Neuroscience curriculum

Dendrites receive most synaptic input, the soma holds the nucleus and pools that input, and the axon carries the output away to another neuron, a muscle cell or other cells. Firing usually starts just past the axon hillock, in the axon initial segment, so a distant input counts only through the voltage change it causes there.

Afferent fibers carry signals toward the central nervous system and efferent fibers carry commands out of it, so the label names direction, not content. Sensory neurons are afferent and motor neurons efferent, and every spinal nerve is mixed, carrying both, so a single nerve usually cannot be called just one or the other.

Sensory neurons bring information in from sense organs such as the eyes and ears, motor neurons carry commands out to muscles, and interneurons connect the two, lying entirely inside the brain and spinal cord. The three labels are broad: within them are hundreds of types that send and receive messages in different ways.

Structural classes count the processes leaving the cell body: a multipolar neuron has more than two, a bipolar neuron two. Dorsal root ganglion sensory neurons end up pseudo-unipolar, with one stem that splits into a branch toward the body and a branch toward the spinal cord, and they start out bipolar before reshaping during development.

The membrane at rest is far more permeable to potassium than to sodium, so its voltage is pulled toward potassium's equilibrium value. A small sodium leak keeps it a little less negative than that value, which is why the resting potential sits near, not at, the potassium figure, and why cells with different leak channels rest at different voltages.

Potassium is concentrated inside the cell and sodium outside, and the sodium-potassium pump spends energy to keep it that way. The gradients are what an action potential spends: opening sodium channels lets sodium rush in, and opening potassium channels lets potassium flow out.

The pump spends cellular energy to push sodium out and potassium in. It maintains the gradients rather than producing signals itself, so the membrane potential depends on constant work and the system is held far from true equilibrium; after spikes, the pump gradually restores the original ion distribution.

At rest, potassium leak channels such as the two-pore-domain family stay open, so the membrane is mostly permeable to potassium and its voltage sits near potassium's equilibrium value. A sodium leak channel nudges it less negative. In one fruit-fly muscle study, adding extra potassium leak channels made the voltage harder to shift, likely by holding it closer to that value.

An ion's equilibrium potential is the membrane voltage that exactly balances the push of that ion's concentration gradient, so the ion has no net flow, and the Nernst equation calculates it for one ion. A real membrane is permeable to several ions at once, so the Goldman equation weights each by permeability, and the resting potential matches no single ion's value.

A graded potential scales with its input, from about 1 to 50 millivolts, and can last up to seconds; synaptic and receptor potentials are both kinds. An action potential differs on every count: roughly 100 millivolts whatever the stimulus, and over in 2 to 4 milliseconds.

Threshold, typically around minus 55 millivolts, is the point where depolarization opens enough voltage-gated sodium channels that the sodium flowing in depolarizes the membrane further on its own. Below threshold, inputs stay graded and can add up; once it is crossed, that feedback carries the spike to completion.

Once threshold is crossed, the action potential reaches its full size, about 100 millivolts, however strong the stimulus was. A bigger push never makes a bigger spike, so stimulus strength has to be carried in another way, such as how often the neuron fires.

In the rising phase, voltage-gated sodium channels open and sodium flows in, briefly reversing the membrane's polarity so the inside turns positive. The phase lasts about a millisecond, and during it the sodium channels inactivate, which helps end the rise.

During repolarization, voltage-gated potassium channels open and potassium flows out, driving the voltage back toward rest, while the sodium channels stay inactivated. Because they are inactivated through this phase, no second spike can start yet, however strong the input.

The outward potassium flow continues briefly past the resting level, so the membrane undershoots and becomes more negative than usual. During this undershoot a neuron is less likely to fire, the relative refractory period. An inhibitory synapse can also hyperpolarize a cell, but that comes from a transmitter, not from the tail of a spike.

Voltage-gated channels open or close when the membrane voltage changes, unlike transmitter-gated channels, which open when a transmitter binds. In a spike, sodium channels open at threshold and then inactivate, potassium channels open to repolarize, and at the axon terminal voltage-gated calcium channels let in the calcium that triggers release.

For a brief moment after a spike starts, no second action potential can fire at any stimulus strength, because the sodium channels are inactivated. This caps the firing rate: specialized interneurons approach about 500 spikes per second, while most neurons fire between 1 and 100 times per second.

After the absolute period, a neuron can fire again, but only to a stronger-than-usual stimulus, because the membrane is still hyperpolarized by lingering potassium outflow. The two periods differ in cause as well as strength: the absolute one comes from inactivated sodium channels, the relative one from that potassium outflow, which a big enough input can overcome.

The stretch of axon a spike has just passed is refractory, so the spike regenerates in the stretch ahead of it and cannot turn back down the axon. Refractoriness also keeps spikes discrete, so successive signals cannot overlap and merge. It does not make every part of a neuron one-way: spikes starting at the initial segment can also spread back toward the cell body.

Because every spike has the same size, one way a neuron can signal a stronger input is by firing more often, within the ceiling the refractory period sets. Muscle control shows the principle: a motor unit fires faster as its drive increases, and a muscle's force rises both with that rate and with how many units are recruited.

A spike travels by regenerating: depolarization at one patch of axon brings the next patch to threshold, where voltage-gated channels produce a full-size spike again. In an unmyelinated axon those sodium channels are spread evenly along the membrane, which permits only slow propagation.

Glial cells wrap myelin around axons, and the cell type doing it differs between the central nervous system and peripheral nerves. Myelin insulates the axon and lowers its capacitance, which speeds conduction and cuts the energy each spike costs. In multiple sclerosis the immune system attacks myelin in the brain and spinal cord.

Nodes of Ranvier are short gaps, usually about a micrometer long, between myelinated segments, and they are rich in voltage-gated sodium channels, so the spike is regenerated only there. Node size matters in a way that depends on the channels: models predict a longer node slows conduction if channel number stays fixed, and the opposite if channel density stays fixed.

In a myelinated axon the spike is regenerated only at the nodes of Ranvier, so it appears to jump from node to node, which greatly raises speed in an energy-efficient way. How much myelin itself contributes is still debated: some reports suggest that clustering sodium channels may be enough to raise conduction speed.

Axon diameter and myelination set conduction speed. A wider axon has less internal resistance, so in myelinated axons speed rises in proportion to diameter, and large myelinated touch fibers exceed 30 meters per second. Unmyelinated C-tactile fibers, which respond to gentle stroking, conduct about 50 times more slowly.

At a chemical synapse an electrical signal is turned into a released transmitter and then back into an electrical change in the next cell. Unlike an electrical synapse, it runs one way, from presynaptic terminal to postsynaptic cell, and what it does depends on the receptors the receiving cell carries.

When a spike reaches the axon terminal it opens voltage-gated calcium channels, and the rise in calcium inside the terminal triggers transmitter release. Calcium is the link from the electrical signal to the chemical one. How soon release follows generally shifts with the shape of the incoming spike and with the terminal's release probability.

