Living-neural computing experiments can be described along four independent axes: the biological substrate, the physical interface, the learning mechanism and the role assigned to the biological network. Keeping those axes separate makes comparisons much clearer.
Biological substrate
2D neuronal cultures
Brain organoids
Engineered 3D neural cultures
Neuron-derived algorithms
Interface
Planar HD-MEA
Microfluidic patterning
3D embedded electronics
Conformal organoid arrays
Optical stimulation
Learning route
Reservoir readout training
Closed-loop neural plasticity
Supervised feedback
Reinforcement-style feedback
Algorithm discovery
System role
Classifier
Temporal generator
Adaptive controller
Reservoir / nonlinear transform
Research instrument
Examples across the map
Brainoware combines an organoid, an electrode interface and reservoir computing. The 2026 temporal-pattern BNN combines a modular planar culture with microfluidics, a high-density MEA and supervised closed-loop feedback. TBC's Neural Optimizer sits elsewhere on the map because the biological experiment is used for algorithm discovery while the deployable artefact is software.
This decomposition also exposes missing combinations. A future commercial system might pair a standardised 3D culture with dense bidirectional I/O and a stable developer abstraction that hides much of the biological variability.