Neuroscience · Unit 6: Cognition and learning

Language and Lateralization

How is language laid out in the brain, and how sure can we be? This lesson starts from the nineteenth-century lesion cases of Broca and Wernicke, re-reads them with modern MRI and stroke cohorts, treats left-hemisphere dominance as a measured probability, and works out the split-brain experiments from the crossings of the visual and touch pathways. Afterwards you can sort the aphasias by four bedside tests, calculate a conditional probability from lateralization rates, design a fair split-brain test, and say what an fMRI activation can and cannot tell you about language.

Watch first Same Rates, Different Base Rate · 2:58

In 1861 Paul Broca examined a man who could say almost nothing beyond one syllable, yet appeared to follow much of what was said to him.

When he died, Broca found damage in the left frontal lobe and argued that speech production lives there. That pair, a symptom and a lesion, started a century and a half of mapping language onto the brain. This lesson asks how good that map is. You will sort the aphasias, see what MRI found in Broca’s own specimens, treat “language is on the left” as a number with a base rate, work out the split-brain experiments from the wiring, and see how imaging of healthy brains has redrawn the picture.

Two patients and a model: four tests sort the aphasias

An aphasia is an impairment of producing or understanding language caused by acquired brain damage, most often a stroke. It is a disorder of language, and it does not mean a loss of intelligence. It must be kept apart from dysarthria, a motor speech disorder in which weak or uncoordinated speech muscles slur articulation while language itself (grammar, word choice, comprehension, writing) is intact.

Clinicians sort aphasias with four questions. Is speech fluent, with normal phrase length and rhythm? Is comprehension of spoken words and sentences preserved? Can the person repeat a phrase? Can they name a pictured object? Naming is impaired in almost every aphasia, so it rarely discriminates on its own.

  • Broca aphasia: nonfluent. Speech is slow and effortful, short phrases with disrupted grammar. Comprehension is relatively preserved, though complex sentences often trip it. Repetition is poor.
  • Wernicke aphasia: fluent, often rapid speech that conveys little meaning, with word substitutions. Comprehension of words and sentences is severely impaired, and so is repetition.
  • Conduction aphasia: fluent speech, comprehension relatively good, but repetition is disproportionately poor, with sound errors the person often tries to correct.
  • Dysarthria: articulation is impaired; all four language tests are normal once speech motor demands are removed, for instance in writing.

Carl Wernicke described the fluent syndrome in 1874 after damage in the posterior superior temporal gyrus, and predicted that cutting the connection between posterior and frontal language cortex would spare comprehension but wreck repetition. In the 1960s Norman Geschwind built these cases into a model: a posterior area for word sounds and meaning, a frontal area for speech planning, and a tract (a bundle of axons in the central nervous system), the arcuate fasciculus, joining them. The model is a good teaching scaffold. Its anatomy is where it strains.

What the lesions really were: deeper than Broca could see

Broca’s area is conventionally the pars opercularis and pars triangularis of the left inferior frontal gyrus. Broca kept his two patients’ brains whole, and in 2007 Nina Dronkers and colleagues imaged them with high-resolution MRI. In both, the damage extended medially beyond the surface lesions Broca had described. His founding cases never isolated a small patch of surface cortex.

Modern cohorts also complicate a one-region story. In 70 people with chronic aphasia, Fridriksson and colleagues (2015) used lesion-symptom mapping and found that damage to the pars opercularis alone was not sufficient to produce Broca aphasia. Wernicke’s area is worse defined: where its boundaries lie is still disputed (Mesulam and colleagues, 2015), so no single outline should be learned as settled anatomy.

The current replacement is a network of two streams, an idea developed by Gregory Hickok and David Poeppel among others. A dorsal stream runs from posterior temporal and inferior parietal cortex to frontal cortex and maps speech sounds onto articulation. A ventral stream runs through the temporal lobe and maps sounds onto meaning. Fridriksson and colleagues (2018) tested this in 159 native English speakers at least six months after a single left-hemisphere stroke, relating damaged regions and estimated connections to 16 speech and language measures. Motor speech impairment went mostly with dorsal-stream damage, comprehension impairment more strongly with ventral-stream damage, and naming, repetition and grammar depended on both. Keep the hedges: strokes follow arterial territories and damage neighbouring regions together, so these are associations.