Transmitter is stored in tiny sacs called vesicles, and the signal reaching the end of the axon makes them fuse with the presynaptic membrane and empty into the synaptic cleft. Not all release works this way: amphetamines drain vesicles into the axon, and the transmitter then escapes passively or through a reversed membrane transporter.

Transmitter crosses the cleft and binds receptors on the receiving cell, often compared to a key in a lock, and the receptor then changes that cell. The effect belongs to the receptor more than to the transmitter: glutamate acts on both channel receptors and G-protein receptors, and dopamine can excite or inhibit.

A transmitter's signal ends when it is cleared from the cleft. Transporters carry many transmitters back into the neuron that released them, and enzymes break others down, as with acetylcholine. Blocking either route prolongs the signal: cocaine interferes with transporters, and some myasthenia gravis medicines slow acetylcholine's breakdown.

Chemical transmission adds a short delay between a presynaptic spike and the postsynaptic response, because calcium must enter and transmitter must be released and diffuse across the gap. The delay is not fixed: it generally shifts with release probability and spike shape. One study of cortical pyramidal synapses saw release probability rise with no change in delay, suggesting the two effects can cancel.

At an electrical synapse, gap junction pores join two neurons so current flows directly between them, usually in either direction, with no transmitter step. They are best known for synchronizing coupled neurons, and because current crosses two membranes they pass slow signals more readily than brief spikes.

An excitatory postsynaptic potential is a small, graded depolarization that moves the membrane toward threshold, usually produced when a transmitter opens channel receptors that let positive ions such as sodium in. A neuron takes input from thousands of others, so whether it fires depends on the sum of its EPSPs and IPSPs, not on any one input alone.

An inhibitory postsynaptic potential makes firing less likely, often by opening receptor channels that let chloride in and hyperpolarize the cell, as GABA-A and glycine receptors do, and it can cancel an EPSP arriving at the same time. Hyperpolarizing is not required: where a cell's internal chloride is high, such channels can depolarize it yet still reduce excitability by shunting current.

Spatial summation adds inputs that arrive at the same time from different synapses on one neuron. Each EPSP alone is small, but several together can carry the membrane to threshold, and IPSPs arriving at the same moment subtract from that total.

Temporal summation adds inputs that arrive in quick succession, often from one synapse firing repeatedly, before the earlier ones have faded. A rapid burst of EPSPs is more likely to trigger a spike than the same inputs spread out. Spatial summation is the counterpart: inputs share a moment but arrive at different places.

Synaptic inputs spread from the dendrites and soma to the axon initial segment, the unmyelinated stretch between the axon hillock and the first myelin, which is the main site where summed input becomes a spike. Its position, length and ion-channel content can change with activity, retuning how excitable the neuron is.

An ionotropic receptor is itself an ion channel that opens when its transmitter binds, so the response is fast. Glutamate's AMPA, NMDA and kainate receptors, GABA-A receptors and glycine receptors all work this way, and channels like these produce most postsynaptic potentials.

A metabotropic receptor is not a channel; it is a G-protein-coupled receptor that sets off chemical steps inside the cell. The same transmitter can use both kinds: glutamate has metabotropic mGluRs and GABA has GABA-B alongside their channel receptors, and many neuromodulators, such as the opioids, work over longer time scales through receptors like these.

A second messenger carries a receptor's signal inside the cell. Some metabotropic glutamate receptors act through G-proteins that inhibit adenylate cyclase, lowering cyclic AMP and so the activity of protein kinase A. In photoreceptors a cascade of this kind amplifies the signal strongly, which is part of why very dim light can be detected.

Glutamate is the main excitatory transmitter, acting on fast ionotropic receptors such as AMPA and NMDA and on slower metabotropic receptors. Its NMDA receptors also let calcium in, which makes glutamate central to synaptic plasticity. Too much glutamate can damage or kill neurons.

GABA is the main inhibitory transmitter in the brain, acting through ionotropic GABA-A and metabotropic GABA-B receptors and making firing less likely. It also helps control muscle activity, and drugs that raise GABA levels are used to treat epileptic seizures, which fits its role in restraining excitation.

Glycine receptors are chloride channels that carry fast inhibition, concentrated in the spinal cord, brainstem, cerebellum and retina. They are blocked by strychnine but not by bicuculline, which blocks GABA-A receptors, and that difference is how researchers tell the two inhibitory systems apart.

At the neuromuscular junction, nerve endings release acetylcholine, which binds receptors on the muscle and triggers contraction; in myasthenia gravis, antibodies block or destroy those receptors. In the brain acetylcholine also works as a neuromodulator, often released diffusely rather than at a single synapse.

Scientists now think dopamine has more to do with getting you to repeat rewarding behavior than with producing pleasure directly. Its bursts of firing are widely read as a reward prediction error, the gap between expected and received reward, but that is contested: one mouse study found movement signals it does not explain. Losing dopamine in some brain areas leads to the rigidity of Parkinson's disease.

Serotonin works as a neuromodulator and can act by volume transmission, released diffusely to reach targets some distance away. It is involved in sleep and temperature regulation and constricts blood vessels. Low levels are often linked to depression, but one umbrella review found no consistent evidence for that account, so the link remains disputed.

In the central nervous system, norepinephrine, also called noradrenaline, comes mainly from the locus coeruleus, whose projections reach the cortex, thalamus, cerebellum and spinal cord and regulate arousal, attention and stress responses. In the body, sympathetic nerves also drive the adrenal medulla to release it with epinephrine under stress.

Beta-endorphin is one of the body's own opioid peptides, which act on four types of G-protein-coupled opioid receptor; it binds the mu receptor with high affinity. Opioid signaling is a foundation of the brain's descending pain control and also shapes reward, which is why opioid drugs, which mimic these natural chemicals, act on both.

An agonist activates a receptor as the natural transmitter does; an antagonist binds the receptor and blocks it without activating it. Naloxone, an opioid antagonist, shows the difference: it reverses opioid effects but does nothing in someone with no opioids present. Strychnine is another antagonist, acting at glycine receptors.

Neuromodulation adjusts how neurons and synapses respond to input rather than passing a specific message, usually over longer time scales and across many neurons at once. Serotonin, dopamine and norepinephrine are classic examples, and they often act by volume transmission, diffusing to targets well away from their release site.

The central nervous system is the brain and spinal cord, encased in bone and surrounded by protective membranes and fluid; the peripheral nervous system is the nerves and ganglia outside it. The divide matters after injury: peripheral axons can regrow, while axons in the spinal cord largely cannot.

Gray matter holds cell bodies and unmyelinated fibers; white matter is bundles of myelinated axons, whose myelin gives it its whitish look. The cortex looks gray because its nerve cells largely lack that insulation. Multiple sclerosis attacks myelin, yet it also damages cell bodies in gray matter, so it is not purely a white matter disease.

Directional terms describe where one structure lies relative to another. Caudal means toward the tail end and rostral names the opposite end of that axis; dorsal means toward the back and ventral toward the front, medial toward the midline and lateral away from it. In the brain they follow the neural axis, which bends almost 180 degrees near the top of the brainstem.

