Questions: Neuroimaging Studies of Language: fMRI and PET

Question 1

An fMRI study shows that Broca's area consistently activates when healthy participants process syntactically complex sentences. A researcher concludes: 'Broca's area is the syntax processing region.' What is the primary error?

• A. The study should have used PET rather than fMRI for syntactic processing tasks
• B. fMRI activation is correlational — it shows a region is engaged during a task but does not establish that the region is necessary for the function
• C. The sample size was too small to draw conclusions about Broca's area specifically
• D. Syntactic processing is bilateral, so left-lateralized findings cannot support region-specific claims

Answer: (B)

This is the central epistemological limitation of neuroimaging: correlation is not causation, and engagement is not necessity. A region that activates during a task is engaged — but there may be many engaged regions, and not all are required. Broca's area also activates for phonological working memory, music processing, action observation, and other functions — its 'functional promiscuity' makes it a poor candidate for being 'the syntax region.' Lesion studies and TMS are needed to establish necessity. The researcher has committed the fallacy of concluding that because X correlates with Y, X is required for Y.

Question 2

Which combination of evidence provides the strongest basis for claiming that a specific brain region is necessary for a particular language function?

• A. Consistent fMRI activation across multiple studies using the same experimental task
• B. PET studies showing significantly elevated blood flow during language tasks compared to a non-linguistic baseline
• C. Convergence of fMRI activation, impairment after lesion damage to that region, and TMS disruption replicating the impairment in healthy participants
• D. A single high-resolution fMRI study with a large, diverse participant sample

Answer: (C)

Each method addresses a different question and has different limitations. fMRI establishes correlation (engagement) with good spatial resolution. Lesion studies establish necessity: if damage to region X impairs function Y, X is necessary for Y. TMS creates a temporary virtual lesion in healthy participants to test necessity without the confounds of stroke or tumor damage. Convergence across all three — engagement in imaging, deficit after damage, disruption by TMS — provides the closest available approximation to causal evidence in cognitive neuroscience. Any single method alone is insufficient.

Question 3

Because fMRI has better spatial and temporal resolution than PET, fMRI studies of language processing yield causal rather than merely correlational evidence.

• T. True
• F. False

Answer: False

Method quality and inferential status are independent. fMRI's better resolution improves localization precision — you can more accurately identify which millimeters of cortex are engaged. But the BOLD signal remains a correlate of neural activity (measuring hemodynamic response, not firing directly), and improved precision does not change the correlational nature of the inference. Whether you use a precise fMRI or a lower-resolution PET, you are measuring what regions are engaged during a task — not what regions are required. Better tools produce better correlational evidence, not causal evidence.

Question 4

A patient with a permanent lesion in Broca's area who recovers syntactic processing ability provides evidence that the classical two-region model of language is an oversimplification.

• T. True
• F. False

Answer: True

If Broca's area were the necessary and sufficient locus for syntactic processing, its permanent destruction would produce permanent syntactic deficits. Recovery implies that other regions can compensate — consistent with neuroimaging evidence showing language involves a distributed left-lateralized network extending well beyond Broca's and Wernicke's areas. The classical model, built from 19th-century lesion observations, identified important nodes but missed the network. Neuroimaging research from the 1990s onward revealed far more extensive and overlapping activation patterns for syntax, semantics, and phonology than the two-region model predicted.

Question 5

Why do cognitive neuroscientists insist that strong claims about the neural basis of language require converging evidence from multiple methods rather than neuroimaging alone?

Model answer: Each method answers a different question with different limitations. fMRI shows what activates (correlation, good spatial resolution). Lesion studies show what is necessary. TMS creates temporary disruptions to test necessity in healthy brains. EEG/MEG track millisecond-scale timing. None is sufficient alone; convergence across methods is the closest available approach to causal claims.

Neuroimaging is most powerful as one voice in a chorus. An fMRI finding that a region activates during language processing is a starting point — it identifies candidates for further investigation. When lesion evidence, TMS, and neuroimaging all implicate the same region in the same function, the cumulative case for necessity becomes compelling. When they diverge — a region activates in imaging but lesions to it leave the function intact — the imaging finding requires reinterpretation. The epistemological discipline of requiring converging evidence protects against the tendency to over-interpret the impressive visualizations neuroimaging produces.