Neurons come in a staggering variety of shapes, sizes, and chemical profiles, and no single classification system captures them all. The traditional approach sorts them by shape or by the direction they carry signals, but modern tools that read gene activity in individual cells have revealed hundreds of distinct neuron types in the human brain alone. Understanding how scientists slice up this diversity matters because different types of neurons do fundamentally different jobs, break down in different diseases, and respond to different drugs.
The Shape-Based Scheme
The oldest way to classify neurons is by their physical structure, specifically the number and arrangement of branches extending from the cell body. This approach goes back to the late 1800s, when staining techniques first let researchers see individual nerve cells under a microscope. The basic categories are still taught and still useful.
Multipolar neurons have one long output fiber and many shorter input branches radiating from the cell body. They are the most common type in the brain and spinal cord, and the classic pyramidal cell of the cerebral cortex is a textbook example. Bipolar neurons have just two extensions, one on each side of the cell body, and are found in sensory pathways like vision and smell. Unipolar neurons, common in invertebrates, have a single branch that splits into two directions. Closely related are pseudo-unipolar neurons, the sensory cells that sit in clusters called dorsal root ganglia along the spinal cord. These start out bipolar during development but their two branches fuse near the cell body, leaving a T-shaped process that sends one arm to the skin or organs and the other into the spinal cord.
1Frontiers in Cellular Neuroscience. Human Dorsal Root GangliaShape is a useful starting point, but it can be misleading. Two neurons with similar branching patterns can use different chemical messengers, fire at different speeds, and connect to completely different circuits. This is why neuroscientists rarely rely on shape alone.
Functional Groupings
A second classic scheme divides neurons by what they do in a circuit. Sensory neurons carry information from the body and the outside world toward the brain and spinal cord. Motor neurons carry commands outward to muscles and glands. Interneurons sit between the two, relaying and processing signals within the central nervous system. Interneurons vastly outnumber the other two categories in the brain, which makes sense given how much internal processing the brain does relative to its input and output traffic.
These categories are intuitive but coarse. A “sensory neuron” that detects a pinprick in your fingertip is doing a very different job from a “sensory neuron” that registers the stretch of your stomach wall, yet both fall into the same functional bin. Researchers have argued that a more meaningful functional classification could be built around how each neuron encodes information, using measures from information theory to define optimal numbers of cell types based on the statistics of their inputs and the noise in their responses.
2PubMed Central. Toward functional classification of neuronal typesChemical Identity
Neurons can also be grouped by the chemical messenger they release at their connections. The two biggest camps are excitatory and inhibitory. Most excitatory neurons in the brain use glutamate, while most inhibitory neurons use GABA. Maintaining the right balance between excitation and inhibition turns out to be critical; growing evidence links disruptions in this balance to neurological and neurodevelopmental disorders.
3PubMed Central. The Yin and Yang of GABAergic and Glutamatergic Synaptic Plasticity: Opposites in Balance by Crosstalking MechanismsBeyond glutamate and GABA, smaller populations of neurons use neuromodulatory chemicals that tune broad swaths of brain activity rather than flipping individual switches. Dopamine neurons in the midbrain are central to reward, motivation, and movement. Noradrenaline neurons clustered in the brainstem, particularly in the locus coeruleus, project widely and support alertness, learning, and the body’s stress response. These brainstem noradrenergic clusters are remarkably heterogeneous; genetic tracing reveals distinct subpopulations that project to different targets, from circuits involved in cognition to pathways that regulate skeletal muscle.
4PubMed Central. Brainstem noradrenergic neurons: Identifying a hub at the intersection of cognition, motility, and skeletal muscle regulationSerotonin neurons in the dorsal raphe nucleus are another major neuromodulatory population. Interestingly, the dorsal raphe also contains dopamine-producing neurons that form an electrophysiologically distinct class with their own signature firing properties, separate from the serotonin cells sitting right next to them.
5eLife. Electrophysiological and morphological properties of dorsal raphe neurons in a mouse model of Parkinson’s diseaseThe amygdala, a region tied to emotion and threat detection, sends projections to many of these modulatory populations. In rats, the central amygdala most heavily innervates specific dopaminergic and noradrenergic cell groups in the brainstem, providing a direct anatomical route through which emotional states can shift the brain’s chemical tone.
6PubMed. Organization of amygdaloid projections to brainstem dopaminergic, noradrenergic, and adrenergic cell groups in the ratPyramidal Neurons and Cortical Layers
The cerebral cortex, the wrinkled outer sheet of the brain responsible for perception, thought, and voluntary action, is organized into layers. The principal excitatory cells in the cortex are pyramidal neurons, named for the roughly triangular shape of their cell bodies. Different layers contain pyramidal cells with strikingly different wiring.