The cerebral cortex is a layer of gray matter about as thick as two or three stacked dimes that coats the cerebrum, and most of the brain's information processing happens there. Each hemisphere's cortex is divided into lobes, four of them named for the skull bones that cover them.

The ridges of the cortex are gyri and the grooves between them sulci. Folding adds surface area, and so gray matter, inside the same skull. When development goes wrong the cortex can form without folds, a condition called lissencephaly, also known as agyria.

The frontal lobes, behind the forehead, do much of the work of planning, imagining the future and reasoning, partly by holding one idea in mind while others are weighed. Their back edge holds the motor cortex, so frontal damage that disrupts speech production can come with weakness on the right side of the body.

The parietal lobes, behind the frontal lobes, begin with the somatosensory cortex, which receives touch, temperature and movement information from the body. They also support reading and arithmetic, and a visual pathway projects to the posterior parietal cortex to guide visually directed action.

The top of each temporal lobe receives information from the ears, and its underside is crucial for forming and retrieving memories. The temporal lobe also supports understanding speech, so damage there can leave someone speaking fluently in long sentences that carry little meaning.

The occipital lobes at the back of the brain process images from the eyes and link them with images stored in memory, and damage to them can cause blindness. From primary visual cortex, information is commonly described as splitting into a ventral pathway for recognizing objects and a dorsal pathway toward parietal cortex for guiding action, though that split is debated.

The primary motor cortex, just ahead of the somatosensory cortex, helps plan, control and execute voluntary movement, and its body map is laid out along the precentral cortex. Territory follows fine motor control rather than body size, and each hemisphere mainly controls the opposite side of the body.

The somatosensory cortex lies just behind the motor areas and receives touch, temperature and movement information from the body. Its map is distorted: regions dense in touch receptors, such as the fingers and the area around the mouth, get disproportionately large territory.

The homunculus is the body map drawn across the sensory and motor cortex, in which territory follows receptor density or fine control rather than body size, so the fingers and the area around the mouth loom large. Penfield's motor version is a caricatured but useful summary, and one surgical mapping study proposes additions to it.

Broca's area, in the frontal lobe, is tied to producing speech. Damage affecting this region can cause Broca's aphasia: short, effortful phrases that drop small words like “is” and “the”. Understanding is usually better than in Wernicke's aphasia, though not untouched, and people are aware of their difficulty, which often frustrates them.

Wernicke's area, in the temporal lobe, supports understanding language. Damage affecting this region can produce Wernicke's aphasia, roughly the reverse of Broca's: fluent, long sentences with little meaning or invented words, poor comprehension of spoken, written or signed language, and often no awareness of the mistakes.

Association cortex is the higher-order territory beyond single-sense and motor areas. Its activity holds information over longer windows than sensorimotor cortex does, a gradient seen across species and recording methods. One brain-imaging study found that gradient still developing through youth before stabilizing in adulthood.

The prefrontal cortex, behind the forehead, supports planning, prioritizing, problem solving and self-control over impulses. It is one of the last parts of the brain to mature, with development continuing into the mid-to-late 20s, which is one proposed reason teens may take more risks when social benefits seem high.

The limbic system is a grouping whose membership has been redrawn over time. Broca first counted mainly olfactory structures, Papez described a circuit through the hippocampus, cingulate cortex, hypothalamus and thalamus believed to control memory and emotion, and later researchers added the amygdala and prefrontal regions.

The amygdala, an almond-shaped structure, plays a central role in detecting threat-related stimuli and in learning and expressing conditioned fear. A related circuit, the extended amygdala, contributes to the anxiety and irritability of drug withdrawal.

NIA describes the hippocampus as a memory indexer that sends new memories to parts of the cortex for long-term storage and retrieves them when needed, though how that transfer works is debated. With the entorhinal cortex, it is usually among the first areas Alzheimer's disease damages, before the disease reaches cortical areas for language and reasoning.

The hypothalamus, a small region below the thalamus, maintains homeostasis by regulating visceral activities, and it connects the nervous system to the hormonal one, for example by sending nerve impulses that drive the adrenal medulla. One of its nuclei receives light information directly from the eyes and helps set daily sleep-wake rhythms.

The thalamus passes sensory signals on to the cerebral cortex, but it is not a passive relay. In the visual thalamus, feedback from cortex and arousal signals change how retinal input is transmitted, so the same input from the eye need not produce the same output to the cortex.

The basal ganglia, clusters of nerve cells around the thalamus, help initiate and integrate movement and help form habits and routines. They are also a key node of the reward circuit, so the same structures matter both in Parkinson's disease and in the compulsive drug use that marks addiction.

The cerebellum, the second-largest part of the brain, coordinates and regulates movement and is involved in learned movements, such as playing the piano or hitting a tennis ball. Most disorders that cause ataxia, a loss of coordination, do so by breaking down cerebellar cells, and many people with ataxia have damage to its Purkinje cells.

The brainstem, made of the midbrain, pons and medulla, joins the brain to the spinal cord, and every ascending and descending tract between them passes through the medulla. It controls heart rate and breathing, which is why opioid overdose, by disrupting it, can depress breathing and cause death.

The reticular formation is a network of nuclei running through the brainstem, home to norepinephrine-producing neurons among others. Part of it feeds the ascending reticular activating system, which links it to the thalamus, hypothalamus and cortex and is considered a main structure for arousal and consciousness.

The corpus callosum is a thick band of nerve fibers at the base of the deep fissure between the hemispheres, and it is the main route by which they communicate. It can be partly or completely missing from birth, called agenesis of the corpus callosum, with effects ranging from mild to severe.

Some functions lean to one hemisphere: in most people the language areas sit mainly on one side, and attention networks lean to the other. But one study of resting brain scans from 1,011 people found that lateralization is local to particular networks, with no sign of people being left-brained or right-brained overall.

Three meninges wrap the brain and spinal cord: the tough outer dura mater, the cobweb-like arachnoid, and the delicate pia mater bound to the brain's surface. Cerebrospinal fluid fills the subarachnoid space beneath the arachnoid, so membranes and fluid protect the tissue together.

The ventricles are interconnected fluid-filled cavities inside the brain. Cerebrospinal fluid flows through them and around the brain and spinal cord, cushioning and nourishing the tissue and removing waste, then is reabsorbed into the blood through arachnoid villi. A blockage along that route can widen the ventricles.

The blood-brain barrier is formed by brain capillary cells sealed together by tight junctions and supported by pericytes and astrocyte end-feet. It shields the brain from circulating toxins and pathogens, but the same selectivity keeps an estimated 98 percent of small-molecule drugs out, a major obstacle for treatment.

Transduction turns physical energy into an electrical signal. In the ear, bending of a hair cell's stereocilia opens channels at their tips; in the eye, light starts a chemical cascade that closes sodium channels, so photoreceptors respond to light by becoming more negative and releasing less transmitter, not by firing.

