Each human neuron performs intricate computations that artificial networks struggle to replicate
Beneath the familiar question of what makes human minds exceptional, a quieter answer has begun to surface — not in the vastness of our neural networks, but in the intricate life of a single cell. Researchers from the Hebrew University of Jerusalem and Vrije Universiteit Amsterdam have found that individual human cortical neurons perform computations of a complexity far exceeding those of rodent neurons, driven by sprawling dendritic architectures and highly reactive synaptic proteins. The finding invites us to reconsider intelligence not as a matter of quantity, but as something woven into the very fabric of our biology at its smallest scales.
- A new measurement tool called the Functional Complexity Index has revealed that human brain cells are dramatically harder for artificial networks to predict than rat neurons — a sign of far greater computational sophistication.
- The sheer physical scale of human dendrites allows different regions of a single neuron to process signals semi-independently, multiplying the cell's overall power in ways that challenge our assumptions about where cognition actually lives.
- Highly reactive NMDA receptors at human synapses amplify incoming signals non-linearly, pushing individual neurons into a tier of processing that simple addition of inputs cannot explain.
- An unexpected layer-by-layer shift emerged: while rat brains concentrate complexity in layer five neurons, human brains appear to have evolved their greatest cellular sophistication in layers two and three — regions known to be disproportionately expanded in our species.
- The study's reliance on simulation rather than living tissue leaves open questions, but the relative superiority of human neurons over rodent neurons is expected to hold as research moves toward verification in active brain tissue.
A team of researchers from the Hebrew University of Jerusalem and Vrije Universiteit Amsterdam has demonstrated that individual human cortical neurons possess computational power that may rival small computer systems — a finding that could reshape how we understand the biological roots of human thought.
The study began with a deceptively simple question: do the well-known physical differences between human and rodent neurons actually translate into functional advantage? To find out, the team developed the Functional Complexity Index, a machine-learning tool that works as a kind of biological ruler. Artificial neural networks were trained to predict the behavior of biological neurons; the harder a biological cell was to predict, the higher its complexity score. Using three-dimensional reconstructions of twelve human and twelve rat neurons, the researchers ran twelve thousand simulations per cell, generating the equivalent of more than a day of continuous neural activity data for each model.
The results were striking. Human neurons scored substantially higher on the index, meaning artificial networks struggled far more to anticipate their outputs. Two structural features drove this gap most powerfully: the total surface area of the dendrites — the branching structures that receive incoming signals — and the reactivity of NMDA receptors at the synapses. Larger, more sprawling dendrites allow different regions of a single cell to process information somewhat independently, while highly reactive NMDA receptors amplify simultaneous signals non-linearly rather than simply summing them. Together, these features push human neurons into a markedly higher tier of processing.
The research also uncovered a telling shift in how complexity is distributed across cortical layers. In rats, layer five neurons are the most computationally sophisticated. In humans, that distinction belongs to layers two and three — regions known to be disproportionately expanded in our species, hinting at an evolutionary adaptation in how human brains allocate their cognitive resources.
The study's limitations are worth holding alongside its ambitions. All findings rest on computer simulations rather than living tissue, and certain electrical properties of human dendrites remain poorly understood, meaning the models may not capture every nuance of real neural behavior. Future work aims to verify these patterns in active human brain tissue and to apply the Functional Complexity Index to other cell types and species, including nonhuman primates. What the research ultimately suggests is that human intelligence may arise not simply from the number of neurons we possess, but from the remarkable sophistication of each one.
A single human brain cell, it turns out, is far more sophisticated than we thought—and vastly more capable than the neurons of a rat. A team of researchers from the Hebrew University of Jerusalem and Vrije Universiteit Amsterdam has demonstrated that individual human cortical neurons possess computational power that rivals small computer systems, a finding that may reshape how we understand the biological basis of human thought.
The discovery emerged from a deceptively simple question: Do the physical differences between human and rodent neurons actually matter for how information gets processed? Anatomists have long known that human cortical pyramidal neurons—the cells responsible for transmitting excitatory signals in the thinking brain—are larger and more elaborately branched than their rat equivalents. But measuring whether those structural differences translate into functional advantage has been elusive. To solve this problem, the research team developed a new tool called the Functional Complexity Index, which uses machine learning as a kind of biological ruler.
