- Atlas
- Natural Sciences
- Systems Neuroscience
Systems Neuroscience
Systems neuroscience is the branch of neuroscience that asks how groups of neurons, wired into circuits, produce perception, movement, memory, and emotion.
11 frameworks, from 1860 to today · Neuroscience · Natural Sciences
- Functional specialization 1860-Present
- Reflex physiology and integrative action 1890-1950
- Equipotentiality 1920-1970
- Single-unit receptive field analysis 1926-Present
- Topographic maps and cortical columns 1940-Present
- Computational circuit modeling 1952-Present
- Population and distributed coding 1980-Present
- Functional neuroimaging 1985-Present
- Cognitive systems neuroscience 1990-Present
- Causal circuit manipulation 2005-Present
- Large-scale population recording 2005-Present
Overview
Systems neuroscience is the branch of neuroscience that asks how groups of neurons, wired into circuits, produce perception, movement, memory, and emotion. Its unit of study sits between the single cell and the whole mind: not "how does one neuron fire?" but "how does a population of neurons, connected in a particular way, compute something an animal actually does?" The field's central wager is that behavior can be explained by the organized activity of identified circuits—that if you know which neurons connect to which, and how their signals combine, you can predict what the animal will sense, decide, and do. What is at stake is nothing less than a mechanistic account of the mind, built from the bottom up.
Reflexes and the Integrative Action of the Nervous System
The intellectual foundation was laid in the late nineteenth and early twentieth centuries by physiologists studying the simplest possible behaviors. Charles Sherrington, working in Britain, studied the spinal reflexes of animals and argued in The Integrative Action of the Nervous System (1906) that the nervous system is not a set of independent pathways but an integrated machine1. His key insight was the synapse—a term he coined—as the site where signals from different sources converge and compete. A single reflex arc, he showed, involves excitation and inhibition interacting: the same stimulus that contracts one muscle relaxes its antagonist. Sherrington's principle of the "final common path"—many inputs funneling onto a shared motor output—established that neural circuits make decisions by weighing competing signals. This was systems thinking before the term existed: behavior explained by the interaction of connected parts.
Localization of Function and Its Discontents
In parallel, a debate raged over whether brain functions are localized to specific regions or distributed across the whole. The phrenologists of the early nineteenth century had claimed that faculties like "combativeness" lived in particular bumps of the skull—a theory discredited as method2. But the underlying question survived. Paul Broca's observation in the 1860s that damage to a specific left-hemisphere region destroyed speech production, and Carl Wernicke's later identification of a region for language comprehension, gave localization a clinical footing. In the 1870s, Gustav Fritsch and Eduard Hitzig showed that electrically stimulating different parts of a dog's cortex produced movements of different body parts, and David Ferrier extended this to monkeys, mapping motor areas. By the early twentieth century, Korbinian Brodmann had divided the cortex into numbered areas based on cell architecture, many of which turned out to correspond to distinct functions.
Yet the opposite camp had evidence too. Karl Lashley, searching for the "engram"—the physical trace of memory—in rats trained on mazes, found that the amount of cortex removed mattered more than its location. He concluded that memory was distributed and that cortex worked equipotentially, a holistic view3. The tension between localization and distribution was not resolved so much as reframed: some functions are sharply localized, others depend on distributed populations, and the question became how local circuits participate in larger networks.
Single Neurons and Their Receptive Fields
The decisive methodological turn came with the ability to record the electrical activity of individual neurons in intact, behaving or lightly anesthetized animals. In the late 1920s, Edgar Adrian in Cambridge showed that a single sensory nerve fiber carries information in the rate of its impulses—stronger stimuli produce more frequent spikes. This established the spike rate as a carrier of information and made single-unit recording a viable window into the brain.
The approach reached its first great triumph in the visual system. In the late 1950s and early 1960s, David Hubel and Torsten Wiesel, recording from the visual cortex of cats, discovered that neurons there respond to specific features of the visual scene. A receptive field—the region of the visual world that influences a neuron's firing—was for retinal cells a small spot of light, but for cortical cells it was a line or edge at a particular orientation, and some cells responded best to edges moving in one direction. Hubel and Wiesel proposed a hierarchy: simple cells combine inputs from center-surround retinal cells to detect edges; complex cells pool simple cells to detect edges regardless of exact position. This was a concrete, mechanistic account of how a circuit builds a feature detector, and it became the model for sensory neuroscience.
Cortical Maps and Columns
Hubel and Wiesel's work also revealed that cortical neurons are organized spatially. Cells with similar orientation preferences sit next to each other, forming orientation columns, and the cortex contains repeating functional units. Earlier, in the 1940s, Wade Marshall, Samuel Talbot, and others had mapped the somatosensory cortex of monkeys, showing that the body surface is represented in an orderly map—the homunculus—with disproportionately large areas for sensitive body parts like the fingers and lips. Wilder Penfield later mapped the human motor and sensory cortices during brain surgery, confirming the principle in people. These topographic maps showed that the brain preserves spatial order, and that the size of a representation reflects behavioral importance. The discovery that maps can reorganize after injury or training—plasticity—made them dynamic rather than fixed.