Receptors are grouped by what they detect: mechanoreceptors respond to deformation, chemoreceptors to molecules, photoreceptors to light, thermoreceptors to temperature and nociceptors to damaging stimuli. The lines blur: the TRPV1 channel responds to noxious heat, acid and capsaicin, which is why chili peppers burn.

A neuron's receptive field is the region where a stimulus changes its activity, such as a patch of skin for a touch fiber. Small fields give fine spatial detail: touch units near the skin surface have small fields, while deeper Pacinian and Ruffini units have large fields and lower precision.

Under constant stimulation many receptors reduce their response, and they differ in how fast. In the skin, fast-adapting units tied to Meissner and Pacinian corpuscles respond mainly when contact starts and ends, while slow-adapting units tied to Merkel complexes and Ruffini endings stay active for as long as contact lasts.

The labeled-line idea, an example of Muller's law of specific nerve energies, holds that a sensation's quality depends on which pathway carries it. In taste it was long disputed, with some researchers arguing that quality is read from patterns across many fibers, and one view holds that labeled lines and across-fiber patterns both operate.

Photoreceptor signals pass through bipolar cells to retinal ganglion cells and leave the eye through the optic nerve. In the lateral geniculate nucleus of the thalamus, each relay cell's main drive comes from only a few ganglion cells, and it passes the signal on to primary visual cortex, which sends feedback to that relay.

Rods and cones capture light, with cones handling color; their signal passes to bipolar cells and then to retinal ganglion cells. On this route the ganglion cells are the first to fire action potentials, since the layers before them use graded voltage changes. Lacking a cone pigment causes color blindness without affecting rod vision.

Many neurons in primary visual cortex respond best to edges at a particular orientation. Hubel and Wiesel found that cells along a vertical path through cat visual cortex shared an orientation preference, the basis of the orientation-column model, though denser recordings later showed preference changing gradually across the surface.

Beyond primary visual cortex, a ventral stream supports recognizing objects and a dorsal stream, projecting to posterior parietal cortex, supports visually guided action. The split is useful but debated: recent evidence, including one brain-stimulation study, suggests the dorsal stream also helps recognize object shape.

Hair cells pass sound signals to spiral ganglion neurons, which carry them to the cochlear nucleus in the brainstem; from there they climb through the inferior colliculus and auditory thalamus to auditory cortex. One mouse study found simple frequency tuning already complete in the cochlear nucleus, with tuning to complex sounds improving toward cortex.

Tonotopy maps frequency to place: hair cells at the base of the cochlea respond to high pitches and those at the apex to low ones. That layout is preserved along the whole ascending auditory pathway, from the cochlea to auditory cortex, much as touch keeps a body map in somatosensory cortex.

The vestibular system in the inner ear has two parts. Three fluid-filled semicircular canals, set roughly at right angles, sense head rotation, while the utricle and saccule use dense calcium carbonate grains to sense head position relative to gravity. When grains drift into a canal, the result is brief positional vertigo.

Each olfactory neuron carries one odor receptor and connects directly to the brain, while taste cells send their signals through three taste nerves. Flavor combines both, and losing smell costs much of it: chewing releases aromas that reach the nose from the throat, which is why a blocked nose makes food taste bland.

Somatosensation covers touch, pressure, vibration, temperature, pain and proprioception, the sense of body position and movement drawn from muscle spindles, tendon organs, and joint and skin receptors. Fast myelinated fibers carry discriminative touch, while slow unmyelinated fibers respond to gentle stroking and are tied to touch's pleasant, social side.

Nociception is the detection of tissue-damaging stimuli by nociceptors, many of them free nerve endings on A-delta and C fibers, with channels such as TRPV1 responding to noxious heat and irritants. Pain is broader: neuropathic pain comes from nerve damage rather than from tissue damage, and the brain systems for pain overlap with those for emotion, attention and memory.

Pain signals are adjusted on their way up rather than passed through unchanged. Gate control theory first framed spinal pain processing as modifiable, and descending pathways from the periaqueductal gray, rostral ventromedial medulla and locus coeruleus can both inhibit and facilitate it; in chronic pain the balance often shifts toward facilitation.

Referred pain is felt next to, or even at a distance from, its real source. A likely explanation, though not a settled one, is that sensory fibers from the same spinal cord segment supply several organs and their signals converge. Pain from an inflamed pancreas radiating to the back is one example.

A reflex arc runs from a sensory receptor through a sensory neuron into the central nervous system, often through interneurons, to a motor neuron and its effector. Because it need not wait for the brain, a withdrawal reflex pulls a body part away before you are aware of the pain; the tendon-jerk reflex skips the interneuron, with spindle sensory fibers synapsing directly on motor neurons.

Upper motor neurons in the brain send signals to lower motor neurons in the brainstem and spinal cord, which drive the muscles. Damage to each looks different: losing lower motor neurons brings weakness, wasting and twitching, while losing upper motor neuron input brings stiffness and overactive reflexes.

A motor unit pairs a single motor neuron with every muscle fiber it innervates, and it is the level at which the nervous system controls muscle force. Units are generally recruited in order from weakest to strongest, the size principle, and once recruited each one fires faster as its drive increases.

The somatic system drives skeletal muscle and is under conscious control; the autonomic system drives cardiac muscle, smooth muscle and glands without conscious effort. Both are efferent divisions, and the autonomic one splits again into sympathetic and parasympathetic branches, with many organs supplied by both.

The sympathetic division mobilizes the body under stress, for example by driving the adrenal medulla to release epinephrine and norepinephrine. It is not simply the parasympathetic division's opposite: one study read heart-rate changes in young women drinking cold water as both branches activating together to limit heat loss and keep heart rate in range.

The vagus nerve, a main parasympathetic route, slows the heart and controls the muscles of the stomach and small intestine. When diabetes damages the vagus, the stomach may empty too slowly, which shows how much routine digestion depends on this branch. Many organs also receive sympathetic fibers, so the two divisions often share targets.

The enteric nervous system is a network of neurons and glia in the gut wall, mostly in the myenteric and submucosal plexuses. When food stretches the gut wall, its own sensory neurons trigger local reflexes that speed or slow movement and secretion without input from the brain or spinal cord, though signals also flow between gut and brain.

Glia support neurons rather than carrying impulses: oligodendrocytes make myelin in the brain and spinal cord and Schwann cells in peripheral nerves, astrocytes handle metabolic, structural and protective work and wrap brain blood vessels, and microglia clear debris and foreign material. Glia can divide, and in some brain regions they far outnumber neurons.

Synaptic plasticity means a connection can become stronger or weaker depending on how it is used. Timing matters because a cell that fires just before another may be helping to cause it. The classic account at excitatory synapses is that the amount of calcium entering the receiving cell sets the direction, large for strengthening and smaller for weakening, though several studies now challenge that simple picture.

Hebb's idea is that connections strengthen between cells that are active together. Taken alone, the rule is a runaway loop: stronger synapses raise joint firing, which strengthens them further. One review argues that homeostatic plasticity as measured in experiments is too slow to stop this in current models, so rapid compensatory processes must work alongside it.