The method works like this: Researchers built artificial neural networks and trained them to predict the behavior of biological neurons. If a biological neuron acts as a simple switch, a small artificial network learns its output easily. If the biological neuron performs intricate computations, the artificial network struggles. The worse the artificial network performs, the higher the complexity score for the biological cell. Using detailed three-dimensional reconstructions of twelve human neurons and twelve rat neurons, the team ran twelve thousand simulations per cell, each lasting ten seconds, generating the equivalent of more than a day of continuous neural activity data for each model. They then tasked artificial networks with predicting the precise millisecond timing of electrical spikes produced by these biological cells.
The results were striking. Human neurons scored substantially higher on the complexity index than rat neurons, meaning the artificial networks had far greater difficulty predicting their output. This suggests that human brain cells perform a much more intricate translation of incoming signals than rodent neurons do. The team then investigated what drives this difference by analyzing fifty-eight physical measurements of the cells' branching structures. The total surface area of the dendrites—the branch-like structures that receive incoming signals—emerged as the single strongest predictor of complexity. The length of secondary branches also mattered significantly. A larger, more sprawling dendritic architecture allows different regions of the cell to process information somewhat independently, which dramatically amplifies overall computational power.
Another crucial factor emerged from examining synapses, the tiny connection points where signals enter the dendrites. The researchers focused on NMDA receptors, specialized proteins that respond to incoming electrical signals in a non-linear way. When enough signals arrive simultaneously, these receptors amplify the electrical current dramatically rather than simply adding the signals together. Scientific evidence suggests that human excitatory synapses contain more NMDA receptors and react more sharply to voltage changes than rat synapses. When the researchers applied these human-like synaptic properties to their models, the functional complexity of the cells increased significantly. The combination of sprawling dendrites and highly reactive NMDA receptors pushes human neurons into a much higher tier of processing power.
The research also revealed an intriguing shift in how complexity is distributed across the brain's layers. In rat models, neurons in layer five were the most complex. In human models, neurons in layers two and three were significantly more complex than those in other layers. Layers two and three are known to be particularly expanded in the human brain, suggesting an evolutionary adaptation in how human brains allocate their computational resources. This finding hints that human cognition may not simply result from having more neurons, but from having neurons that are individually more sophisticated information processors.
The study does carry limitations worth noting. The research relied entirely on computer simulations rather than living tissue in an active brain. Because certain electrical properties of human dendrites remain poorly understood, the digital models could not include every active ion channel found in real cells, meaning the simulations may behave slightly differently than biological neurons in an actual human brain. Additionally, the Functional Complexity Index depends heavily on the specific design of the artificial neural network used for testing. Different computer architectures could yield different numerical results, though the relative ranking of human versus rat neurons would likely remain consistent.
Future work may verify these computational patterns in living human brain tissue and explore how other cellular features, such as the tiny protrusions on dendrites known as spines, affect signal processing. The researchers also plan to apply this measurement tool to other types of brain cells and to different species, including nonhuman primates. What emerges from this work is a new way of thinking about human intelligence—not as a product of sheer numbers of neurons, but as arising from the remarkable sophistication of individual cells.
Citas Notables
The unique cognitive abilities of humans might stem from the complex structure and function of individual cells, rather than just the vast number of cells in the brain network.— Research team, Hebrew University of Jerusalem and Vrije Universiteit Amsterdam
La Conversación del Hearth Otra perspectiva de la historia
So you're saying a single human brain cell is smarter than a rat's brain cell. How do you even measure that?
You can't measure it directly, so the researchers built artificial neural networks and used them as a measuring stick. They trained these artificial systems to predict what a biological neuron would do. If the artificial network struggles to keep up, that tells you the biological neuron is doing something complex.
That's clever. But why does it matter? We have billions of neurons anyway.
That's the point—it might change how we think about where human intelligence comes from. Maybe it's not just about having more neurons. Maybe it's that each individual neuron in a human brain is doing more sophisticated work.
What makes a human neuron so much more complex?
Two things, mainly. Human neurons have much larger, more elaborately branched structures for receiving signals. And the connection points—the synapses—are more reactive. They amplify signals in a non-linear way, which means they're not just adding things up. They're doing something more like computation.
Did the researchers test this on actual human brains?
No, that's a limitation. They used computer models based on detailed reconstructions of real neurons. They can't yet measure all the electrical properties of living human dendrites, so the simulations are incomplete. The next step would be testing these patterns in actual living tissue.
If this holds up, what does it mean for understanding how we think?
It suggests that human cognition might emerge not just from the architecture of the network—how neurons are wired together—but from the intrinsic computational sophistication of individual cells. That's a different way of thinking about what makes us human.