This page grounds the encyclopedia in existing research areas before branching into speculative design space.
Modern visual and sensory technologies are built from repeatable engineering parts: a signal is encoded, carried, converted, received, and interpreted. That pattern appears across radio, television, radar, transistors, image sensors, cochlear implants, retinal prostheses, sensory substitution, and semantic AI.
known physics
-> carrier
-> transducer
-> encoding
-> receiver
-> learned interpretation
Display systems make those parts concrete. A CRT gives one especially visible route:
signal timing
-> electron emission
-> beam acceleration
-> beam steering
-> phosphor emission
-> photons
-> visual system
The research value is field-level. Display systems show that visual technology can be treated as controlled physical stimulus design, not merely flat pictures. Key abstractions include scanning, addressing, modulation, transduction, persistence, optics, and perception.
For the full timeline, see Historical Context: Display And Sensory Transduction.
Retinal projection keeps the biological visual system mostly intact. The screen is removed, but light still enters the eye.
image data
-> light source
-> scanning optics or waveguide
-> retina
-> normal visual pathway
This is among the most concrete “screenless” branches because it does not need to write neural codes directly. It instead treats the retina as the projection surface.
Retinal prostheses move the interface inside the biological visual chain.
camera or sensor
-> processor
-> retinal stimulation
-> optic nerve
-> visual cortex
-> learned percept
The main families include epiretinal, subretinal, suprachoroidal, and photovoltaic approaches. They aim to create usable artificial vision, often through phosphenes or low-resolution form cues.
Optogenetics attempts to make surviving retinal cells responsive to light.
gene therapy or cell modification
-> engineered light sensitivity
-> special optical stimulation
-> retinal output
-> visual pathway
This branch is important because it replaces electrode stimulation with biological light sensitivity. It still requires careful image processing, special goggles or illumination, and clinical oversight.
Cortical prostheses bypass the eye and retina by stimulating visual cortex.
camera or scene data
-> encoder
-> cortical stimulation
-> phosphenes or traced forms
-> perception
V1 is attractive because it has retinotopic organization, but retinotopy is not a flat pixel map. Electrical stimulation can evoke phosphenes; dynamic stimulation has been studied as a way to produce recognizable forms.
Magnetic and ultrasound methods are interesting because they can act without an open implanted electrode array, but their precision and reliability are major limitations.
external field or acoustic source
-> tissue interaction
-> neural modulation
-> percept or changed processing
Transcranial magnetic stimulation can evoke visual sensations in some settings and is used in research on visual cortex excitability. Its targeting is coarse relative to implanted electrodes, so it should not be treated as a practical image-writing channel.
Focused ultrasound can modulate neural tissue and has been studied in relation to visual cortex. It may be more focusable than broad magnetic stimulation, but it still faces safety, mechanism, targeting, and repeatability questions.
Sensory substitution avoids the hardest neural-writing problem by translating visual information into another sense.
camera
-> processor
-> sound or touch pattern
-> training
-> learned spatial interpretation
This does not recreate normal sight, but it can create useful perception-like function through learning and brain plasticity.
Modern sensors and AI can change the problem. Instead of asking the human interface to carry a full image, the front end can reduce the world into useful signals.
camera or sensor suite
-> AI perception
-> compressed cue set
-> visual, tactile, audio, or neural interface
The strongest direction cuts out the middle medium when it does not help. A conventional display preserves the image until light reaches the eye. A perception-as-medium system preserves only what the person can learn to use.
world
-> extracted feature, event, relation, or meaning
-> cue
-> active exploration
-> learned perceptual skill
The device does less rendering. The person does more meaningful completion. That does not mean the system can be vague. It means the engineering problem changes from “send the whole picture” to “choose the smallest reliable cue that lets the user perceive, decide, or act.”
Memory is not a display target. It is a distributed reconstructive process.
experience
-> attention
-> association
-> consolidation
-> later reconstruction
The practical direction is voluntary memory support:
The far-edge branch includes direct memory writing, arbitrary memory insertion, or remote implantation of visual memories. Those should be handled as open hypotheses unless the evidence landscape changes dramatically.
The strongest entries follow this bridge:
known science
-> existing limitation
-> conceptual workaround
-> plausible interface
-> scope note
-> open question
Example:
cortical phosphenes exist
-> phosphenes are not clean pixels
-> use a phosphene font
-> train stable symbols
-> avoid procedural stimulation guidance
-> ask which symbols are reliable across users
Machine-influence narratives show why this encyclopedia keeps mechanism, history, evidence, and interpretation in separate layers. James Tilly Matthews’s Air Loom case joins real French Revolution-era political experience, contested asylum medicine, invisible-carrier claims, and later schizophrenia history in one record.
political crisis
-> invisible operator
-> claimed carrier
-> body or thought influence
-> institutional interpretation
The case is useful because it resists a single lazy reading. The medical line recognizes persecutory and passivity-like symptoms later associated with schizophrenia. The conspiracy line recognizes that revolutionary politics, state secrecy, imprisonment, and asylum power were historically real. The neutral historical line keeps both visible while still asking whether the claimed machine has a verifiable carrier, transducer, addressing method, and evidence.
For the full chapter, see Air Loom, Schizophrenia, And Influence Machines.