Lateralization by the numbers: a probability, not a rule

Lateralization is a functional difference between the two cerebral hemispheres. Language is left-dominant in most people, but how most? Knecht and colleagues (2000) measured it in 326 healthy adults with functional transcranial Doppler during word generation. Right-hemisphere dominance occurred in 4% of strong right-handers, 15% of ambidextrous people and 27% of strong left-handers. Left-handedness raises the odds of right-sided language; it does not make it typical.

Worked problem. Use Knecht’s rates as fixed probabilities. The cohort sizes below are assumptions for the exercise, not measured prevalences, and they ignore everyone between the handedness extremes.

Part (a). Take 100 strong right-handers and 100 strong left-handers.

  • Expected right-dominant right-handers: 100 × 0.04 = 4.
  • Expected right-dominant left-handers: 100 × 0.27 = 27.
  • Right-dominant overall: (4 + 27) / 200 = 31 / 200 = 15.5%.

Now pick one right-dominant person. The chance that they are left-handed is

P(left-hander | right dominance) = 27 / (27 + 4) = 27 / 31 = 0.871, or 87.1%

Compare the reverse conditional: P(right dominance | left-hander) = 27%. A right-dominant person in this sample is very likely left-handed, yet most left-handers are left-dominant. The two statements answer different questions.

Lateralization is a probability — the base rate decides the answer⚙ original · interactive
Question
Right-language-dominant right-handers
Right-language-dominant left-handers
Right-language-dominant overall
The other conditional
Share of right-language-dominant people

In this sketch: Knecht’s rates are fixed probabilities (4% and 27%), the cohort sizes are assumed, and ambidextrous people and everyone between the extremes are left out. Rates are for language dominance by functional transcranial Doppler during word generation, in healthy adults (Knecht and colleagues, 2000); left-language-dominant means not right-language-dominant here. Counts are expected values per 1,000; icons are rounded to whole people.

The rates did not change, the base rate did: P(right-language dominance | left-hander) stays 27%, while P(left-hander | right-language dominance) depends on how many left-handers there are. Most left-handers are still left-language-dominant, and lateralization is function by function. The personality claim of “left-brained” or “right-brained” people is not supported by the evidence (a resting-state fMRI analysis of more than 1,000 people, Nielsen and colleagues, 2013, found no global left or right dominance).

Part (b). Now assume a population that is 90% strong right-handers and 10% strong left-handers.

  • Right-dominant right-handers: 0.90 × 0.04 = 0.036.
  • Right-dominant left-handers: 0.10 × 0.27 = 0.027.
  • Right-dominant total: 0.063, or 6.3%.

P(left-hander | right dominance) = 0.027 / 0.063 = 0.429, or 42.9%

Check: 0.036 / 0.063 = 0.571, and 0.429 + 0.571 = 1. The rates did not change; the base rate did, and the answer fell from 87% to 43%. Rightward language is now more often found in a right-hander, simply because right-handers are so many.

Lateralization is function by function. That is different from the popular claim that people are “left-brained” (logical) or “right-brained” (creative) in personality, which the evidence does not support: a resting-state fMRI analysis of more than 1,000 people (Nielsen and colleagues, 2013) found no sign that individuals are globally left- or right-dominant. How age at exposure shapes language is a developmental question, treated in sensitive periods and behaviour.

Cutting the commissures: what split-brain patients show

A commissure is a tract that crosses the midline to link the two hemispheres; the corpus callosum is the largest (the map of the nervous system). How commissural axons find the midline during development belongs to crossing the midline. Cutting the forebrain commissures was used to treat epilepsy that drugs could not control.

To read the results, you need the crossings. Contralateral means on the opposite side; ipsilateral, on the same side.

  • Vision. At the optic chiasm only the fibres from the nasal half of each retina cross. So everything in the left visual hemifield (the half of space left of the fixation point) reaches the right hemisphere, from both eyes (visual cortex). The variable is the hemifield, not the eye.
  • Touch. Fine touch from the hand ascends in the dorsal columns and crosses in the medulla, so the left hand reports to the right hemisphere (touch and pain).
  • Movement. Most corticospinal fibres cross at the pyramidal decussation, so each hand is controlled mainly by the opposite hemisphere (the basal ganglia).
  • Hearing reaches both hemispheres (hearing and balance), which makes sound a leak in any lateralized test.

