The code — capacity is the theta/gamma ratio

The chain's strongest result: working-memory capacity equals the theta/gamma ratio — about 7±2 items — because each gamma cycle nested in one theta cycle holds one item. This is the one link with causal validation: manipulating theta with tACS changes realised capacity in the predicted direction. Everything downstream is built on this.

The strongest result in the chain. Working-memory capacity equals the theta / gamma ratio — about 7±2 items — because each gamma cycle nested inside one theta cycle holds one item. Gamma crosses the encode threshold at lower energy and so carries more information per cycle (~19×), and a binding read carries ~0.65 bits per channel when on versus ~0 when off. This is the one link that reaches causal validation: manipulating theta with tACS changes the realised capacity in the predicted direction. Everything downstream is built on this.

The code

The code is theta-nesting. A theta cycle is a frame; the gamma cycles packed inside it are slots; each gamma packet holds one item. To hold a list is to fill the gamma slots of one theta frame in order. This is the structure the §3 coupling makes available — a slow carrier with fast packets — now read as a memory buffer.

Why gamma

Gamma is not arbitrary. Crossing the encode threshold (§5) costs energy, and a gamma packet crosses it at lower energy than a slow one would, so each gamma cycle carries more information for the same metabolic price — on the order of 19× more information per cycle in the combined threshold-plus-energy read. A binding read makes the same point in bits: ~0.65 bits per channel when the binding is on versus ~0 when it is off. Gamma is the efficient carrier.

Capacity = θ/γ → 7±2

If a theta frame holds as many items as it has gamma slots, then capacity is simply the ratio of the two periods, θ / γ. Put in the measured band values, that ratio is about 7±2 — the classic working-memory span, here derived rather than asserted. The famous number falls out of the two clocks.

The causal test

This is the one link that goes past model and direction to causal. If capacity is the θ/γ ratio, then changing theta should change capacity. Driving theta with transcranial alternating-current stimulation (tACS) does exactly that: the realised span moves in the predicted direction when theta is manipulated. This is the hardest evidence in the paper, and it is why the chapters downstream are allowed to lean on the code — while remaining, themselves, model or direction (§9).