You've probably seen the diagram. A neuron drawn like a cartoon tree: a round cell body, a long tail shooting off to one side, and a spray of branches at the other end. That said, textbook stuff. But here's the thing — most people label the parts and move on without ever asking why it's built that way.
The branches? And they're not just passive antennae waiting for a signal to land. Also, those are dendrites. They're where the real computation happens Worth keeping that in mind. Took long enough..
What Is a Dendrite
Dendrites are the branched extensions of a neuron that receive incoming signals from other neurons. The word comes from the Greek dendron, meaning tree. And branch. Because of that, fitting, because they branch. And branch again — sometimes into thousands of tiny twigs per cell.
But calling them "receivers" sells them short.
Each dendrite is covered in microscopic protrusions called dendritic spines. Consider this: a single pyramidal neuron in your cortex might have 10,000 of them. These tiny knobs — shaped like lollipops, mushrooms, or thin filaments — are the actual contact points. Think about it: ten thousand independent inputs. Consider this: one spine, one synapse. All converging on one cell.
And they're not static. Spines grow. Shrink. Appear. Disappear. They change shape in response to activity. This is structural plasticity in real time — the physical basis of learning That's the part that actually makes a difference..
The membrane matters
Dendritic membranes aren't just passive wire. Others act like leaky cables, passively spreading current toward the soma. Some dendrites can generate their own action potentials. They're studded with voltage-gated ion channels — sodium, calcium, potassium — distributed in precise gradients. The mix varies by cell type, by brain region, even by branch order.
A Purkinje cell in the cerebellum has a dendritic tree so vast and flat it looks like a coral fan. Consider this: same basic job. A hippocampal granule cell has short, claw-like dendrites that barely leave the cell body layer. Wildly different geometries Simple, but easy to overlook..
Why It Matters / Why People Care
If you're a student, this is the answer to a test question: "Dendrites receive information." Memorize it, pass the quiz, move on.
But if you're trying to understand how the brain actually works — how memory forms, why Alzheimer's destroys cognition, how psychedelics rewire circuits — dendrites are where the action is.
The integration problem
A neuron doesn't just sum up inputs like a calculator. It integrates them. Nonlinearly. Two inputs arriving on the same dendritic branch within a few milliseconds can trigger a local spike — a dendritic spike — that never reaches the soma but still changes synaptic strength. So inputs on different branches? That said, they might sum linearly. Or they might not interact at all Easy to understand, harder to ignore. But it adds up..
This means a single neuron can perform multiple computations in parallel. Different branches, different logic gates. One cell, thousands of tiny processors No workaround needed..
Disease lives here
In Alzheimer's, dendritic spines vanish before neurons die. So they regrow them. In depression, chronic stress shrinks dendritic arbors in the prefrontal cortex and hippocampus. In Fragile X syndrome, spines are too long, too thin, too many — immature. Antidepressants? Ketamine does it in hours.
Schizophrenia shows reduced dendritic branching in cortical pyramidal cells. Autism spectrum disorders often show increased spine density — a pruning deficit.
The dendrite isn't just a part of the neuron. It's where neuropsychiatric disease writes its signature.
How It Works
Let's walk through the journey of a signal — from another neuron's axon terminal to the dendritic spine, and what happens next.
Step 1: The synapse
An action potential arrives at the presynaptic terminal. In real terms, voltage-gated calcium channels open. Vesicles fuse. It diffuses. In real terms, glutamate (usually) spills into the synaptic cleft — a 20-nanometer gap. Binds to receptors on the spine head.
Two main receptor types matter here:
- AMPA receptors — fast, sodium-permeable, mediate the bulk of excitatory transmission
- NMDA receptors — slower, calcium-permeable, voltage-dependent (blocked by magnesium at rest)
The AMPA current depolarizes the spine. In practice, if it's strong enough, it kicks out the magnesium block from NMDA receptors. Calcium floods in.
That calcium is the trigger. Not just for this synapse — for everything that follows.