Gazzaniga, Bogen and Sperry (1962) reported patient W.J., whose corpus callosum and anterior and hippocampal commissures had been cut. Blindfolded, he could use familiar objects with his left hand but could not name them. With brief tachistoscopic flashes, his left hand could select some simple patterns shown in the left field that he could not identify verbally; without that control the deficit was not shown satisfactorily. The authors warned that possible earlier brain injury limited how far one patient generalizes.

Later testing showed that the disconnected right hemisphere understands some spoken language, so it is mute more often than it is deaf to language (Sperry, Nobel lecture, 1981). The tidy hand-by-field rule also has exceptions: Pinto and colleagues (2017) found that two patients with complete callosal section recognized stimuli in either field above chance with any response, while still failing to compare across fields. Whether the surgery creates two minds is unsettled; a 2020 collective review judged the evidence insufficient to decide.

Design problem

Design a test showing that a split-brain patient’s right hemisphere understood a written word it cannot say. List your controls and what the result would and would not show.

One way to do it

Track the eyes and start each trial only when gaze holds on a central dot. Flash a noun, such as “spoon”, in the left visual field for 150 ms or less, shorter than the time needed to launch a saccade, so the word reaches only the right hemisphere. Behind a screen, the left hand searches a tray of objects by touch; the patient also says what he saw. The prediction is a verbal report of “nothing” or a guess, with the left hand choosing the spoon above chance.

Controls: right-field trials, where he should name the word; right-hand responses to left-field words, which should fail if the hand-by-field rule holds; silent objects and a screen, so neither sound nor sight of the hand cues the speaking hemisphere; randomized fields and objects. A positive result shows right-hemisphere comprehension of that word class in this patient. It does not show two persons, and it may not generalize: patients are few, all had epilepsy, some had earlier damage, and Pinto and colleagues found the hand-by-field rule is not universal.

The language network in living brains

Lesions show what a brain cannot do without some tissue. Imaging shows what healthy tissue does. Fedorenko, Behr and Kanwisher (2011) located language regions in each person’s own brain, by contrasting sentences with nonword lists, and then found those regions responded little or not at all to arithmetic, working memory, cognitive control and music. Malik-Moraleda and colleagues (2022) ran a similar localizer in 86 native speakers of 45 languages from 12 families, contrasting native-language passages with acoustically degraded ones, and found a broadly shared fronto-temporo-parietal network whose regions’ activity was strongly correlated. In left language regions, native speech produced a 2.13% blood-oxygen-level-dependent (BOLD) signal change from fixation against 0.84% for degraded speech, about 2.5 times larger. All participants were also fluent in English, and each language was represented by two native speakers (four languages by one), so fine differences between languages stay open.

Neither result shows necessity: an activation means involvement, and only damage or disruption tests need. The strongest account combines the two kinds of evidence, and the next lesson, attention and decisions, applies the same discipline to choosing what to process.

How we measure it

Lesion-symptom mapping

Structural MRI outlines each patient's damage, and the damaged voxels or connections are related statistically to scores on language tests across a cohort. It shows which damage goes with which deficit, but strokes damage neighbouring regions together, so it cannot by itself isolate the one region a function needs.

Tachistoscopic divided-visual-field testing

A person fixates a central point while a word or picture is flashed briefly to one side, too briefly for an eye movement to bring it to the centre. Because each visual hemifield projects to the opposite hemisphere, the method sends input to one hemisphere at a time, which is what made the split-brain experiments readable.

Functional transcranial Doppler ultrasonography

Ultrasound probes measure blood-flow velocity in the left and right middle cerebral arteries while a person performs a language task such as word generation. The side with the larger rise is taken as dominant, which lets lateralization be measured quickly and without injection in healthy people.

Individual-subject fMRI localizers

A short contrast (sentences against nonwords, or native speech against degraded speech) defines language-responsive regions in each person's own brain, and those regions are then tested on new tasks. This avoids blurring across people, but an activation shows involvement, not that the region is needed.

