the epigenetic hypothesis is correct. If inhibition of methyl transferases prevents memory storage, that would point toward epigenetic memory. There are, however, too many epigenetic mechanisms independent of methylation for this experiment to be dispositive.
Gated ion channels can be manipulated with drugs and with certain frequencies of radio waves.
If physical structures within the cell are the seat of memory, the best way to study this might be to apply AI algorithms to photographs of cells that store different memories, before and after these memories are recorded. This is virgin territory, AFAIK.
Levin has found planaria to be his most productive model organism. They readily learn and they have extraordinary power of regeneration. Where is the memory stored while an amputated piece of a planarian is regenerating its brain? If it is stored in patterns of electric potential, then the same arsenal of chemicals that Levin has used to manipulate electric potentials in other contexts should be effective for studying the mechanism of extra-neural memory.
Plant memory is well-established, thanks to the meticulous work of Gagliano, but inertia is keeping researchers from flocking to the field. There is some foundational work, and I was surprised to learn that plant electrical signaling has been studied for 200 years. [Review] Electrical signals would probably be the best place to begin a search for memory in plants. Unlike in animals, electrical signaling in plants is decentralized, a departure from the command-and-control model that we think about so naturally. This suggests that plant memory may have properties of fractal scaling and self-organization that will make for future paradigm shifts concerning flow of biological information.