Sensors that glow when neurons speak
The most mature molecular read channel is the genetically-encoded sensor. A calcium indicator (GCaMP) brightens when intracellular calcium rises after spikes; a voltage indicator tracks membrane potential far faster; a neurotransmitter sensor (for dopamine, glutamate, acetylcholine and more) reports chemical signalling that electrodes cannot see at all. All are read out optically, so they inherit the depth limits of Guide 1 and pair naturally with fibre photometry or a GRIN-lens microscope.
An indicator's fractional fluorescence change to an event, with rise τ_on and decay τ_off. Calcium indicators integrate spikes with slow τ (tens to hundreds of ms); voltage indicators are far faster but dimmer — the brightness–bandwidth tradeoff.
A newer, all-electronic variant is the aptamer-based electronic neurochemical sensor: a nucleic-acid aptamer that changes conformation on binding its target, transducing a specific molecule into a current on a field-effect transistor. It promises label-free, implantable, real-time chemical sensing — a genuinely emerging modality, not yet a standard tool.
Recording into DNA
Now the most radical idea in the whole track. What if activity did not stream out at all, but was written into the genome and read back later by sequencing? A DNA molecular ticker tape proposes an engineered polymerase whose error rate rises with local calcium, so the pattern of mutations along a synthesised strand encodes a cell's activity history — a molecular flight recorder in every neuron.
Relatives already work in simpler systems. A CRISPR molecular recorder writes ordered records into a genomic array; activity-dependent transcriptional recorders (integrating a light- or drug-gated calcium signal into gene expression) tag the neurons that were active during a defined window. These are demonstrated proofs-of-concept — mostly in cells, invertebrates and mice, for coarse or offline records — not yet a brain-wide readout.
The information a molecular recorder can hold scales with the number of writable positions L and the bits stored per position b. Vast static capacity is imaginable — but every bit is retrieved only by later, destructive sequencing.
Two meanings of 'recording'
This exposes a deep split. An electrode or a sensor gives you a live stream: activity now, usable for closed-loop control. A molecular recorder gives you a massively parallel archive: potentially every cell at once, but read only after the fact, usually by fixing and sequencing the tissue. Combined with spatiotemporal molecular readout (spatial transcriptomics, in-situ sequencing), it could in principle map activity across a whole structure at cellular resolution — the dream behind whole-brain activity mapping.