Distributed and Population Coding
Single-unit recording had a limitation: it listened to one voice at a time. By the 1970s and 1980s, researchers recognized that information is often carried by populations. In the monkey motor cortex, Apostolos Georgopoulos and colleagues showed in the 1980s that the direction of an upcoming arm movement is encoded not by any single neuron but by the population vector—the weighted sum of many neurons' preferred directions. Each neuron "votes" for its preferred direction, and the collective vote predicts the movement. This was a direct demonstration that a circuit computes by distributed representation, reconciling the localization and holistic traditions: individual neurons are tuned, but the code lives in the ensemble.
Computational Circuit Modeling
As recordings accumulated, so did the ambition to explain circuits mathematically. The Hodgkin–Huxley model (1952) had described how a single neuron's membrane generates spikes from ion currents; systems neuroscience extended this to networks45. David Marr's 1969 work on the cerebellum and his 1970s framework for understanding vision as computation—asking what problem a circuit solves, by what algorithm, and in what hardware—gave the field a vocabulary. In the 1980s, connectionist models and the study of attractor networks (circuits that settle into stable activity patterns, proposed as a basis for memory) linked circuit structure to computation. Modeling became a way to test whether a proposed wiring diagram could actually produce the observed behavior, and to predict what recordings should find.
Functional Neuroimaging and the Human Circuit
A parallel revolution came from being able to watch the human brain at work. In the early 1990s, functional magnetic resonance imaging (fMRI)—detecting changes in blood oxygenation as a proxy for neural activity—made it possible to see which brain regions are active during a task in a living person6. Positron emission tomography (PET) had preceded it in the 1980s. These methods brought systems neuroscience into human cognition: researchers could ask which networks support face recognition, navigation, or decision-making. The limitation was resolution in time and space—fMRI measures a slow hemodynamic signal, not spikes—so imaging identified where and when roughly, while animal electrophysiology supplied the cellular detail. The two approaches became complementary, and the concept of the large-scale network—distributed regions cooperating—became central.
Cognitive Systems Neuroscience
By the 1990s and 2000s, systems neuroscience increasingly addressed cognitive questions directly: how does a circuit hold a memory in working storage, choose between options, or represent the value of a reward? Recordings in the prefrontal cortex, hippocampus, and basal ganglia revealed neurons whose activity tracked task rules, spatial location, or expected reward. The hippocampus became a showcase: place cells, discovered by John O'Keefe in the 1970s, fire when an animal is in a specific location, and grid cells, found by Edvard and May-Britt Moser in the 2000s, provide a coordinate system. Together they suggested a circuit for navigation and memory. This fusion of systems physiology with cognitive questions—sometimes called cognitive systems neuroscience—treated the circuit as the substrate of thought.
Causal Circuit Manipulation
A persistent problem was correlation: recording a neuron's activity during a behavior shows association, not causation. The solution arrived with tools to switch specific neurons on or off. Optogenetics, developed in the mid-2000s by Karl Deisseroth and others, uses light-sensitive proteins inserted into genetically targeted neurons so that shining light can activate or silence them. Combined with chemogenetics (drugs that activate engineered receptors), researchers could test whether a circuit is necessary or sufficient for a behavior. This moved systems neuroscience from description to intervention: if silencing a population abolishes a behavior, that population is causally involved. The approach, alongside genetic tools for labeling cell types, became a standard method for linking circuit activity to function.
Population and Large-Scale Recording
The most recent turn is technological: the ability to record from hundreds or thousands of neurons simultaneously. Multi-electrode arrays, tetrodes, and later high-density silicon probes (such as Neuropixels, introduced in the late 2010s) let researchers observe entire populations at once, revealing dynamics invisible to single-unit studies—sequences of activity, population states, and low-dimensional structure in the neural code. Calcium imaging with fluorescent indicators allowed optical recording of large populations in behaving animals7. These methods, combined with machine learning for analysis, made it possible to ask how circuits represent information in the joint activity of many cells, and to connect population dynamics to behavior.
The Present
Today systems neuroscience is defined by the integration of these strands: genetic tools to identify and manipulate cell types, high-density recording to observe populations, imaging to see human networks, and computational models to interpret the data. The field descends directly from Sherrington's integrative reflexes, Hubel and Wiesel's receptive fields, and the population-coding tradition. Open questions remain: how do circuits learn and adapt; how do local computations scale to whole-brain networks; what is the relationship between the activity of identified neurons and subjective experience; and how do the maps and columns of the cortex actually implement the computations that behavior requires. The tools have changed, but the central problem—how connected neurons produce what an animal does—remains the field's organizing question.