Finish line

Test yourself

8 questions

Practice and MCAT-style, with explained answers.

Next · Lesson 25

Attention, Executive Control and Decisions

A parietal neuron's firing climbs toward a fixed level just before a monkey announces its choice. Silence that area and the monkey chooses just as well. What was the neuron doing?

Continue →

← Lesson 23 · Plasticity, Networks and Consolidation

Paul Broca

Paul Broca 1824–1880

Described in 1861 two patients who had lost most speech and whose autopsies showed damage in the left inferior frontal lobe, the founding cases for localizing language; in 2007 Nina Dronkers and colleagues re-examined both preserved brains with MRI.

In April 1861 Paul Broca, a Paris surgeon and anthropologist, examined a patient at the Bicêtre hospital named Leborgne, who had lost nearly all speech years before. Hospital accounts called him 'Tan', after the one syllable he could reliably produce, yet he appeared to understand much of what was said to him. Leborgne died within days, and Broca found a lesion on the surface of the left frontal lobe. Later that year a second patient, Lelong, who had only a few words left, proved to have damage in a similar place. Broca kept both brains whole rather than slicing them, and they were preserved in a Paris museum. By the mid-1860s he was arguing that articulate speech depends mainly on the left hemisphere. In 2007 Nina Dronkers and colleagues scanned the two brains with high-resolution MRI and found that both lesions extended medially, beyond the surface damage Broca could see. That study made the founding cases of language localization readable in modern anatomical terms.

✓ verbatim Dronkers, Plaisant, Iba-Zizen & Cabanis, Brain 130:1432–1441 (2007), which reviews Broca's 1861 reports and images both brains.
Roger Sperry and Michael Gazzaniga

Roger Sperry and Michael Gazzaniga

With the neurosurgeon Joseph Bogen, tested patient W.J. after section of the forebrain commissures and showed that his left hand could use objects and select some patterns presented to his right hemisphere that he could not name (PNAS, 1962).

Roger Sperry had shown in amphibians that regenerating optic axons find their proper targets, and in the 1950s he and his students, at Chicago and then at Caltech, studied cats in which the optic chiasm and corpus callosum had been cut, so that what one eye learned stayed in one hemisphere. In 1960, by his own later account, Michael Gazzaniga arrived at Caltech on a summer fellowship from the National Science Foundation to work with Sperry. Joseph Bogen, then a neurosurgical resident, was evaluating people with epilepsy that drugs could not control for surgery to cut the forebrain commissures. Gazzaniga, Bogen and Sperry tested one of these patients, W.J., after the operation, using brief flashes from a tachistoscope Gazzaniga built and blindfolded tests of each hand. Their 1962 paper reported that W.J. could use familiar objects with his left hand that he could not name. Lateralized testing of this kind became the standard method of split-brain research, and Sperry shared the 1981 Nobel Prize in Physiology or Medicine for discoveries about the functional specialization of the cerebral hemispheres.

✓ verbatim Gazzaniga, Bogen & Sperry, PNAS 48:1765–1769 (1962); Gazzaniga's retrospective article in PNAS (2014); the 1981 Nobel Prize in Physiology or Medicine (nobelprize.org).
Stefan Knecht

Stefan Knecht

With colleagues, measured language lateralization in 326 healthy adults with functional transcranial Doppler and showed that right-hemisphere dominance rises with left-handedness but stays the minority pattern (Brain, 2000).

Most of what was known about which hemisphere holds language came from people with strokes, or from patients given an anaesthetic injection into one carotid artery before epilepsy surgery. Both groups have damaged or abnormal brains, which can move language. Knecht and colleagues wanted rates for healthy people. They used functional transcranial Doppler ultrasonography: a probe over each temple measures blood-flow velocity in the left and right middle cerebral arteries while the person generates words. The side whose flow rises more is scored as language-dominant. They tested 326 healthy adults spread across the range of handedness. Right-hemisphere dominance appeared in 4% of strong right-handers, 15% of ambidextrous people and 27% of strong left-handers. Their paper in Brain in 2000 turned a clinical rule of thumb into measured rates, and showed that left-handedness shifts the odds of right-sided language without reversing them.