Pyramidal neurons in layers 2 and 3 send their output fibers to other cortical areas, connecting distant patches of cortex to integrate sensory information with motor planning. They tend to have smaller cell bodies, more compact branching trees, and lower firing rates. Pyramidal neurons in layer 5, by contrast, are a major source of cortical output to subcortical structures. Their fibers reach the thalamus, the striatum, the midbrain, and even the spinal cord. Layer 5 cells have larger bodies, more elaborate branches, and higher spontaneous firing rates.
7Frontiers in Neural Circuits. Pyramidal Neurons in Different Cortical Layers Exhibit Distinct Dynamics and Plasticity of Apical Dendritic SpinesWithin the deep layers themselves, researchers distinguish at least two broad classes. Intratelencephalic neurons keep their projections within the cerebral hemispheres, while extratelencephalic neurons also project to targets outside, such as the brainstem and spinal cord. These subpopulations differ in their electrical behavior, the synaptic inputs they receive, and their roles in generating behavior.
8PubMed Central. Specialized Subpopulations of Deep-Layer Pyramidal Neurons in the Neocortex: Bridging Cellular Properties to Functional ConsequencesCortical Interneurons
If pyramidal neurons are the cortex’s long-range messengers, interneurons are its local editors. These inhibitory cells shape cortical activity by dampening, timing, and filtering the signals that pyramidal cells send. They are wildly diverse, and researchers have identified several major molecular subtypes based on the proteins they contain.
Parvalbumin-positive interneurons are fast-spiking basket cells that form tight connections around pyramidal cell bodies, giving them powerful control over when a pyramidal cell fires. Somatostatin-positive cells, often called Martinotti cells, send branches upward through the cortical layers and tend to target the input-receiving branches of pyramidal neurons rather than the cell body. A third group expresses calretinin or vasoactive intestinal polypeptide and often inhibits other interneurons rather than pyramidal cells directly, creating a disinhibitory circuit: they quiet the cells that quiet the output cells, effectively releasing the brakes. A fourth group consists of large basket cells positive for cholecystokinin.
9PubMed. Parvalbumin, somatostatin and cholecystokinin as chemical markers for specific GABAergic interneuron types in the rat frontal cortexThese subtypes are not just molecular curiosities. Each one targets a different part of the pyramidal neuron and fires with a different rhythm, giving the cortex a toolkit for precisely sculpting activity patterns. Disruptions in specific interneuron populations have been implicated in epilepsy and schizophrenia.
Specialized Cells Worth Knowing
Purkinje Cells of the Cerebellum
Purkinje cells are among the most visually dramatic neurons in the brain. Each one has an enormous, flat, fan-shaped branching tree that collects input from hundreds of thousands of other cells. In humans, their total dendritic length is roughly eleven times greater than in mice, averaging about 63,600 micrometers compared to about 6,000 in mice. Human Purkinje cells also have nearly four times the number of branch segments, reflecting branching that reaches almost double the maximum complexity seen in mice.
10eLife. Human Purkinje cell dendrites scale beyond allometric limits and host unique morphological featuresAll of that elaborate wiring serves the cerebellum’s role in coordinating movement, timing, and motor learning. Purkinje cells are the sole output neurons of the cerebellar cortex, meaning everything the cerebellum computes ultimately funnels through them.
Von Economo Neurons
Von Economo neurons are large, spindle-shaped cells found in specific regions of the cortex in great apes and humans, primarily the frontoinsular cortex and the limbic anterior cingulate cortex. Their unusually elongated cell bodies and simple branching pattern suggest they are built for speed, rapidly relaying basic information to distant brain regions while slower neighboring pyramidal cells send more detailed signals.
11PubMed Central. The von Economo neurons in the frontoinsular and anterior cingulate cortexTheir selective destruction in the early stages of frontotemporal dementia, a condition marked by loss of empathy and social awareness, hints that these cells play a role in social cognition and self-control. Evolutionary analyses suggest they may have originated in the frontoinsular cortex before the split of hominids and Old World monkeys, becoming abundant in the anterior cingulate cortex only in the hominid lineage.
12Frontiers in Mammal Science. Von Economo neurons as a specialized neuron class of the human cerebral cortexNeurons That Build Internal Maps
Some of the most celebrated neuron types discovered in recent decades are involved in spatial navigation. Place cells in the hippocampus fire when an animal occupies a specific spot in an environment, essentially tagging “you are here.” Grid cells in the entorhinal cortex, one step upstream from the hippocampus, fire in a repeating hexagonal pattern across space, providing a metric, almost like graph paper, that the hippocampus can use to build maps.
13PubMed. How do spatial learning and memory occur in the brain? Coordinated learning of entorhinal grid cells and hippocampal place cellsGrid cells come in different scales: some tile small spaces with fine resolution, others tile large spaces with coarser spacing, and they are organized along a gradient in the entorhinal cortex. The hippocampal place fields they feed into represent much larger spaces, supporting the kind of flexible navigational behavior that lets you find your way around a new city. Beyond navigation, these same circuits appear to support long-term memory storage through offline consolidation processes tied to sharp-wave ripple activity during sleep and rest.