References
- Charles Scott Sherrington, Wikipedia
- Phrenology, Wikipedia
- Equipotentiality, Wikipedia
- Single-unit recording (History), Wikipedia
- Computational neuroscience (History), Wikipedia
- Functional magnetic resonance imaging, Wikipedia
- Calcium imaging, Wikipedia
The frameworks of Systems Neuroscience
Each framework is a way of working the field has used: when it held, and the question it answers.
- 1860-Present
- 1890-1950
Reflex physiology and integrative action
How do connected nerve cells produce coordinated movement?
- 1920-1970
Equipotentiality
Is memory stored in particular places or spread across the brain?
- 1926-Present
Single-unit receptive field analysis
What features of the world make a single neuron fire?
- 1940-Present
Topographic maps and cortical columns
How is the body and the world laid out in the cortex?
- 1952-Present
Computational circuit modeling
Can a wiring diagram mathematically produce observed behavior?
- 1980-Present
Population and distributed coding
How do groups of neurons together carry information?
- 1985-Present
Functional neuroimaging
Which brain regions are active during a mental task?
- 1990-Present
Cognitive systems neuroscience
How do circuits hold memories, choose actions and represent value?
- 2005-Present
Causal circuit manipulation
Does turning a circuit on or off change behavior?
- 2005-Present
Large-scale population recording
What can thousands of neurons recorded at once tell us?
How the ideas connect
Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.
Equipotentiality reacted against Functional specialization
Lashley's equipotentiality was formulated explicitly against the localization of function, using his rat maze experiments to argue that memory depends on the amount of cortex rather than specific regions. He defined his holistic position by rejecting the phrenological and Broca-style localization he had been trained in.
Carried by Karl Lashley.
Topographic maps and cortical columns grew out of Single-unit receptive field analysis
Hubel and Wiesel's single-unit recordings in cat visual cortex directly produced the discovery of orientation columns and repeating functional units, extending receptive field analysis into a spatial map of cortical organization. Their microelectrode method and hierarchy of simple and complex cells became the basis for columnar models.
Carried by David Hubel; Torsten Wiesel.
Large-scale population recording grew out of Population and distributed coding
Large-scale population recording is the technological generalization of population coding: instead of summing the tuning of a few dozen neurons, high-density probes and imaging record thousands at once. The theoretical commitment that information lives in ensembles drove the push to record entire populations.
Single-unit receptive field analysis influenced by Reflex physiology and integrative action
Adrian's demonstration that a single sensory fiber encodes stimulus intensity in spike rate grew from the reflex physiology tradition of recording from isolated nerve-muscle preparations. It supplied the method and the concept of the spike as a unit of neural signaling that made single-unit receptive field analysis possible.
Carried by Edgar Adrian.
Computational circuit modeling influenced by Single-unit receptive field analysis
Marr's computational models of the cerebellum and visual cortex were built to explain the response properties of single neurons revealed by receptive field recording. The data and questions of single-unit analysis set the target that circuit modeling sought to explain mathematically.
Carried by David Marr.
Population and distributed coding influenced by Single-unit receptive field analysis
Population coding arose from single-unit recording: Georgopoulos and colleagues recorded many motor cortex neurons individually and combined their tuning curves into a population vector. The method and the concept of the tuned neuron were taken directly from receptive field analysis.
Carried by Apostolos Georgopoulos.
Functional neuroimaging influenced by Topographic maps and cortical columns
Functional neuroimaging adopted the topographic map as its interpretive framework: fMRI and PET activations were read as changes in the same cortical areas and maps that Penfield, Marshall and Hubel–Wiesel had defined. The maps provided the anatomical reference system for human imaging studies.
Carried by Wilder Penfield; Wade Marshall.
Population and distributed coding influenced by Computational circuit modeling
Attractor network models from computational circuit modeling gave population coding a theoretical language: distributed activity patterns in recurrent networks were proposed as the substrate for memory and computation. The population vector and ensemble codes were interpreted through these network models.
Carried by John Hopfield.
Cognitive systems neuroscience grew out of Population and distributed coding
Cognitive systems neuroscience extended population and distributed coding to cognitive questions, recording ensembles in prefrontal cortex and hippocampus to show how populations represent rules, locations and rewards. The ensemble framework became the tool for asking how circuits support cognition.
Carried by John O'Keefe; Edvard Moser; May-Britt Moser.
Cognitive systems neuroscience influenced by Functional neuroimaging
Human functional imaging identified large-scale networks for face recognition, navigation and decision-making, which set the cognitive questions that animal circuit recordings then addressed. The network concept from imaging carried directly into cognitive systems neuroscience.