✓ verbatim Knecht et al., Brain 123:2512–2518 (2000).
Evelina Fedorenko

Evelina Fedorenko

Defined language regions separately in each person with fMRI and showed that they respond little or not at all to arithmetic, working memory, cognitive control and music (with Behr and Kanwisher, PNAS 2011); later extended the approach to 45 languages.

Fedorenko, working with Michael Behr and Nancy Kanwisher at MIT, asked whether the brain regions that respond to language also do other demanding mental work. Group-averaged fMRI studies had reported overlap between language and arithmetic, music or working memory. But averaging across people can merge neighbouring regions whose exact positions differ from one brain to the next. So in each participant they first located language regions with a localizer scan, reading sentences contrasted with lists of nonwords, and then measured how those same regions responded to arithmetic, working-memory, cognitive-control and music tasks. The language regions showed little or no response to the nonlinguistic tasks they tested (PNAS, 2011). In 2022 Saima Malik-Moraleda, Fedorenko and colleagues applied the same logic to 86 native speakers of 45 languages and found a broadly shared fronto-temporo-parietal network. The approach gave the field a way to define language regions person by person before asking what else they do.

✓ verbatim Fedorenko, Behr & Kanwisher, PNAS 108:16428–16433 (2011); Malik-Moraleda et al., Nature Neuroscience 25 (2022).
Test yourself
Question 1 of 8 Score 0 / 0

A patient speaks in long, fluent, well-articulated sentences that convey little meaning. She cannot follow simple spoken commands, and she cannot repeat a short phrase. Which syndrome fits best?
Sources & Integrity

Every structure, slide, simulation, and clip on this page is listed below with the source it came from. None of the 3D structures is drawn by hand or by AI: each one loads by accession ID from RCSB PDB or PubChem. This panel is generated from the lesson's own data file, so what it lists is what the page loads.

Simulations

  • Base rates and laterality · Original interactive — Biochemistrypedia (Deppmann Lab).

Scientist profiles & quotes

  • Paul Broca — Dronkers, Plaisant, Iba-Zizen & Cabanis, Brain 130:1432–1441 (2007), which reviews Broca's 1861 reports and images both brains.
  • Roger Sperry and Michael Gazzaniga — Gazzaniga, Bogen & Sperry, PNAS 48:1765–1769 (1962); Gazzaniga's retrospective article in PNAS (2014); the 1981 Nobel Prize in Physiology or Medicine (nobelprize.org).
  • Stefan Knecht — Knecht et al., Brain 123:2512–2518 (2000).
  • Evelina Fedorenko — Fedorenko, Behr & Kanwisher, PNAS 108:16428–16433 (2011); Malik-Moraleda et al., Nature Neuroscience 25 (2022).

Quotes are verbatim from the cited source. Profile text is original to this site and was checked against the sources cited.

Animated explainer

Same Rates, Different Base Rate · narration: an AI narrator voice (Gemini, Charon) reading narration written from this lesson's text · animation: AI-generated in code (Manim) by Claude, a schematic with no drawn molecular structures, plots computed from their equations · captions from the narration script.

Images & figures

This lesson carries no raster figures. Every structure is rendered live from RCSB PDB or PubChem, and every plot is drawn from its equation in the browser.

integrity Machines check the mechanics. The build confirms that every local file exists, that every question's answer key points at a real choice, and that any AI-generated scientific figure names the person who checked it; a weekly job re-resolves every accession ID against its database. Questions are original; MCAT-style items are written for this site and are not real AAMC questions. Banned sources (the Tymoczko textbook, publisher decks, copyrighted figures) are not used.