14PubMed Central. Place cells, grid cells, and memoryA Third Kind of Photoreceptor
Most people learn that the eye has two types of photoreceptors, rods and cones. But the mammalian retina contains a third type: intrinsically photosensitive retinal ganglion cells, or ipRGCs. Unlike rods and cones, these cells are not primarily involved in forming images. Instead, they contain a light-sensitive pigment called melanopsin and signal overall environmental light levels to the brain’s circadian clock.
15PubMed Central. Mood, the Circadian System, and Melanopsin Retinal Ganglion CellsThese neurons are the reason bright light at night suppresses melatonin and disrupts sleep, and why totally blind people who have lost all rod and cone function can still synchronize their body clocks to the day-night cycle. In humans, the ipRGC-driven pupil response shows its own circadian rhythm, independent of external light cues, with the minimum response occurring after melatonin onset.
16PLoS ONE. The Circadian Response of Intrinsically Photosensitive Retinal Ganglion CellsClinically, the pupil light reflex driven by ipRGCs has become a readily accessible marker of their activity, useful for studying retinal diseases that spare or damage different photoreceptor classes.
17PubMed. Melanopsin-expressing intrinsically photosensitive retinal ganglion cells in retinal diseaseWhy Some Neuron Types Die First in Disease
Neurodegenerative diseases do not kill neurons at random. Each disease has a characteristic pattern of damage, and within a targeted region, certain neuron subtypes are far more vulnerable than others. In Parkinson’s disease, dopaminergic neurons in the ventral tier of the substantia nigra pars compacta are devastated, with up to about 90% lost by the time of death, while dopaminergic neurons in the dorsal tier of the same nucleus lose only around 25%.
18PubMed. Selective cell death in neurodegeneration: why are some neurons spared in vulnerable regions?This selective vulnerability is one of the strongest arguments for why classifying neurons matters beyond academic curiosity. If you want to develop treatments that protect the cells most at risk, you first need to know what makes them different from their neighbors. The same logic applies to frontotemporal dementia’s early targeting of von Economo neurons and to the selective loss of certain cortical interneuron subtypes in disorders like schizophrenia. Understanding the molecular and physiological identity of each neuron type is a prerequisite for understanding why it is vulnerable.
How Modern Genomics Is Redrawing the Map
Traditional classification relied on what you could see under a microscope, record with an electrode, or stain with a chemical marker. Each approach gave a partial picture, and the pictures did not always agree with each other. Two neurons that looked identical could turn out to express different genes; two neurons with different shapes could share a molecular profile.
Single-cell RNA sequencing has transformed this landscape. By reading the gene activity of individual neurons, researchers can now build molecular taxonomies across entire brain regions. One large-scale effort profiled over 70,000 individual nuclei across five human brain areas, including the hippocampus, amygdala, and prefrontal cortex, to map the cell-type diversity within circuits involved in reward.
19PubMed Central. Single-nucleus transcriptome analysis reveals cell-type-specific molecular signatures across reward circuitry in the human brainThese molecular surveys consistently find more types than anyone expected. Regions that were thought to contain a handful of neuron classes turn out to harbor dozens, distinguished by subtle but reproducible differences in which genes are switched on.
Combining All the Data at Once
The real frontier is not any single classification axis but the integration of all of them. A technique called Patch-seq combines electrical recording, cell-shape reconstruction, and gene-expression profiling from a single neuron in one experiment. This lets researchers directly ask whether molecular subtypes match up with electrical subtypes and structural subtypes, or whether the different axes cut the population in conflicting ways.
20PubMed Central. Multimodal profiling of single-cell morphology, electrophysiology, and gene expression using Patch-seqEarly results are encouraging: in many cases, molecular types do correspond to distinct electrical and morphological signatures, suggesting that the brain’s neuron types are not arbitrary labels but reflect genuinely distinct biological identities. The approach has been described as taking neuroscience into a “multimodal era,” where classification is no longer constrained to one lens at a time.
21PubMed Central. Patch-seq: Advances and Biological ApplicationsThe Debate That Started It All
It is easy to take for granted that the brain is made of discrete cells, but that idea was itself controversial. In the late 1800s, Camillo Golgi, the man whose staining technique made individual neurons visible for the first time, believed that the nervous system was a continuous network of fused filaments rather than a collection of separate units. Santiago Ramón y Cajal used Golgi’s own method to argue the opposite: that each neuron is an anatomically, functionally, and developmentally separate individual. The two shared the 1906 Nobel Prize despite never resolving their disagreement.
22PubMed. Reticular theory versus neuron theory in the work of Camillo GolgiCajal’s view, now called the neuron doctrine, won out and forms the foundation of modern neuroscience. But the tension between holistic and atomistic views of the brain echoes in today’s classification debates. A single neuron’s identity depends on its context within a circuit as much as on its intrinsic properties. As classification tools become more powerful, the field is grappling with where to draw lines in what is, ultimately, a continuum of diversity shaped by development, experience, and evolution.