Causal circuit manipulation grew out of Cognitive systems neuroscience
Causal circuit manipulation grew out of cognitive systems neuroscience's need to test whether recorded activity in identified circuits actually causes the behavior. Optogenetics and chemogenetics were developed to move from correlation in cognitive recordings to causal intervention.
Carried by Karl Deisseroth; Ed Boyden.
Beyond the field
Where Systems Neuroscience's ideas came from in other fields, and what they led to there: the strongest 12 of 68 links.
Functional specialization led to Lateralization of brain function in Neuropsychology
Lateralization research is a specialization of the broader principle that neural tissue is functionally differentiated. The split-brain experiments of the 1960s carried that principle to the level of the cerebral hemispheres.
Carried by Roger Sperry.
Functional specialization influenced Epilepsy Surgery in Epilepsy
Functional specialization in systems neuroscience, built on cortical stimulation and ablation experiments, provided the conceptual basis for identifying and removing a localized epileptogenic zone. Epilepsy surgery rests on this principle.
Carried by David Ferrier; Gustav Fritsch.
Reflex physiology and integrative action led to Systems Neuroscience in Neuroscience
Systems neuroscience grew out of reflex physiology and Sherrington's integrative action, taking its concept of coordinated neural circuits and extending it to whole systems.
Carried by Charles Scott Sherrington.
Reflex physiology and integrative action grew out of Reflex Doctrine in Neuroscience
Reflex physiology and integrative action grew directly out of the older reflex doctrine, generalizing the simple reflex arc into a model of coordinated nervous action across the whole neuraxis.
Carried by Charles Scott Sherrington.
Reflex physiology and integrative action influenced Reflex-chaining motor control physiology in Motor Learning
Sherrington's integrative action of the nervous system, with its chaining of reflexes and reciprocal inhibition, is the direct source of reflex-chain theories of motor control.
Carried by Charles Scott Sherrington; The Integrative Action of the Nervous System (1906).
Single-unit receptive field analysis influenced Feature-detector and hierarchical feedforward models in Computational Neuroscience
Hubel and Wiesel's single-unit recordings of simple and complex cells in cat visual cortex supplied the receptive-field hierarchy that Fukushima's Neocognitron and later feedforward feature-detector models formalised. The physiological findings were the direct template for the model architecture.
Carried by David H. Hubel; Torsten Wiesel; Kunihiko Fukushima.
Computational circuit modeling influenced Computational Neurostimulation Modeling in Neural Engineering
Computational circuit modelling provided the network-level equations and simulation tools that neurostimulation models use to predict how applied currents propagate through and recruit neural circuits.
Large-scale population recording influenced by Multielectrode array neurophysiology in Neural Engineering
Multielectrode array neurophysiology is the direct technical basis of large-scale population recording: high-density silicon probes and arrays enabled simultaneous recording from hundreds to thousands of neurons.
Functional specialization influenced Systems Neuroscience in Neuroscience
Functional specialization of cortical regions, established by clinical and lesion work, gave systems neuroscience its founding premise that distinct circuits serve distinct functions.
Carried by Paul Broca.
Functional specialization reacted against Holistic and Network Conceptions of Brain Function in Neurology
Functional specialization defined itself against the holistic and network views of brain function, arguing from circumscribed lesion deficits that distinct regions carry distinct functions.
Carried by Paul Broca.
Functional specialization led to Hemispheric Specialization in Cognitive Neuroscience
Hemispheric specialization is a branch of the older functional specialization principle: the idea that distinct functions are carried out by distinct neural structures was applied to the two hemispheres as separate specialized systems.
Functional specialization influenced Lesion and stimulation mapping in Behavioral Neuroscience
Functional specialization provided the premise that different cortical regions serve different functions, which lesion and stimulation mapping then tested experimentally. The mapping program operationalized specialization.

Systems Neuroscience is a planet in Nousfarer
Its city has a tower for each of these 11 frameworks. Land, walk in, and climb one concept at a time.
Land on Systems NeuroscienceRelated fields
Fields its ideas are linked to
- Neuroscience 12 links
- Behavioral Neuroscience 9 links
- Neurology 7 links
- Computational Neuroscience 6 links
- Neural Engineering 4 links
- Cognitive Neuroscience 4 links
- Neurophysiology 4 links
- Epilepsy 4 links
- Engineering Physiology 3 links
- Action And Motor Cognition 2 links
- Neurolinguistics 2 links
- Neuroimaging 2 links
More in Neuroscience
- Behavioral Neuroscience
- Cellular Neuroscience
- Cognitive Neuroscience
- Computational Neuroscience
- Developmental Neuroscience
- Molecular Neuroscience
- Neuroimaging
- Neurophysiology
- Neuroscience
See where Systems Neuroscience sits on the Map of Human Knowledge