Step 2: Local computation
The spine head is tiny — femtoliter volume. Calcium concentration spikes fast. This activates calcium-dependent enzymes: CaMKII, calcineurin, PKC. They phosphorylate AMPA receptors (making them more conductive), recruit more receptors to the membrane, or trigger structural changes in the spine cytoskeleton.
This is long-term potentiation (LTP) — the cellular correlate of memory. It happens in the spine. Locally. Input-specific.
But the signal doesn't stop there Worth keeping that in mind..
Step 3: Dendritic propagation
The depolarization spreads from the spine into the dendritic shaft. How far it goes depends on:
- Dendritic diameter (thinner = more resistance)
- Ion channel density (more channels = more leak)
- Distance from soma
- Branch geometry
In many pyramidal neurons, distal inputs (far out on the apical tuft) barely reach the soma on their own. On top of that, a nonlinear event. But if many distal inputs arrive together — or if a backpropagating action potential from the soma meets them halfway — you get a dendritic calcium spike. A local decision No workaround needed..
This is why dendrites are called "compartments." They're electrically semi-independent.
Step 4: Somatic integration
All these local events — EPSPs, dendritic spikes, inhibitory inputs (GABAergic synapses often target the shaft or soma directly) — converge at the axon initial segment. That's the spike initiation zone. If the summed depolarization crosses threshold there, an action potential fires.
Worth pausing on this one.
But the dendrites already did the heavy lifting. The soma just reads the result.
Backpropagation matters
When the axon fires, that action potential doesn't just go forward. It backpropagates into the dendritic tree. Not perfectly — it attenuates, especially in thin distal branches. But it reaches far enough to coincide with incoming synaptic input Surprisingly effective..
Why does this matter? " That's a teaching signal. In practice, **Spike-timing-dependent plasticity (STDP). If it's active just after, it weakens (LTD). The dendrite compares "what I heard" with "what the cell decided to do.Now, ** If a synapse is active just before the backpropagating spike arrives, it strengthens (LTP). Right there in the branch.
Common Mistakes / What Most People Get Wrong
"Dendrites are just wires"
No. But wires don't compute logarithms of input timing. Wires don't grow new connection points when you learn a language. Wires don't change their resistance based on recent activity. Dendrites do all three Not complicated — just consistent..
"The cell body is the neuron"
The soma is tiny compared to the dendritic tree. In a cortical pyramidal cell, the dendritic membrane area is 95% of the total. And the soma is a rounding error. Most synapses are on dendrites. Now, most ion channels are on dendrites. Most plasticity happens on dendrites. The cell body is the readout unit. The dendrites are the processor It's one of those things that adds up..
"All dendrites work the same"
A cerebellar Purkinje cell receives 200
...million parallel fiber synapses — but they're arranged in a precise, columnar pattern. Each Purkinje dendrite integrates inputs from hundreds of parallel fibers in a way that's completely different from how a cortical pyramidal neuron handles its apical tuft versus basal dendrites.
A hippocampal pyramidal cell's dendrites in CA1 have entirely different ion channel compositions than those in CA3. Even within the same neuron, proximal and distal dendrites can behave like different computational units.
"Synapses are just points"
A synapse isn't a static dot. It's a dynamic structure that changes size, shape, and receptor density based on activity. A single spines can enlarge by 300% within minutes of strong stimulation. The entire dendritic branch can remodel its arborization over days.
The Real Story
Neurons don't compute like classical logic gates. They compute like ecosystems — with feedback loops, emergent properties, and local decisions that cascade upward.
The dendrite is where the first "yes" or “no” gets made. Not in the cell body. Because of that, not in the synapse. In the branch.
Memory isn't stored in the nucleus. It's encoded in the electrical behavior of dendritic segments. Think about it: learning isn't just strengthening connections. It's teaching dendrites how to listen, integrate, and respond.
And that’s why neuroscientists who still draw neurons with neat little circles for somas and straight lines for dendrites are missing the point entirely.
The brain doesn’t run on cell bodies.
It runs on dendrites.
