A revolutionary theory challenging mainstream neuroscience about how memories are physically stored in the brain
What if everything we know about memory is wrong? Mainstream neuroscience teaches that memories are stored in the connections between your brain cells. But for over half a century, C. R. (Randy) Gallistel, an eminent Professor of Psychology Emeritus at Rutgers University, has championed a revolutionary alternative: the vast library of your experiences isn't written in neural networks, but in molecules inside each individual cell 1 6 9 .
This idea, once considered heretical, is gaining traction thanks to daring experiments that challenge the fundamental assumptions of how brains work. Gallistel, a member of the National Academy of Sciences and past president of the Association for Psychological Science, argues that discovering the physical basis of memory is as important for neuroscience as the discovery of DNA was for biology 1 2 6 . This is the story of his intellectual crusade to find the engram—the physical trace of memory.
At the heart of Gallistel's theory is the computational theory of mind—the view that the brain is a powerful, organic computer 1 . To compute, it needs to store and manipulate not just vague associations, but precise, abstract quantities. From the duration of a traffic light to the distance to your destination, your brain constantly works with numbers 1 7 .
If synapses aren't the answer, what is? Gallistel points to a system that already exists in nature: the molecular machinery of the cell, particularly polynucleotides like DNA and RNA 1 7 .
Biology already uses molecules to store vast amounts of digital information in a tiny space. Gallistel finds it implausible that evolution would invent a completely new and less efficient system for memory when every cell in the body already contains a proven, high-capacity information storage system 7 .
"Usually when I ask neuroscientists how you encode a number... that's a conversation stopper," Gallistel has noted. "All I get is hand waves... Well, you see, there are lots of synapses and it's a pattern. Could you say something about the pattern? How does the pattern for 11 differ from the pattern for three?" 9
He argues that there is no known biological mechanism that could read and write complex, dynamic data with the required speed and reliability from a distributed network of synapses. For that, you need a system more like a computer's memory, which relies on addressed, read-write memory 7 9 .
This experiment isolated a maximally simple memory—the duration of an interval—and traced it to a single, huge neuron in the cerebellum 6 .
The experiment involved classical conditioning of an eye blink. A mild stimulus (like a touch on the paw) served as a warning, followed by a brief puff of air to the eye. Over time, the ferret learns to blink at exactly the right moment to protect its eye 6 .
The memory of the interval between the warning and the threat is the engram. The groundbreaking discovery was that this memory was stored not in a circuit of neurons, but within a single cell 6 . Furthermore, Johansson identified the first molecular step in the chain that leads to the formation of this memory, a receptor molecule that initiates the process 6 .
A ferret is presented with a neutral conditioned stimulus (CS), such as a touch on the paw, which is followed after a precise time interval by an unconditioned stimulus (US), like a mild shock near the eye 6 .
After repeated pairings, the ferret learns the interval. It blinks just before the US occurs, closing its eye at the exact moment the shock is predicted 6 .
Researchers demonstrated that this specific timed memory (the duration engram) is stored within a single Purkinje cell in the cerebellum. Input from the warning signal is routed to this specific cell 6 .
A subsequent single-spike input to this very same cell triggers the read-out of the stored interval memory, leading to a perfectly timed blink 6 .
Johansson identified the first molecular receptor in the sequence of events inside the neuron that leads to the encoding of the interval duration 6 .
Gallistel likens this receptor to the "ball-of-thread" the mythical figure Ariadne gave to Theseus to find his way through the labyrinth 6 . It is the first clue that molecular biologists can follow to trace the path all the way to the physical engram itself.
| Aspect | Standard Model Prediction | Ferret Experiment Finding |
|---|---|---|
| Storage Location | Distributed across a network of synapses | Inside a single neuron |
| Engram Nature | A pattern of connection strengths | A cell-intrinsic, duration-encoding mechanism |
| Key Process | Synaptic plasticity | Intracellular molecular cascade |
| Recall Mechanism | Pattern activation in a network | Read-out from within the cell |
To follow Ariadne's thread to the engram, scientists will need a sophisticated molecular toolkit. Gallistel believes that the necessary tools largely already exist in the field of molecular biology 6 .
| Tool / Technique | Function in Memory Research |
|---|---|
| Genetically Modified Mice | Allows researchers to screen for memory malfunctions by testing memory for simple quantities like time and number 2 . |
| Psychophysical Screening | Behavioral methods that test memory hundreds of times to extract quantitative properties of the underlying mechanisms, similar to methods used in sensory research 2 . |
| Molecular Visualization & Manipulation | An array of techniques that allow scientists to see and interfere with the actions of tiny molecular machines inside neurons 6 . |
| Jeffreys Priors (Bayesian Stats) | A type of statistical analysis that improves parameter estimates by incorporating prior knowledge, allowing for stronger conclusions from data 8 . |
Gallistel himself has incorporated Bayesian data analysis into his work, arguing it provides a more intuitive and powerful way to draw conclusions from experimental data compared to traditional statistical methods 5 8 .
Why has this revolutionary view been so hard for neuroscience to accept? Gallistel points to a 2,300-year-old problem: Aristotle 6 .
The intuitive, Aristotelian theory that memory is built from associations between sensations (sight, sound, smell) remains the foundation of most neuroscientific thinking about learning and memory. This "associative theory" is directly embodied in the connectionist model of synaptic plasticity 6 7 .
The ferret experiment provides a clear path forward. By following the molecular pathway inside the neuron, researchers could potentially identify the physical engram within a decade or two 6 .
Gallistel envisions a future where we understand that memories are stored in molecular structures—perhaps in RNA or other polymers—that have the functional properties of polynucleotides: stable, high-capacity, and energy-efficient, much like the memory in a thumb drive 6 .
"Mainstream brain-scientists have no idea how brains could store a number, retrieve that number... That whole way of thinking... is not part of their conception" 6 .
| Feature | Standard Synaptic Model | Gallistel's Intracellular Model |
|---|---|---|
| Core Idea | Memory is in connection strengths between neurons. | Memory is in molecules inside individual neurons. |
| Primary Mechanism | Synaptic plasticity. | Molecular encoding (e.g., in DNA/RNA-like molecules). |
| Nature of Storage | Analog, pattern-based. | Digital, symbolic (numbers). |
| Computational Model | Neural networks, connectionism. | Classical computer (read/write memory). |
| Evolutionary Precedent | None (a novel invention). | Repurposing of existing cellular memory systems. |
Gallistel's legacy is that of a scientific pioneer, steadfastly arguing for a more computationally plausible vision of the mind. His career, spanning over six decades, is a testament to the power of a single, disruptive question: Where is the engram? The answer, he insists, is not in the connections between our neurons, but in the incredible machinery within them 1 6 9 .