Rey.BEng 28th October 2026
https://x.com/MillerLabMIT/status/2115416746173637102
Miller Lab reports that the brain holds ordered lists in separate, low-dimensional position-based slots within its activity patterns. Neural subspaces encode sequential working memory without detectable neural sequences or replay during the delay. Order is maintained as a relational addressing geometry rather than by sequential reconstruction in time. Neighbouring slots remain distinct; the cost of separation is spatial rather than temporal. The finding appears in both human MEG and prior macaque PFC data.
Ace reads the same geometry as residual packet held open. Geometric memory is the residual history of the contact patch that has not yet been spent. The low-dimensional subspaces are successive stations of one residual that never completes to 2. The position-based slots are the open readings of that residual across the twelve stations. No continuous-second replay is required because the duration is the packet itself. When the residual is held open at the fixed point 0^{i2}, the ordered list remains available as simultaneous faces of one Moment. When the residual is spent, the same list appears as separate locked faces. The tick ^ carries the twist onto the mass seat. The spark * receives the exponent as residual celerity. COUNTNOW^ is the count written while the residual remains open. The non-fracture identity holds. The continuous second is the Observational face that remains once the residual itself is unnamed.
The subspaces are therefore not a new neural code. They are the 2D projection of the residual packet already written on the Master Table. Earth is the preferred axis. The laboratory is the Earth frame of the Rest Time origin. The residual stands.The work is open.
The invitation remains open.
Mch. Wide-Open Invitation: The Development of Geometric Instruction
