Lecture 16: Music

What this lecture covers

This lecture turns from speech to music, asking why humans across every culture create and enjoy it, whether that capacity is an evolutionary adaptation or a byproduct of other brain machinery, and whether music has its own dedicated neural circuitry separate from language. It briefly recaps the previous lecture on hearing and speech before building the case for music as a serious object of cognitive neuroscience study.

By the end you'll understand the classic evolutionary debate over music (Darwin's sexual-selection hypothesis versus Pinker's "auditory cheesecake" view), what developmental and cross-cultural evidence says about music's universality, what brain-damage and neuroimaging studies reveal about amusia and pitch processing, and why a data-driven fMRI study and intracranial recordings provide some of the strongest evidence yet for brain regions selective for music, separate from language and speech.

Key ideas

• The evolutionary question: Darwin speculated music evolved through sexual selection, while Steven Pinker argued it is "auditory cheesecake," a pleasurable byproduct of neural machinery that evolved for other purposes.
• Specialization does not prove innateness: even if a brain region responds selectively to music, that alone does not show the capacity is innate, since learned selectivity (like the visual word form area) is possible.
• Early and cross-cultural evidence for music: infants show sensitivity to beat and relative pitch within days to months of life, and cross-cultural surveys find no single universal feature of music but many common properties, such as a small set of discrete pitches and a regular pulse.
• Perceptual narrowing in rhythm: infants around six months can appreciate both isochronous and non-isochronous rhythms, but by twelve months they favor the rhythms typical of their own culture, paralleling narrowing seen in face and phoneme perception.
• Consonance preference is learned, not universal: a study of the Tsimane', a Bolivian group with minimal exposure to Western music, found no preference for consonant over dissonant sounds, unlike Western listeners.
• Amusia and pitch: brain-damaged and congenital amusia often reflect a broader deficit in pitch-contour perception that also affects speech intonation, though some amusic individuals also show rhythm deficits that complicate a purely pitch-based explanation.
• Group-analysis pitfall: fMRI studies claiming overlapping brain regions for music and language often used group analyses, which can create the appearance of overlap even when no individual subject actually shows it.
• Data-driven discovery of music selectivity: a large-scale fMRI study using unsupervised component analysis on responses to 165 natural sounds found a distinct music-selective component in auditory cortex, separate from a speech-selective component, later confirmed with intracranial electrodes showing highly music-selective and even song-selective responses.

Walkthrough

Recap and why music deserves a lecture (0:10)

After a brief recap of auditory computational problems (the cocktail party problem, reverb) and speech perception, Kanwisher explains why music, unlike speech, is uniquely and universally human, with no close animal analog. She notes its long evolutionary history (40,000-year-old bone flutes), its early appearance in infancy, and its cultural and economic significance as reasons to take it seriously as a subject of study.

Is music an evolved adaptation? (10:16)

The lecture presents competing hypotheses: Darwin's view that music evolved through sexual selection, a related proposal that infant-directed song evolved to manage parent-offspring conflict, and Pinker's contrasting claim that music is auditory cheesecake, repurposing neural machinery that evolved for speech or language. This last claim, Kanwisher notes, is empirically testable, setting up the rest of the lecture.

Developmental and cross-cultural evidence (15:19)

Because fetuses can hear music in the womb, true innateness is hard to establish, so the lecture instead looks at how early musical abilities appear. Two- to three-day-old infants show beat-induction responses in EEG, five- to six-month-olds recognize melodies transposed in pitch, and infants show perceptual narrowing in rhythm appreciation between six and twelve months, mirroring narrowing effects seen earlier in the course for faces and phonemes. A cross-cultural survey of recorded music finds no single universal feature but widespread patterns like discrete pitch scales and regular pulse, with notable exceptions such as certain Papua New Guinean music. A study of the Tsimane' in Bolivia shows that a preference for consonance over dissonance is culturally specific rather than universal.

Brain damage and amusia (37:54)

Patients with acquired or congenital amusia often show more general pitch-contour deficits that extend to speech intonation, suggesting a shared pitch-processing problem rather than a purely music-specific one. However, the lecture also discusses a study showing some amusic individuals have rhythm deficits too, complicating that clean story and leaving the underlying dissociations unresolved.

The pitfalls of group fMRI analysis (52:10)

Kanwisher explains why prior claims of overlapping brain machinery for music and language, based on group-level fMRI analyses, are unreliable: apparent overlap can arise purely from blurring together individually variable regions across subjects. Evelina Fedorenko's within-subject study, which functionally localizes language regions in each individual and tests their response to music, finds no response to music in language regions and no response to language in music-selective regions.

A data-driven discovery of music-selective cortex (59:16)

The lecture's centerpiece describes an unsupervised analysis of fMRI responses to 165 natural sounds, sampled broadly from everyday auditory experience. Without imposing any hypothesis, the analysis extracts six components, most reflecting basic acoustic properties, but including a speech-selective component and a music-selective component with distinct locations in auditory cortex. This music selectivity replicates in people with and without musical training and is confirmed with intracranial electrodes, some showing striking selectivity for music and even specifically for song.

Before you watch

• Watch the previous lecture on hearing and speech, since this lecture opens with a direct recap of the cocktail party problem, phonemes, and primary auditory cortex.
• Recall the concept of perceptual narrowing from the face-perception lectures, since it reappears here for musical rhythm.
• Be familiar with the visual word form area example from the development lectures, since it is used to explain why brain specialization does not prove innateness.

Check your understanding

1. Why doesn't finding a music-selective brain region prove that music perception is innate?
2. What developmental and cross-cultural evidence suggests musical abilities emerge early and are widely, though not perfectly, universal?
3. What did the Tsimane' study reveal about consonance preference, and why does it matter for theories of music?
4. Why are group-level fMRI analyses unreliable for establishing that two functions share the same brain region?
5. How did the unsupervised, data-driven fMRI study differ from earlier hypothesis-driven approaches, and what did it find?