The Screenless Sight Codex

08. Neural Interfaces

Neural interfaces try to introduce information into the nervous system directly or indirectly. In this encyclopedia they are organized by interface point, carrier, and evidence level.

Interface Families

Family	Target	Status	Typical Percept
Retinal prosthesis	Retina	Research	Spots, light patterns, crude form cues.
Optogenetic visual therapy	Modified retinal cells	Research	Partial restored visual behavior with special goggles.
Optic-nerve interface	Optic nerve	Research/speculative	Phosphenes or coarse visual sensations.
Cortical visual prosthesis	Visual cortex	Research	Phosphenes and traced forms.
TMS	Cortex via magnetic induction	Research	Phosphenes under some conditions.
Focused ultrasound	Brain tissue	Research	Modulation or phosphene reports in studies.
Sensory substitution	Touch or hearing	Research	Learned spatial interpretation, not direct sight.

Why “Neural Pixels” Are Difficult

A neuron is not a pixel.
One electrode can affect many cells.
One percept can be shaped by multiple sites.
Tissue changes with time.
Stimulation can spread.
Perception depends on state, attention, and learning.
Natural vision is processed before it becomes conscious.

More Plausible Engineering Direction

The most realistic path is layered:

sensors
-> AI or signal preprocessor
-> compressed task-relevant code
-> calibrated stimulation or cross-modal output
-> user training
-> feedback and recalibration

This is less dramatic than “uploading images,” but far more coherent.

Creative Concepts

Concept	Description
Closed-loop phosphene mapper	The user reports percepts, and the system updates its stimulation map.
Cortical handwriting	Dynamic stimulation traces lines, letters, or shapes over time.
Phosphene icon set	A small vocabulary of reliable visual symbols.
Hybrid residual-vision overlay	Adds artificial cues only where natural vision is missing.
Semantic prosthesis	Uses object detection to communicate meaning through simple cues.
Training-first interface	Treats the device like a new sensory language rather than failed normal vision.

Scope

This module stays at the level of concepts, research summaries, interfaces, encoding strategies, and open problems. The useful contribution is the map: where the signal enters, what code it carries, and what kind of perception it might support.