Functional neuroimaging (fMRI measures blood oxygenation; PET measures metabolic activity) reveals neural substrates of language by measuring regional brain activation during linguistic tasks. These methods have localized syntax, semantics, and phonology to partly overlapping but dissociable regions.
Survey landmark neuroimaging studies distinguishing syntactic (Broca's area, left anterior insula) and semantic (temporal lobes) processing; learn technical details of spatial resolution and temporal assumptions.
Neuroimaging reveals correlations, not causation; activation does not prove a region is necessary for a function—lesion and patient studies provide stronger evidence of necessity.
From your study of language and the brain, you know the classical neurological picture: Broca's area (left inferior frontal gyrus) for production and syntactic processing, Wernicke's area (left superior temporal gyrus) for lexical-semantic comprehension, connected by the arcuate fasciculus. This picture was built almost entirely from lesion studies — observing which language capacities break down when specific brain regions are damaged. Lesion studies tell you about necessity: if damage to region X impairs function Y, then X is necessary for Y. But they have a fundamental limitation: strokes and tumors don't respect the boundaries of cognitive functions. By the 1990s, neuroimaging offered something entirely new — the ability to watch healthy brains process language in real time.
PET (Positron Emission Tomography) was the first widely used functional neuroimaging method. Participants are injected with a radioactive tracer that concentrates in metabolically active tissue; the scanner detects gamma rays emitted by tracer decay and reconstructs regional blood flow maps. Higher blood flow indexes neural activity. PET has reasonable spatial resolution (about 5–10mm) but poor temporal resolution — a scan integrates activity over 40+ seconds, far too slow to track the millisecond-scale dynamics of real-time language processing. fMRI (functional Magnetic Resonance Imaging) measures the BOLD signal (Blood Oxygen Level Dependent), exploiting the fact that oxygenated and deoxygenated hemoglobin have different magnetic properties. When neurons fire, oxygenated blood rushes in; the local magnetic field shifts slightly; the scanner detects this as increased signal. fMRI offers better spatial resolution (2–3mm) and temporal resolution (2–4 seconds), with no radiation. These technical parameters matter: what you can discover about language is constrained by what each method can measure.
The landmark findings from the 1990s onward revealed that the classical two-region picture was radically incomplete. Syntactic processing activates not just Broca's area but a left-lateralized network including the left anterior insula, supplementary motor area, and posterior superior temporal sulcus. Semantic processing involves extensive bilateral temporal lobe activation, with greater left lateralization for combinatorial meaning. The two systems overlap substantially — pure syntax versus semantics is not a clean neural division. Crucially, Broca's area activates across multiple functions: phonological working memory, hierarchical structure building, action observation, and music processing. This functional promiscuity means that "Broca's area is the syntax region" is an oversimplification: it is more accurately a region that contributes to a family of computations, language being one of them.
The epistemological constraint you need to hold onto is the one flagged in the misconceptions: neuroimaging evidence is correlational. A region that activates during syntactic processing is engaged during that processing — but this does not establish that it is *necessary* for it. Patients with Broca's area damage show syntactic deficits, which does converge with the imaging evidence; but there are also reports of patients who recover syntactic capacity despite permanent lesions, suggesting other regions can compensate. The strongest claims in cognitive neuroscience require converging evidence from multiple methods: neuroimaging to localize, lesion studies to establish necessity, TMS (transcranial magnetic stimulation) to create temporary virtual lesions in healthy participants, and EEG/MEG to track temporal dynamics. Neuroimaging is most powerful as one voice in a chorus, not as a stand-alone oracle about what the brain "does" for language.
No topics depend on this one yet.