Long-term potentiation (LTP) was first described in the hippocampal formation and is usually induced in the lab with brief bursts of high-frequency stimulation. One review argues that what looks like one effect is a family: a short-term potentiation that decays, plus longer-lasting forms that overlap in time. So a single LTP measurement can mix processes with different lifetimes.

Two events must coincide for this channel to pass current: transmitter bound at its site, and a receiving cell already depolarized enough to clear a magnesium block. Because neither is enough alone, it acts as a coincidence detector, and the calcium it admits is needed for many forms of plasticity. In one well-studied form, the lasting change is a rise in AMPA receptor sensitivity.

Long-term depression (LTD) is a lasting drop in synaptic strength, the counterpart of LTP. In timing experiments it follows when the receiving cell fires just before the input arrives, a pairing that lets in less calcium. LTD is not one mechanism: some forms depend on presynaptic NMDA sites, while others rely on metabotropic or cannabinoid signaling instead.

In spike-timing-dependent plasticity, order sets the direction: input arriving just before the cell fires tends to strengthen the synapse, while firing just before the input tends to weaken it. A 2012 overview warned that most studies were done in a dish, so the rule should not be assumed to run the same way in an intact brain, and neuromodulators can gate it.

Plasticity is also physical. Dendritic spines, the small protrusions that carry most excitatory synapses, form and disappear: animals kept in enriched versus impoverished surroundings differ sharply in spine density, and sensory deprivation changes how fast spines turn over. Even so, in adult mice most spines persist for long periods, so the bulk of wiring is stable.

Young brains, including human ones, carry many more dendritic spines than adult brains. In mice, the drop between adolescence and adulthood comes mainly from removing existing synapses, called pruning, rather than from making fewer. Pruning tracks experience: trimming young mice's whiskers slowed spine elimination in the matching cortex and left a higher net spine density.

A critical period is a window when experience reshapes a circuit unusually strongly. Blocking vision in one eye early in life causes lasting loss of acuity (amblyopia), yet the same deprivation in an adult does not. The window closes partly through molecular brakes, such as perineuronal nets, and recovery from amblyopia in adulthood is limited across species.

Circuits keep adjusting to what an individual actually encounters, well past any critical window. In adult mice, whisker trimming removed some long-lasting spines and formed new ones, and older people can still learn new skills and form new memories. The capacity is real but bounded, since most adult spines stay stable over long periods.

After arm deafferentation in monkeys, the face representation in sensory cortex takes over the territory of the missing hand. In people, such remapping was long blamed for phantom pain. One imaging study of upper-limb amputees found only small shifts of the lip area toward, not into, the hand area, and no link to pain, so remapping's role in symptoms is contested.

Most of your neurons are already created by the time you are born, but adult brains add some new ones, classically in two main niches. How much of this continues in adult humans, especially at older ages, is still disputed, as are reports of niches in other regions, so findings from other species do not settle the question for people.

Memory for facts and events is the kind you can report, and it is supported mostly by medial temporal structures. A classic case shows the split from other kinds of memory: an amnesic patient pricked by a pin hidden in a handshake later refused to shake hands yet could not recall being stuck. She had learned the association without storing the episode.

A motor skill or habit shows up in what you can do rather than what you can report. This kind of memory belongs to the non-declarative family, alongside priming and conditioning, and skills and habits depend largely on the striatum rather than on medial temporal structures. The split is a working distinction rather than a wall: one review notes that an intrusive aversive memory could, in principle, stem from both systems.

To treat severe seizures, a surgeon removed medial temporal lobe structures on both sides of H.M.'s brain. His seizures eased, but he was left with severe, lasting anterograde amnesia: he could hold information for about 30 seconds yet could not turn it into long-term memory, while knowledge from before surgery stayed largely intact. One lesion separated those memory processes.

Holding information over seconds was long tied to persistent delay-period firing in prefrontal neurons, which survives distraction and whose disruption raises errors. Later studies often decoded the remembered item from posterior sensory areas but not from prefrontal activity, so one view treats prefrontal delay activity as a top-down control signal rather than the store itself.

A new memory starts out easy to disturb and becomes stable through consolidation. The first stage, cellular or synaptic consolidation, is a cascade of events inside neurons that ends within hours at most and depends on new gene expression. Until it finishes, the memory is easier to disrupt, which is why recent memories are more vulnerable than older, consolidated ones.

The standard model says the hippocampus first binds an event's scattered cortical pieces, then over time strengthens their links until the cortex can retrieve the memory alone; in rodents this takes weeks. The model is contested. One critical review argues its premises fall short, and that detailed, relational memories keep needing the hippocampus however old they are.

Retrieving a memory can reopen it to disruption, and it must then re-stabilize through gene expression much as a new memory does; blocking that step in animals disrupts the memory. Retrieval does not always reopen a trace, though: in rodents, older or very strong fear memories needed longer reminders, and a much longer re-exposure produces new learning that reduces fear rather than erasing it.

Memory consolidation most likely needs both non-REM and REM sleep. In one nap study, auditory cue words tied to learned pairs set off fast sleep rhythms, the type of material could be decoded during those rhythms, and recall later improved. The authors call that evidence correlative, so it does not show that the rhythms themselves do the consolidating.

Dopamine neurons in the ventral tegmental area project to the nucleus accumbens in the ventral striatum. Dopamine was once thought to create pleasure directly; it now looks more like a signal that makes you repeat pleasurable acts. Drugs overdrive this pathway, and with repeated use it adapts by becoming less sensitive.

Midbrain dopamine neurons fire a brief burst to a reward that is better than predicted, show no response to a fully predicted one, and dip below baseline when a reward is worse than predicted or omitted. That pattern matches the prediction error of learning theory: the gap between expected and received, not the reward itself, drives the signal.

Early in training, animals press a lever purposefully and stop once the food reward is devalued; after extensive training they keep pressing anyway. Responding that ignores the outcome's current value is the working definition of a habit, and that shift is tied to the dorsolateral striatum. Whether habit explains human addiction is disputed by several human studies.

In fear conditioning, a neutral cue such as a tone is paired with a footshock until the cue alone triggers freezing. Tone and shock inputs converge on single neurons in the lateral amygdala, and the generally accepted model is NMDA-dependent strengthening of the tone input there. Even so, no mechanism yet fully explains the changes behind the learning.

After a stroke, activity and connections shift in tissue around the lesion, in distant regions and even in the opposite hemisphere, and these changes are thought to drive unaided recovery. Plasticity can also misfire: leaning on the less-affected side can build movement patterns that limit recovery. And after spinal cord injury, neurons that lose their axon connections may survive yet stop communicating.

The central nervous system begins very early in the embryo as a flat plate of tissue that narrows, bends into folds and fuses into a closed structure. When closure fails, the result can be spina bifida at the spinal end or anencephaly at the cranial end. Folic acid fortification and supplements have helped prevent these defects, though how folic acid works is poorly understood.

