Auditory Processing
Your Brain Is Writing the Music Before It Arrives
The sound you think you just heard is mostly a prediction your brain made a fraction of a second ago.
The Idea
Hearing feels passive — sound waves enter, you perceive them. But the auditory cortex doesn't simply receive information; it actively anticipates it. The brain is running a continuous generative model of the world, constantly predicting what sound should come next based on context, memory, and pattern. When reality matches the prediction, very little neural processing is actually triggered. It's only the surprise — the deviation — that demands real attention. This is known in neuroscience as predictive coding, and it reshapes how we understand perception fundamentally. You are not hearing the world; you are hearing your model of it, corrected in real time by incoming data. The implications get strange quickly. Because your brain is completing patterns before all the acoustic information has arrived, what you perceive can diverge from what was physically produced. Phonemes you've already heard alter your perception of phonemes you haven't yet heard, a phenomenon called the phonemic restoration effect. Drop a single syllable from a sentence and replace it with a cough — listeners don't notice the gap. They hear the full word. The brain simply fills it in. More unsettling: the version of a sound you consciously experience is slightly backdated, smoothed into a coherent narrative that never quite happened that cleanly in the real world. Listening, it turns out, is an act of creative reconstruction, not passive recording.
In the World
In 1970, psychologist Richard Warren conducted a deceptively simple experiment at the University of Wisconsin-Milwaukee. He took a recording of the sentence 'The state governors met with their respective legislatures convening in the capital city,' excised a single phoneme — the 's' in 'legislatures' — and replaced it with the sound of a cough. He then played it to participants and asked them to identify where the cough occurred. They couldn't. Not only did listeners fail to locate the cough; many insisted they had heard a perfectly intact, uncoughed sentence. The brain had received no acoustic information for the missing sound and yet reported hearing it clearly. Warren called this the phonemic restoration effect. The auditory system, fluent in the statistical regularities of language, simply voted to insert the most likely phoneme and erased the cough from conscious awareness entirely. Decades later, brain imaging confirmed what Warren had inferred behaviourally: the auditory cortex genuinely activates for sounds that weren't there, as though it heard them. Neurologist David Poeppel and colleagues have traced how the brain uses multiple timescales of oscillation — rapid gamma waves tracking individual phonemes, slower theta waves tracking the rhythm of speech — to synchronise its predictions with incoming sound. The whole system is tuned less like a microphone and more like a musician who already knows the score, following along and filling in any notes the orchestra drops.
Why It Matters
Once you understand that hearing is predictive, a lot of ordinary experience becomes newly legible. Why do you struggle to understand someone with a strong unfamiliar accent? Your model hasn't updated yet — the prediction keeps misfiring. Why does a familiar song sound richer than an unfamiliar one, even the first time you hear it again after years? The predictions are landing cleanly, creating a satisfying sense of fit. Why do mishearing lyrics stick so stubbornly — singing the wrong words even after you've been corrected? Because your prediction engine has updated its local model and keeps overriding the acoustics. It also matters beyond the personal. People who live in noisy environments, or who have hearing loss, aren't simply receiving less information — they're running a prediction system on degraded input, which is cognitively expensive. The fatigue that hard-of-hearing people describe isn't about effort in the way we usually imagine it. It's the cost of a generative model working overtime. Knowing this changes how we might think about accessibility, about design, about what we owe each other in terms of acoustic environments. The ear is just the beginning of a very long story that mostly happens inside.
A Question to Ponder
If so much of what you hear is your brain's prediction rather than the sound itself, what does it mean to truly listen to someone?
Get a new one of these every morning.
Start learning with Thinkable