Cortical Function: What It Is and How It Works

Cortical function refers to everything the cerebral cortex does, and the list is vast: seeing, hearing, touching, planning a movement, holding a phone number in mind, understanding a sentence, deciding what to eat for dinner, and daydreaming about tomorrow. The cortex is the roughly three-millimeter-thick sheet of neural tissue covering the outer surface of the brain, and it houses the circuitry behind perception, voluntary action, language, abstract thought, and much of what we consider conscious experience. What makes it work is not one mechanism but a constellation of structural features, electrical dynamics, chemical signals, and feedback loops operating across scales from individual synapses to brain-wide networks.

How the Cortex Is Built

The cortex has a layered architecture. Neuron cell bodies are organized into six main layers stacked from the surface inward, each with distinct cell types and connection patterns. Neurons in layers two through six called intratelencephalic cells connect within the cortex itself and to the basal ganglia. A different population in one sublayer of layer five sends long-range projections down the entire length of the central nervous system, while cells in layer six project specifically back to the thalamus. Layer four, meanwhile, is the main landing zone for incoming sensory information from the thalamus.¹ This layered design is not decorative; it separates incoming signals, local processing, and outgoing commands into physically distinct tiers.

Running perpendicular to these layers is a radial organization often described as “cortical columns.” A column is a vertical slice of cortex in which neurons from all six layers share response properties and tend to be activated together. Whether columns are literally the “basic functional units” of cortical processing has been debated for decades, and the term itself is used loosely to refer to everything from a cluster of cells to a pattern of connectivity or gene expression.² Still, the basic insight holds: the cortex is organized both horizontally across its surface and vertically through its depth, and both dimensions matter for how it processes information.

The Balancing Act Between Excitation and Inhibition

Cortical neurons communicate by either exciting or inhibiting their neighbors. Roughly one in five cortical neurons is inhibitory, and the rest are excitatory. That ratio sounds lopsided, but it turns out to be close to optimal for keeping the network stable and responsive.³ If excitation overwhelms inhibition, activity spirals out of control and the network can tip into seizure-like oscillations. If inhibition dominates, the network goes quiet and stops carrying useful signals.

Computational modeling shows that when excitatory and inhibitory inputs are both strong and roughly cancel each other out, neurons become far better at selectively responding to specific stimuli even in the presence of noisy, unpredictable input.⁴ Think of it like two powerful forces pulling in opposite directions: the tension between them makes the system exquisitely sensitive to small tilts. This excitatory-inhibitory balance is not static. It shifts with arousal, attention, and learning, and its disruption is implicated in conditions ranging from epilepsy to schizophrenia.

Several structural and physiological features keep this balance tuned. The speed at which inhibitory synapses decay and how densely inhibitory neurons connect to each other both turn out to be critical control points. Interestingly, the balance is robust enough against changes in individual synaptic strengths to accommodate learning and memory, which require those strengths to change.³ The system can store new information by tweaking individual connections without destabilizing the whole network.

How Brain Regions Talk to Each Other

Neurons do not fire in isolation. Across the cortex, large groups of neurons fall into rhythmic patterns of activity, oscillating between more and less excitable states many times per second. These rhythms serve as a communication mechanism. When two brain regions oscillate in sync, their neurons are excitable at the same moments, so signals sent from one arrive at the other during a window of high receptivity. The result is stronger, more effective communication. When the two regions fall out of sync, the same signals arrive at random phases and have much less impact.⁵

This “communication through coherence” idea helps explain how the brain can selectively route information. A cortical area receiving input from multiple sources does not have to physically disconnect from one to listen to another. It just synchronizes its oscillations with the source it needs at that moment and falls out of sync with the rest. Attention, for instance, may work partly by adjusting which regions are rhythmically coupled. This mechanism operates on top of the anatomical wiring; the physical connections set the menu of possible conversations, while oscillatory coherence selects which conversations actually happen.

Sensing the World and Controlling Movement

Some of the best-understood cortical functions involve sensory processing and motor control. In the primary visual cortex, individual neurons respond preferentially to edges at particular orientations. How sharply a neuron distinguishes its preferred orientation from others depends not just on the total synaptic input it receives, but on how that input is physically arranged across its dendrites. When synapses tuned to the same orientation cluster together on a dendritic branch, they can trigger local electrical events that amplify the signal, sharpening the neuron’s selectivity beyond what a simple summing of inputs would predict.⁶ The geometry of a single neuron’s wiring tree, in other words, matters for how precisely the cortex encodes what you see.

Motor cortex has been mapped since the 1870s, and the classic textbook picture shows a tidy map of the body draped across the cortical surface: face here, arm there, leg over there. The reality is messier. Within those broad zones, the representations of individual muscles or body parts overlap extensively. Rather than a neat point-to-point map, the motor cortex uses a distributed, overlapping code, where the same patch of cortex participates in controlling multiple nearby body parts.⁷ This overlap is probably what allows the motor cortex to coordinate complex movements that require multiple body parts to work together.

