6/ The biggest lever? The neural data used to fit the model→brain mapping.
Alignment grows log-linearly with the number of stimulus–response pairs, reliably across all modalities — and several datasets show no plateau yet. More neural data still pays off.
