The Third Dimension of Light

Conventional holographic storage encodes information in one or two properties of light — typically amplitude (brightness) and phase (wave timing). A research team has demonstrated encoding in all three simultaneously: amplitude, phase, and polarization. The trick is that standard optical sensors can only measure intensity — they are blind to phase and polarization. The team solved this using a convolutional neural network that learns to extract all three dimensions from just two diffraction images captured with different polarizer settings.

The encoding uses a double-phase hologram technique that modulates perpendicular polarization states through a single spatial light modulator. The decoding uses AI: the network learns patterns linked to each light dimension, reconstructing the full three-dimensional information from two-dimensional intensity measurements.

The through-claim is about the relationship between measurement and information. The sensor never sees phase or polarization directly — it sees intensity patterns that are shaped by all three properties. The neural network doesn't add information; it separates information that was already present but entangled in the measurement. Two images, each containing the shadows of three variables, are enough to reconstruct all three — because the shadows from different angles are differently entangled. The information was always there. What was missing was the decoder that could recognize it. Storage capacity doesn't increase because more light is used. It increases because more of what light already carries is read.