Working Memory and the Prefrontal Cortex

Working memory, the ability to hold and manipulate a piece of information over a few seconds, has long been associated with the prefrontal cortex. Neurons there show sustained activity during delay periods when an animal or person must remember something briefly. For years, this persistent firing was interpreted as the cortex literally storing the memory on-site. More recent work suggests the story is more nuanced. The prefrontal cortex may not be where the memory content is stored so much as the control center that keeps the memory alive in posterior sensory areas.⁸ Studies examining neural activity during working memory tasks have sometimes found stronger stimulus-related information in sensory areas than in the prefrontal cortex itself, suggesting the prefrontal cortex sends a top-down signal that keeps relevant representations active elsewhere.

That said, the prefrontal cortex clearly acts as a kind of executive allocator. It distributes limited memory resources across different tasks depending on demand.⁹ When you are trying to remember a phone number while also navigating a conversation, your prefrontal cortex is managing the tension between those competing demands. Damage to this region does not so much erase stored memories as impair the ability to juggle and control them.

Language, Spatial Awareness, and Multisensory Integration

Language processing involves a network of cortical regions connected by at least four major white-matter pathways. Two run along the top (dorsal) route linking temporal and frontal cortex: one supports the ability to repeat speech, while the other handles complex grammatical structure. Two more run along the bottom (ventral) route and handle meaning and basic sentence structure.¹⁰ Language is not a single faculty lodged in one spot but a distributed process that depends on specific highways between regions. Damage to different pathways produces different kinds of language impairment, which is why strokes affecting different parts of the brain can leave a person unable to find words, unable to understand grammar, or unable to repeat what they hear, each selectively.

The posterior parietal cortex handles a different integrative challenge: combining information from your eyes, your skin, your ears, and your sense of body position into a unified map of where things are. This area stitches together visual, somatosensory, auditory, and vestibular signals to represent the locations of objects relative to your body and within the surrounding environment.¹¹ When this region is disrupted, people lose the ability to remap visual and tactile information based on limb position, something that matters every time you reach for an object you can see but cannot yet feel.¹²

Predictive Processing

One influential framework for understanding cortical function treats the cortex as a prediction machine. Rather than passively receiving sensory data and building a picture from scratch, the cortex constantly generates predictions about what it expects to sense next and compares those predictions against incoming signals. Only the mismatches, the “prediction errors,” get propagated upward for further processing. This is called predictive coding, and it offers an elegant account of why the cortex has separate feedforward and feedback pathways operating at different speeds.¹³

High-resolution brain imaging has provided direct evidence for this scheme in the somatosensory cortex. When people receive predictable touches, the middle layer of the primary somatosensory cortex, the main recipient of incoming sensory signals, shows less activation than when touches are unpredictable. Meanwhile, the superficial and deep layers, which handle cortex-to-cortex communication, are more engaged during predictable sequences, consistent with the idea that they carry the predictive feedback signals.¹⁴ The layered architecture described earlier is not just an organizational quirk; different layers play different computational roles in this prediction-and-correction cycle.

What the Brain Does at Rest

Even when you are not doing anything in particular, the cortex is active. A set of regions including the medial prefrontal cortex, the posterior cingulate cortex, and the inferior parietal lobule form what is known as the default mode network. This network tends to become more active during rest and internal reflection and quiets down when you focus on an external task.¹⁵ There is even a reciprocal relationship: regions involved in demanding cognitive work show significant inverse correlations with the default mode network, as if the brain has a seesaw between outward attention and inward reflection.

The default mode network is not just an idle hum. Patterns of connectivity within it correlate with psychological states. A study of 148 healthy adults found that people who reported lower levels of happiness showed greater resting-state connectivity within the default mode network, and this stronger connectivity tracked with a tendency toward rumination, the repetitive, self-focused thinking that often accompanies unhappiness.¹⁶ The cortex’s baseline activity, in other words, is not neutral. It reflects something about the person’s habitual mental stance.

Plasticity and Rewiring

The cortex is not fixed once it matures. When part of the retina is damaged, the corresponding zone in the visual cortex loses its normal input. Within hours, axons from neighboring, unaffected cortex begin sprouting into the deprived zone. Older connections get pruned, new ones grow, and over weeks the density of connections from intact cortex into the affected area steadily increases. This rewiring provides a route for visual signals to propagate into territory that would otherwise go silent.¹⁷ Similar turnover of both axonal branches and dendritic spines accelerates during learning, not just after injury. The cortex remodels its wiring constantly, and the rate of remodeling ramps up when circumstances demand it.

Chemical Signals That Set the Cortex’s Operating State

The cortex’s moment-to-moment behavior depends heavily on chemical modulators originating from small clusters of neurons deep in the brainstem and basal forebrain. These neuromodulatory systems release substances like norepinephrine, acetylcholine, serotonin, and dopamine broadly across the cortex, shifting its overall operating state rather than transmitting specific sensory information. Norepinephrine, for example, provides a signal that links transitions in behavioral state to changes in cortical network interactions.¹⁸ When you go from drowsy to alert, it is partly because norepinephrine levels have shifted across the cortex, changing how readily neurons fire and how strongly they communicate.

