Researchers at Oregon State University have developed a new light-sensitive electronic device that works more like the human brain.
The innovation could lead to artificial intelligence systems that process information faster while using much less electricity.
The new device, described in the journal Advanced Functional Materials, combines three important functions into a single component: it can sense light, store information and process signals.
According to the researchers, bringing these functions together could make AI systems much more efficient.
Today’s AI hardware typically handles these tasks using separate components. Information must constantly move back and forth between sensors, memory units and processors.
This movement requires a great deal of energy and slows down the system.
The new device takes a different approach. Instead of separating these tasks, it performs them in one place, similar to how the human brain works.
Professor Larry Cheng, an electrical engineering and computer science researcher at Oregon State University, says the technology introduces a new way to process information directly where it is detected.
The device is called a phototransistor, meaning it responds to light. When light shines on it, electrical charges are created and stored inside the device. These stored charges act like memories.
What makes the technology especially interesting is that the memories are not fixed. Their strength can be electronically adjusted, much like memories in the human brain become stronger or gradually fade over time.
In the brain, chemical signals help determine which memories are reinforced and which are forgotten. The new device mimics this process. By applying a small electrical signal, researchers can control whether the stored information lasts longer or disappears more quickly.
This ability to manage memory strength is considered an important step toward neuromorphic computing, a field that aims to build computer systems that function more like biological brains.
The device is made by combining two different materials. One material, called an oxide semiconductor, carries electrical current. On top of it is an organic light-sensitive material that absorbs light and produces electrical charges.
Some of these charges become trapped inside the light-sensitive layer. Even after the light is turned off, the trapped charges continue to influence the electrical current flowing through the device. In effect, the device remembers that it has previously seen light.
Researchers can then control this memory by moving the trapped charges closer to or farther from the pathway through which electrons travel. Bringing the charges closer strengthens their effect and allows the memory to last longer. Moving them farther away weakens the effect and causes the memory to fade more quickly.
The researchers say this programmable memory lifetime could create a flexible time window for processing visual information and other sensor signals directly at their source.
In the future, this capability could lead to more energy-efficient computer vision systems, smarter sensors and AI technologies that think and respond more like the human brain.
