Skip to main content

Verified by Psychology Today


Radical Modification of the Standard Model of Memory Storage

The demise of the synapse.

Key points

  • Recent research subverts the hegemony of the synapse.
  • The synapse acts as a conduit to neurons, not as a depository for memories.
  • Cortical neurons as well as glial cells and neurons in the spinal cord also contribute to information processing and storing.
  • The time has arrived for responsible neuroscientists to seriously consider modifying or abandoning the cortico-central hypothesis of memory.
Genny Anderson/Wikipedia Commons, CC BY-SA 4.0
Source: Genny Anderson/Wikipedia Commons, CC BY-SA 4.0

According to the reigning neuroscientific view, short-term memory is linked to functional changes in existing synapses, while long-term memory is associated with a change in the number of synaptic connections and strengthening of the brain’s existing circuitry. As we shall see in this article, this is a highly problematic notion.

Recent Discoveries

Interestingly, it is recent work in this exact domain that has put into doubt the idea of synaptic conductance as the basic memory mechanism. Studies by David Glanzman at the University of California, Los Angeles, exposed Aplysia (a type of sea slug) to mild electric shocks, creating a memory of the event expressed as new synapses in the brain. Then, they transferred neurons from the mollusk into a petri dish and chemically triggered the memory of the shocks in them.

Next, they added propranolol to the neurons. The drug wiped out the mollusk’s synapses formed during learning. When the neuroscientists examined the brain cells, they found that even when the synapses were erased, molecular and chemical changes indicated that the engram, or memory trace, was preserved. These studies suggest that memories are stored inside of neurons in Aplysia and, very likely, in all animals.

In the same vein, researchers at the University of Pennsylvania have discovered in the mouse brain that a key metabolic enzyme, called acetyl-CoA synthetase 2, or ACSS2, works directly within the nucleus of neurons to turn genes on or off when new memories are being established. Using mouse models, researchers in the laboratory of Carlos Lois, at Caltech, determined that strong, stable memories are encoded not, as has been up til recently postulated, by strengthening of the connections to an individual neuron but rather by teams of neurons all firing in synchrony.

Finally, in a 2016 paper entitled “The Demise of the Synapse as the Locus of Memory: A Looming Paradigm Shift?” Patrick Trettenbrein from the Language Development and Cognitive Science Unit at the University of Graz in Austria reviewed the evidence and conclude that the synapse is an ill fit when looking for the brain's basic memory mechanism. It has been repeatedly shown that memory persists despite destruction of synapses and synapses are turning over at very high rates even when nothing is being learned. Taking into consideration all of the preceding, the case against synaptic plasticity is convincing.

To enable cognition and the storage of memories, the present most evidence-based scientific view is that interactions between three moving parts—a binding protein, a structural protein, and calcium—are necessary for electrical signals to enter neural cells and remodel the cytoskeleton. Cytoskeleton is a dense network of various filamentous proteins in all cells with a nucleus, like humans and other animals have, which are essential for cell shape, cell division, and cell migration.

Actin filaments, microtubules, and intermediate filaments form the major components of the cytoskeleton. It is in these cytoskeletons inside neurons where it is thought memories are stored.

For decades, scientists lacked the experimental tools to record the various components of a single nerve cell. Several years ago, researchers realized that a single neuron could function as a logic gate, akin to those in digital circuits.

Recently, researchers in Germany discovered that individual dendrites might process the signals they receive from adjacent neurons before passing them along as inputs to the cell’s overall response. It seems that tiny compartments in the dendritic arms of cortical neurons can each perform complicated operations in mathematical logic.

In theory, almost any imaginable computation might be performed by one neuron with enough dendrites, each capable of performing its own nonlinear operation. A neuron behaving like a multilayered network has much more processing power and can therefore learn or store more data. “Very few people have taken seriously the notion that a single neuron could be a complex computational device,” said Gary Marcus, a cognitive scientist at New York University.

Further research into how neurons perform their tasks by neuroscientist Jeffrey Macklis at Harvard Medical School has challenged the dogma that the nucleus and cell body are the control centers of the neuron. What Macklis’ results suggest is that growth cones—the outermost tips of the axonal strands—are capable of receiving information from the environment, making signaling decisions locally, and functioning semi-autonomously without the cell body.

The growth cones contain much of the molecular machinery of an independent cell, including proteins involved in growth, metabolism, signaling, and more. In other words, here is another example of how our bodies operate with an intricate web of decision-making and the existence of semi-independent units far from the cell nucleus.

According to Douglas Fields, writing in Scientific American, “Neurons are elegant cells, the brain’s information specialists. But the workhorses? Those are the glia.” Astrocytes, which are the largest glial cells, were long considered “junk DNA,” sideline players in the brain. Not anymore.

Human astrocytes are more complex than those of other animals, which suggests that their role in neural processing has expanded with evolution. Because of their size astrocytes span thousands of neurons and millions of synapses and seem to contribute another level of functioning to neural networks.

The spinal cord, which is part of the CNS, like the brain, consists of neurons and supporting glial cells. Researchers at the University of Montreal recently showed that the spinal cord might learn motor skills independently of the brain. Of course, this study also demonstrates, rather persuasively, that neurons outside of the brain retain memories.


The widely held belief that memory is stored in synapses is badly in need of an upgrade. The research studies cited here prove that synapses provide “access points” to neurons. Information flows through the synapses to the neurons. The more information of a particular type enters an assembly of neurons, the stronger will their synapses grow, and the more memory will be stored in these functional ensembles of neurons and glial cells.


Eyal, G., Verhoog, M. B., ... & Segev, I. et al., (2016). Unique membrane properties and enhanced signal processing in human neocortical neurons. eLife, 5, e16553.

Fields, R. Douglas (2011). The Other Brain. Simon and Schuster, New York.

Delgado-García, J. M. (2015). Cajal and the conceptual weakness of neural sciences. Front. Neuroanat. 9:128. doi: 10.3389/fnana.2015.00128

Firestein, S. (2012). Ignorance: How it drives science. OUP USA.

Chen, Shanping, Cai, Diancai, Glanzman, David L. et al., (2014). Reinstatement of long-term memory following erasure of its behavioral and synaptic expression in Aplysia. eLife, 3, e03896.

Fusi, S., & Abbott, L. F. (2007). Limits on the memory storage capacity of bounded synapses. Nature neuroscience, 10(4), 485

More from Thomas R. Verny M.D.
More from Psychology Today