Like I'm five

Talking and understanding words use many parts of your brain working together, mostly on the left side. If one part gets hurt, a person might understand you but struggle to talk, or talk easily but not make sense. The two halves of your brain talk to each other through a thick bridge in the middle. When doctors cut that bridge to stop bad seizures, scientists learned that the right half can understand some words even though it usually cannot say them.
High school

Damage to the left frontal lobe can leave someone able to understand much of what they hear but speaking slowly and with effort (Broca aphasia). Damage farther back, near the temporal lobe, can leave speech fluent but empty, with poor understanding (Wernicke aphasia). Language is on the left in most people, but not all: in one healthy-adult study about 4% of strong right-handers and 27% of strong left-handers had it on the right. In split-brain patients, whose connecting fibres were cut to treat epilepsy, a picture shown only to the right hemisphere classically cannot be named but can be picked out by the left hand.
College

Aphasias are acquired language impairments sorted at the bedside by fluency, comprehension, repetition and naming: Broca aphasia is nonfluent with relatively preserved comprehension, Wernicke aphasia fluent with severe comprehension loss, conduction aphasia fluent with disproportionately poor repetition, and dysarthria a motor speech disorder with language intact. The mapping to areas is approximate. MRI of Broca's two preserved brains showed damage reaching medially beyond the surface lesion, pars opercularis damage alone did not produce chronic Broca aphasia in a 70-patient cohort, and in 159 stroke survivors motor speech went mostly with dorsal-stream damage and comprehension more with ventral-stream damage, with naming and repetition drawing on both. Lateralization is a probability that shifts with handedness. Split-brain results follow from the crossings: each visual hemifield reaches the contralateral hemisphere through the chiasm, and the hand's touch and motor pathways are mainly contralateral.
Graduate

Modern accounts replace the Wernicke-Geschwind box-and-arrow model with distributed streams: a dorsal route linking posterior temporal and inferior parietal cortex to frontal cortex for mapping sound onto articulation, and a ventral temporal route for mapping sound onto meaning. Chronic stroke data (Fridriksson and colleagues, 2018; 159 survivors, 16 measures, lesion and probabilistic tractography) support partly separate stream contributions, but vascular lesions are spatially correlated and tractography estimates links rather than counting axons. Individual-subject fMRI localizers define a fronto-temporo-parietal network that responds weakly to arithmetic, working memory and music (Fedorenko and colleagues, 2011) and recurs in native speakers of 45 languages (86 people, all fluent in English; Malik-Moraleda and colleagues, 2022), where native passages gave a 2.13% and degraded speech a 0.84% BOLD change from fixation in left language regions. Lateralization estimates depend on method (fTCD, intracarotid amobarbital, fMRI laterality indices) and on sampling: Knecht's rates are for healthy adults at the handedness extremes, so any population-level conditional probability needs an assumed base rate. In split-brain work, two patients with radiologically complete callosal section responded above chance to stimuli in either hemifield with either response type while still failing cross-field comparisons (Pinto and colleagues, 2017). Open question: whether callosotomy divides consciousness, which a 2020 collective review judged the evidence insufficient to decide.
Two brains kept whole for 146 years

Broca did not slice the brains of Leborgne and Lelong. He kept them intact, and they stayed in a Paris collection into the twenty-first century. That choice, made for preservation, left a test open. In 2007 Nina Dronkers and her colleagues placed both brains in an MRI scanner and imaged them at high resolution. Both lesions ran medially, beyond the surface damage that Broca had drawn and that later textbooks labelled as Broca's area. The famous cases had never isolated a small patch of cortex.

— Dronkers, Plaisant, Iba-Zizen & Cabanis, Brain 130:1432–1441 (2007).
A homemade tachistoscope

Split-brain testing depends on timing. A picture must reach one visual hemifield and vanish before the eyes can move to it, or both hemispheres will see it. Gazzaniga's retrospective article about his early years at Caltech includes a photograph of a tachistoscope he built for the first tests. The 1962 report on W.J. noted that without this kind of control the visual deficit could not be shown satisfactorily. A piece of apparatus decided whether the disconnection was visible at all.

— Gazzaniga, PNAS (2014), a retrospective autobiographical article.
Two patients who broke the rule

The textbook split-brain result is tidy: left-field stimuli can be reported only with the left hand. Yair Pinto and colleagues tested two patients whose callosal section had been confirmed by imaging. Both still failed to compare stimuli across the two hemifields, yet both recognized stimuli in either field above chance whether they answered by pointing with either hand or by speaking. The authors read this as evidence against divided consciousness. A 2020 review written jointly by researchers of differing views concluded the question was not yet settled.

— Pinto et al., Brain 140 (2017); de Haan et al., Neuropsychology Review 30 (2020).

 ↑ Top