A newborn neuron must travel to where it will work. Some glide along the long fibers of radial glia, while others follow chemical signals from adhesion molecules on nearby cells. The trip is risky: scientists think only about a third arrive, and mutations in migration genes can leave misplaced neurons linked to disorders such as childhood epilepsy.

The growth cone at an axon's tip steers by reading cues: diffusible attractants and repellents, plus molecules anchored to cell membranes or the extracellular matrix. The same molecule can attract one axon type and repel another, and an axon's response changes during development. Slit-Robo repulsion, for example, lets commissural axons cross the midline only once.

Development removes surplus neurons on purpose: elimination of extra neurons is one of the few remodeling jobs that clearly relies on a built-in death program. The trigger appears to be too little neurotrophic support, and it can strike before or after a neuron reaches its target. Large numbers of neurons die during migration and differentiation.

Cortical myelin does not arrive all at once. In one MRI study of 484 people aged 8 to 85, using a marker linked to myelin, primary sensory and motor areas peaked in growth before puberty, while association, insular and limbic areas peaked after it. Most regions reached a stable plateau only in their 30s. The marker is an MRI proxy, not a direct count of myelin.

The brain stops growing in size by early adolescence, then spends years on fine-tuning, with the prefrontal cortex among the last regions to mature, into the mid-to-late 20s. Dual Systems models pin teen risk-taking on a fast socioemotional system outpacing slower cognitive control. One review argues that this imbalance claim has not yet been rigorously tested.

With age, some brain regions shrink, communication between neurons can weaken, blood flow may fall and inflammation may rise. Slower word finding and trouble multitasking are common, yet many studies find older adults know more words than younger adults, and given enough time to learn a new task they usually perform just as well. Telling typical change from disease in one person takes a clinician.

People with similar amounts of brain pathology can differ widely in dementia symptoms, and cognitive reserve is the presumed reason. Reserve cannot be observed directly, so studies rely on proxies such as education and occupational complexity. One large study tied each added year of education to lower odds of an Alzheimer's diagnosis after accounting for autopsy pathology; that is association, not proof.

The brain is about 2 percent of body mass yet takes 20 to 25 percent of the body's energy. Much of that bill goes to restoring ion gradients and running synaptic transmission, including recycling neurotransmitters. Energy supply is so tightly linked to neural activity that it provides the signals functional imaging methods such as fMRI and PET detect.

Glucose is the adult brain's main fuel, and little is kept in reserve: its small carbohydrate store sits in astrocytes and is used up within a few minutes if glucose stops arriving. Steady blood flow matters as much, since reduced flow and the resulting lack of oxygen can kill brain cells. Other fuels can step in under particular conditions, such as ketone bodies during fasting.

Cerebral blood arrives mainly through the two internal carotid arteries and the basilar artery, which join in a ring, the Circle of Willis. That ring offers backup routes, but the full version is uncommon: one population study found it in 11.9 percent of people. A blockage in the head or neck causes ischemic stroke, about 80 percent of strokes.

When neurons change their activity, local blood flow follows, and that link is the physiological basis of BOLD fMRI. The obvious explanation, that tissue calls for blood once fuel runs low, has weak support: one review argues against metabolic feedback as the trigger and favors feed-forward chemical signals set off by activity. How the coupling works is still debated.

The glymphatic hypothesis, named in 2012, proposes that cerebrospinal fluid enters tissue along arteries, helped by aquaporin-4 water channels, and carries solutes out along veins. Mouse tracer work reported far less marker entry awake than asleep. One detailed review disputes much of this: it finds no evidence that veins are the main exit, argues flow is too slow to beat diffusion, and offers another reading of the mouse result.

Microglia, the brain's resident immune cells, enter an activated state when they sense something harmful. The response is meant to protect and usually shuts down once the damage is dealt with, and their clearing of dead cells is generally helpful. The harm comes when activation is uncontrolled or prolonged, so timing and duration matter.

Most Parkinson's symptoms arise as nigrostriatal cells weaken, become damaged and die, removing a chemical signal that helps produce smooth, purposeful movement. By the time symptoms appear, most people have already lost 60 to 80 percent or more of them. Levodopa, which the brain converts into the missing messenger, can ease movement symptoms, but it does not replace lost cells or stop the disease from worsening.

Alzheimer's brains show deficits in choline acetyltransferase, the enzyme behind cholinergic signaling, tied to degeneration of cholinergic neurons such as those in the nucleus basalis of Meynert. That finding gave rise to the cholinergic hypothesis. Cholinesterase inhibitors have only a modest influence on progression, and the disease also involves amyloid plaques, tau tangles and other factors.

Myelin helps signals travel faster and farther, so losing it can disrupt or block conduction. In multiple sclerosis the immune system targets myelin in the central nervous system and leaves scar-like plaques. The damage goes beyond myelin: axons and nerve cell bodies are hurt too, and shifts in ion channels may add to the conduction failure.

Too much glutamate can kill the neurons it excites, an effect named excitotoxic in the 1970s. In the classic account, calcium overload activates enzymes and causes acute cell death, and brain ischemia raises extracellular glutamate within minutes. Yet most NMDA-blocking drugs have failed to show benefit in stroke trials, so a sound mechanism has not become a treatment.

A seizure reflects hyperexcitable neurons and hypersynchronous networks, a population firing at the same time. The standard account is net excitation exceeding net inhibition in a cortical area. That account is being widened: some epilepsy mutations do not overtly alter the balance, and some antiseizure medications do not work by decreasing excitation or increasing inhibition.

Repeated drug use can push sensitivity in opposite directions. Tolerance is a shrinking effect, partly because the reward circuit turns down its own signaling, so larger amounts are needed for the same high. Sensitization is a growing effect. One theory proposes that it inflates wanting without raising liking, which is a model, not a settled fact.

Tight junctions between the endothelial cells of brain vessels leave passive diffusion open mainly to lipid-soluble molecules under about 400 to 600 daltons. One review says the barrier excludes more than 98 percent of small-molecule drugs and all large-molecule therapeutics, and efflux pumps such as P-glycoprotein return some drugs to the blood. Making a drug more fat-soluble helps only sometimes.

The suprachiasmatic nucleus (SCN) acts as the master clock. It receives light information directly from the eyes, controls melatonin production and synchronizes the rhythms that nearly every tissue keeps. Light has the biggest influence, though food, stress and activity also matter, and SCN damage can leave sleep scattered through the day.

Reasoning from a deficit back to the job of the injured tissue is an inference to the best available explanation. One philosophical analysis holds that its hard problems are factual rather than logical: injury can also impair distant areas through lost connections, called diaschisis and disconnection, and a finding in one person may not generalize to all normal brains.

A double dissociation pairs opposite deficits: damage at one site impairs task A but not B, while damage at another impairs B but not A. In one mouse study, silencing one structure impaired delay but not trace fear conditioning, and silencing another did the reverse. One human lesion-mapping study tied agrammatic speech to frontal damage and paragrammatic speech to posterior temporal damage.