These same subcortical circuits control the sleep-wake cycle, and when they malfunction, the cognitive consequences are broad: impaired attention, disrupted arousal, poor decision-making.¹⁹ The cortex, for all its sophisticated circuitry, depends on these relatively simple chemical broadcast systems to function properly. Neuromodulatory nuclei in the brainstem and basal forebrain are key drivers of widespread cortical activity and communication, and vigilance, the continuously fluctuating state of cortical activation that influences everything from reaction time to complex reasoning, is regulated through these centers.²⁰

Blood Supply and Energy Demands

The cortex is an energy-hungry organ, consuming a disproportionate share of the body’s glucose and oxygen. To meet this demand, cerebral blood flow is dynamically regulated by neural activity itself, a process called neurovascular coupling.²¹ When a patch of cortex becomes active, local blood vessels dilate within seconds, delivering more oxygenated blood precisely where it is needed. This tight regional and temporal linkage between neural firing and blood flow is what makes functional brain imaging possible: techniques like fMRI detect the blood-flow changes that accompany neural activity.²² When neurovascular coupling breaks down, as it can in aging, hypertension, or neurodegenerative disease, the cortex loses its ability to fuel active regions on demand, and cognitive performance suffers even if the neurons themselves are still intact.

The Thalamic Relay

The cortex does not operate in isolation from subcortical structures. Nearly all sensory information passes through the thalamus before reaching the cortex, and the cortex sends extensive projections back to the thalamus, creating loops rather than a simple one-way pipeline. These thalamocortical circuits act as a kind of signal translator. Sensory organs often encode information in the timing of their signals, while the cortex tends to represent information through firing rates. Thalamocortical loops function as oscillation-based systems that convert temporally coded sensory information into rate-coded cortical signals, bridging the two coding schemes.²³ This translation happens across senses, and the same loop architecture allows information from different sensory and motor channels to be integrated.

When Cortical Function Breaks Down

Damage to the cortex produces remarkably specific deficits depending on what region is affected. In visual agnosia, damage to temporal and parietal cortex can leave basic vision intact while destroying the ability to recognize objects. One well-studied patient showed normal overall responsiveness to visual stimulation in occipital cortex, but significantly reduced object-related activity in temporal and parietal regions.²⁴ In a larger study of patients with posterior cerebral artery strokes, about two-thirds showed impairments on visual recognition tasks, affecting all categories of stimuli tested.²⁵ You can see perfectly well but be unable to tell a cup from a shoe.

Stroke damage also disrupts the large-scale functional networks the cortex depends on. Modeling and experimental data show that regions most closely connected to the damaged tissue experience the most severe drops in functional connectivity, while more distant regions may be spared.²⁶ This pattern helps explain why a small stroke can sometimes cause widespread cognitive problems: the damage propagates through the network topology, degrading communication in regions that were not themselves injured.

At the cellular level, developmental disorders like schizophrenia and autism have been linked to problems with how excitatory synapses form, function, and are maintained. Research using neurons derived from patient stem cells supports the idea that disrupted synapse development is a shared biological substrate across both conditions.²⁷ The excitatory-inhibitory balance that keeps cortical networks stable may be thrown off from early development, producing cascading effects on perception, social cognition, and thought.

Reading Cortical Signals With Brain-Computer Interfaces

The fact that cortical neurons encode specific information in their firing patterns has opened the door to brain-computer interfaces. By recording from populations of neurons in the motor cortex and decoding their activity, researchers can extract signals representing intended movements. Motor cortex neurons have “preferred directions,” firing most vigorously when a movement is planned in a particular direction. Algorithms that sum each neuron’s preferred direction scaled by its firing rate can reconstruct the intended trajectory of a hand movement, and this approach has been used to allow a person with paralysis to control a robotic arm.²⁸

Beyond motor signals, cognitive parameters like the goal of an action and the expected value of a choice can also be decoded from cortical activity.²⁹ This suggests that brain-computer interfaces could eventually go beyond restoring movement to enabling paralyzed individuals to communicate intentions and decisions directly. The practical applications remain limited by recording stability, signal quality, and the sheer complexity of cortical coding, but the underlying principle is clear: the cortex’s information is there to be read, if you can get close enough to the neurons and interpret their collective language.

Why the Human Cortex Folds

One of the most visible features of the human cortex is its extensive folding. The grooves and ridges that give the brain its wrinkled appearance are not random; they reflect evolutionary pressures that favored packing more cortical surface area into a skull constrained by the size of the birth canal. Both brain volume and the degree of cortical folding have increased over primate evolution, and they are strongly correlated across species. Humans are exceptional in both measures. But within a species, the genetic relationship between volume and folding turns out to be unexpectedly negative: genes that promote larger brains tend to be associated with less folding, and vice versa.³⁰

This finding suggests that at least two separate evolutionary pressures were at work. One favored increasing brain volume, while a second, possibly related to the constraints of birth and neonatal head size, independently favored more folding. Without that second pressure, natural selection for bigger brains would have been expected to produce smoother, less folded cortices. The folds, then, are not a simple byproduct of size but an independently selected feature, likely because more folding means more cortical surface area and more room for the layered, columnar circuitry that underlies cortical function.