Module 1 — Neural codes: the format of representation

arabic african folk, dakar new wave, dreamy soul · 4:41

Listen on 93

Lyrics

[Verse 1]
In your skull there's a billion neurons firing away
Each one's got a story, but not what scientists used to say
No grandmother cell for grandma's face alone
It's populations dancing, codes they've always known
Quiroga found the truth in twenty-oh-five
Jennifer Aniston neurons, sparse but still alive
Not one cell screaming "That's the star I see!"
But clusters whispering in distributed harmony

[Chorus]
Neural codes, neural codes, how does information flow?
Rate and time and spike patterns, there's so much we need to know
Sparse or dense, the tradeoffs dance, capacity meets the cost
In the manifold geometry, no signal should be lost
Neural codes, neural codes, mixing high dimensionally
Population doctrine rising, that's the brain's reality

[Verse 2]
Rate codes count the spikes per second, slow but reliable
Temporal codes use precise timing, fast and quite pliable
Spike-timing rides the millisecond waves of neural chatter
Each format carries different loads, and bandwidth's what matters
Sparse codes save on energy, few neurons active at once
Dense codes pack more information, but metabolism confronts
The tradeoff between efficiency and what the brain can hold
Decodability versus cost, the story that unfolds

[Chorus]
Neural codes, neural codes, how does information flow?
Rate and time and spike patterns, there's so much we need to know
Sparse or dense, the tradeoffs dance, capacity meets the cost
In the manifold geometry, no signal should be lost
Neural codes, neural codes, mixing high dimensionally
Population doctrine rising, that's the brain's reality

[Bridge]
Mixed selectivity's the secret sauce that Rigotti revealed
Prefrontal cortex neurons mix variables unrepealed
Nonlinear combinations in the high-dimensional space
Give flexibility for reasoning, cognitive interface
Why neurons mix, Fusi explained the theory behind
Dimensional expansion lets the cortex redesign

[Verse 3]
Representational similarity maps the neural terrain
Kriegeskorte showed us how to measure what the patterns contain
Activity flows on manifolds, geometry defines
How concepts sit in neural space, mathematical designs
Chung and Lee proved capacity lives in the manifold's shape
Saxena synthesized the doctrine, population landscape
From single cells to群体codes, the paradigm has shifted
Neural space holds all the answers, dimensionally gifted

[Chorus]
Neural codes, neural codes, how does information flow?
Rate and time and spike patterns, there's so much we need to know
Sparse or dense, the tradeoffs dance, capacity meets the cost
In the manifold geometry, no signal should be lost
Neural codes, neural codes, mixing high dimensionally
Population doctrine rising, that's the brain's reality

[Outro]
In the format of representation, populations reign supreme
Mixed selectivity and manifolds fulfill the neural dream

← Module 0 — Framing: what counts as an answer | Module 2 — Concepts and semantic memory: how meaning is stored →