Single-unit recording uses a microelectrode to read one neuron's action potentials directly, with timing near a few milliseconds. In humans most of it comes from invasive epilepsy monitoring, which is how cells responding to a concept in any format were found. The catch is sampling: very small populations may not represent the larger group sharing a function.

Dense probes record many neurons at once. Neuropixels probes, with 960 sites on one thin shank, deliver low-noise recordings from hundreds of neurons, and spikes from different cells can be separated by amplitude and waveform across sites. Scale brings new bottlenecks: processing the data, working out where each site sits, and only 384 channels readable at a time.

EEG follows brain electrical activity on a millisecond scale without surgery. Its weak point is location: skull and scalp smear and weaken the signal, so working out where a source sits requires a detailed model of head tissues. It tends to favor activity in the gyri, though strong sources in the sulci can also be detected at the scalp.

An event-related potential is what remains after averaging many EEG trials locked to a stimulus or action; ongoing activity unrelated to the event averages away as noise. That is also the method's blind spot. Brain processes whose relationship to the stimulus is not constant cannot be seen by direct averaging, which is why later work turned to oscillations and spontaneous activity.

MEG measures magnetic fields from neural currents. Skull and scalp do not distort those fields, which is said to simplify source estimation; one review table puts its spatial resolution near 5 mm against about 10 mm for EEG. It has its own blind spot, because radial currents produce no magnetic field outside the head, so the two methods complement each other.

Functional MRI tracks blood oxygen during a task, a proxy for neural activity on a timescale of seconds. It uses no ionizing radiation, and one review table puts its resolution near a millimeter. Because the signal is an indirect, slow readout, fMRI says where a change happened more clearly than when, or what the neurons did.

BOLD reflects hemodynamics, including deoxyhemoglobin levels and the volume and rate of local circulation, that respond to neural metabolic demand, and it reaches its peak only about 4 to 6 seconds after activity. It is not a pure spike readout: one study in anesthetized monkeys tied it more closely to local field potentials, usually read as inputs, than to firing rate. Whether that holds in humans is unclear.

Noninvasive methods trade timing against location. One review table lists a scalp electrical method at about 0.05 seconds and 10 mm and a blood-oxygen method at about 1 second and 1 mm. Implanted electrodes do well on both, near 3 milliseconds and 1 mm or finer, but need surgery. Pick the method by whether your question is when or where.

PET follows a tracer: positrons from it annihilate with electrons and give off paired gamma photons that the scanner records. The tracer sets the question. FDG, a glucose analog, tracks energy use in active cells, while tracers aimed at dopamine or serotonin receptors estimate receptor density, which fMRI's signal cannot do directly.

Diffusion imaging exploits the way white matter constrains the random motion of molecules in tissue, then tractography stitches preferred directions into pathways. The results are reconstructions, not traced axons. Connection strength is usually estimated by counting recovered tracts or averaging a map such as fractional anisotropy, and those numbers relate only indirectly to real connectivity.

A rapidly switching magnetic field induces weak currents in the cortex under the coil. Because it can create a temporary virtual lesion in a healthy brain, TMS tests causation: if an area contributes to a behavior, stimulating it should worsen performance. As a therapy, repetitive TMS is FDA-cleared for a specific list of conditions, including treatment-resistant depression.

tDCS passes a weak current across the scalp. The premise is that anodal stimulation raises excitability and cathodal lowers it, as shown in motor cortex, but reported effects vary hugely, with contradictory results even at the same settings. Small samples add to the uncertainty: average power has been estimated below 50 percent, which can skew estimates in either direction.

Neurons are genetically engineered to carry an opsin, a protein that responds to light, and light of the right frequency then activates or silences them, depending on the opsin. Older tools were weaker on one axis: lesions and electrical stimulation hit regions without choosing a cell type, while drugs and gene manipulation are slow. One caveat: extremely responsive opsins can be driven from a source several millimeters away.

Chemogenetic tools add designer receptors, called DREADDs, to chosen neurons so that a designer drug activates or silences them. Timing follows the drug, so effects can last hours, unlike a light pulse. A known caveat: evidence suggests the classic drug, CNO, works in rodents through its conversion to clozapine, which is why newer ligands were developed.

Genetically encoded indicators such as GCaMP fluoresce as calcium changes, so you can watch large groups of neurons in behaving animals. Calcium is a fast but indirect readout of activity, since it also takes part in many other cell processes. Wide-field versions are limited to activity near the brain's surface and typically do not report single cells.

Animal models let you measure behavior under tight control, remove or silence a region and test again, which human work rarely allows. They fail where a capacity has no accepted animal analog, such as language, and where results do not carry over: most NMDA-blocking drugs failed stroke trials, and habits seen in drug-exposed animals often do not appear in complex decision settings.

Reverse inference reasons from activity back to a mental operation: this region was active, so that process was engaged. It is weak since an active region may reflect a wide variety of cognitive processes. Forward inference runs the other way, from an engaged function to predicted activity. The same pattern can appear in different conditions, so it cannot by itself diagnose a disorder.

A whole-brain fMRI analysis tests every voxel, sometimes over 200,000 of them, so skipping correction can produce an extremely high false positive rate. Correction has its own traps. A 2016 analysis reported inflated false positives in common software, and a commentary argued that results also depend on a preprocessing choice: how finely data are resampled.

Common fMRI samples of 20 to 30 people can have very low power for weak, diffuse effects, and results shift widely between replications. Effects that clear significance also tend to overestimate the true size. One analysis across eleven tasks found that even samples near 100 fell well short of perfectly replicable. How severe this is for fMRI is still debated.

Imaging records activity without manipulating it, so a region active during a task is not thereby shown to be needed for it. In consciousness research, one analysis notes that correlated states include causes, effects and enabling conditions. One mouse paper raises a parallel possibility: a region might be engaged during a type of learning without its activity being required. Disruption methods such as TMS test this more directly.

Decoding, or multivoxel pattern analysis, trains a classifier to link distributed activity patterns to experimental conditions, then scores it on separate test data. It can use information across voxels that a voxel-by-voxel analysis ignores. Overfitting is the trap: a classifier tuned too tightly to training data generalizes poorly, which is why the test data must be kept apart.

A connectome is a map of neural elements and the connections between them, usually stored as a matrix. Scale is the central choice: tracing some 10^15 connections is not only technically impossible but may be unnecessary, so a first human draft was proposed at the scale of hundreds of regions and thousands of pathways. That level leaves out subcircuits within regions and differences between people.

Ask four things of any brain claim. Method: does it measure spikes or a blood oxygen proxy? Sample: 20 to 30 people gives low power for weak effects. Effect size: significant results tend to overestimate. Inference: does it reason backward from activation to a mental state, or treat correlation as cause? Extra brain detail alone should earn no credit.

The usual definition is a minimal neural system sufficient for an experience. The word minimal matters, since saying the whole brain suffices explains nothing. Mere correlation falls short: an experience's causes, its effects and background needs such as oxygen all correlate with it without explaining it.

In global workspace theory, a state is conscious when its content enters a global neuronal workspace that makes it available to memory, motor, evaluative and attentional systems. One preregistered adversarial test in 256 people supported some predictions but found a general lack of the predicted ignition when a stimulus ended, and limited prefrontal coding of some conscious features.

IIT starts from claimed axioms of experience and identifies consciousness with integrated information, symbolized Phi: a system integrates when the whole carries more than the sum of its parts. One critique notes it implies a form of panpsychism and that a simple grid of XOR gates could score high. In a 2025 adversarial test, posterior cortex lacked the sustained synchronization it expected.

Higher-order theories say a state is conscious only when you represent yourself as being in it; a visual state you are in no way aware of would not be conscious. Tying that representation to prefrontal activity makes the idea testable, since prefrontal lesions should affect consciousness. Some versions allow a conscious state without any matching visual activity, which is debated.

Chalmers's hard problem asks why physical processing is accompanied by experience at all, the felt quality of deep blue, rather than how the brain performs a function. He argues empirical explanations are structural or functional and experience reduces to neither. Some philosophers deny the problem, and neuroscience often targets what turns consciousness on and off instead.

After the corpus callosum was cut to treat intractable epilepsy, patients classically denied seeing a stimulus in the left visual field yet picked it out with the left hand, suggesting a divided mind. That reading is disputed: patients behave ordinarily, report feeling unchanged, and several studies show accurate location reports across the whole visual field.

One split-brain patient, shown the word Texas in the left visual field, drew a cowboy hat with his left hand while his speaking hemisphere showed no insight into why. Healthy people show a related gap: in choice blindness studies they justify choices they did not make, and one study found such confabulation most prominent when internal evidence was weak.

Some people with damage to primary visual cortex report seeing nothing in part of their visual field yet do far better than chance when made to guess about it. In one study of 17 patients, those with residual vision all kept a pathway from the lateral geniculate nucleus to motion area hMT+. Whether this vision truly lies outside awareness is contested; one case suggests degraded but still aware vision.

If separate neuron groups code color and form, a red circle is easy to represent. A red circle and a blue square are not: superimposed patterns for the two colors blur into a pattern that is neither, a superposition catastrophe. Neural synchrony is one proposed fix, others avoid it, and how the brain actually solves this remains an outstanding question.

Chemically diverse anesthetics all disrupt consciousness, and how they do so has been asked since the nineteenth century. An influential idea blames blocked sensory relay through the thalamus; one review says recent evidence suggests a top-down process may be involved, and the two remain in dispute. Waking is not the mirror image of going under: in insects and mammals, the two paths differ beyond what drug clearance explains.

Coma, the vegetative state and the minimally conscious state are sorted by degrees of arousal and awareness. People in a vegetative state, or unresponsive wakefulness, show sleep/wake cycles but no response to their surroundings; a minimally conscious state adds inconsistent command following. Telling them apart is hard: up to about 40 percent of patients can be misdiagnosed when specially designed scales are not used properly.

Default mode regions show more activity in a task-free state than during demanding tasks, and they deactivate more as attention load rises. Hubs include the posterior cingulate, linked to autobiographical memory and self-referential processes. What the network does at rest is unsettled: one review reports that neither the introspection account nor the sentinel account has been fully validated.

Some authors treat attention as the gate to consciousness; others present evidence that the two are separate. Experiments show attention can be deployed to a stimulus that stays out of awareness, and neglect and blindsight show unconscious processing guiding behavior. The reverse question, whether awareness can occur without attention, is also debated.

In Libet's task, people flexed a wrist and reported when they felt the urge to move. Reported intention came about 200 ms before movement but lagged the readiness potential by about 300 to 500 ms, which some took to mean the brain decides first. That interpretation has been severely criticized, and what the timing reports actually reflect is still debatable.

A brain-computer interface records neural activity and uses algorithms to produce signals that control a tool, such as a robotic limb or wheelchair. Noninvasive EEG systems are the most common. In one trial of implanted arrays in three people with paralysis, the top participant averaged nearly 8 words per minute, a clinical research result rather than an everyday product.

Implanted intracortical electrodes and electrocorticography give better signal-to-noise and better localization, and intracortical recordings outperform EEG for restoring mobility. They also require surgery. For a locked-in patient the benefits can outweigh implantation risks, while invasive systems should generally not be considered for neurologically intact people.

In one clinical study, implanted silicon probes in the hand area of motor cortex recorded spiking and high-frequency field potentials. Algorithms such as a Kalman filter turned those signals into point-and-click control on a screen. The mapping is not fixed: baselines drifted over time in one participant, and long-term stability without recalibration remains a key challenge.

Neural prostheses work in two directions. Output devices read brain activity to control tools such as robotic limbs or powered wheelchairs; input devices stimulate nerve tissue, as a cochlear implant does for the auditory nerve. Restoration has limits: a cochlear implant does not restore normal hearing, and users need time to learn what its signals mean.

A cochlear implant bypasses damaged parts of the ear: a microphone and speech processor feed an electrode array that stimulates different regions of the auditory nerve. By late 2019 about 736,900 devices had been registered worldwide. It does not restore normal hearing, and hearing through one takes time to learn, yet it lets many people understand speech.

DBS sends electrical pulses through surgically implanted electrodes, driven by a pacemaker-sized generator in the chest. It was first developed for movement disorders such as tremor and Parkinson's disease. Exactly how it works is unclear; one idea is that pulses reset malfunctioning networks. It does not cure Parkinson's, and it leaves symptoms such as balance problems and dementia.

Brain signals change within and across people over short and long periods, so systems need subject-specific calibration that users can find tedious. Some users struggle to gain control, sometimes because of technical limits rather than their own signals. Implants add open questions of tissue acceptance and long-term signal stability, and moving from lab results to plug-and-play tools remains a challenge.

Franz Joseph Gall's phrenology read mental faculties from bumps on the head, on the false premise that soft brain tissue shapes the skull. One MRI study in almost six thousand people found no significant or meaningful link between local scalp curvature and measures standing in for his faculties. When a bump happens to sit near a region with a known job, the match is likely accidental.

The claim that people use only 10 percent of their brain is a classic neuromyth; a survey of the Brazilian public found it the most prevalent misconception. One systematic review of neuromyths answers that no single brain area is entirely out of work, even during sleep. The brain also takes 20 to 25 percent of the body's energy at about 2 percent of its mass.

Some functions really are lateralized: language hubs and attention-control hubs sit mainly in opposite hemispheres. The personality theory is the overreach. A resting-state study of 1,011 people aged 7 to 29 found lateralization to be a local property of networks, with no whole-brain left-brained or right-brained type. Sorting people into left- and right-brain learners is a known neuromyth.

Adding irrelevant neuroscience can make an explanation seem better, the seductive allure effect. In one preregistered online study of 430 people, superfluous neuroscience boosted ratings mostly among those with less expertise, and good explanations beat bad ones by a surprisingly small margin. Credit brain detail only when it actually bears on the claim.