In the late 1990s, a Princeton research lab created software that let ordinary people attempt the impossible — influencing a computer with the power of thought. It disappeared. Twenty years later, someone went looking for it.
Late at night, somewhere in the Nevada desert, a radio host changed what millions of people thought was possible.
For anyone who grew up listening to late-night radio in the 1990s, the voice of Art Bell was the sound of the unknown. Broadcasting from his home studio in Pahrump, Nevada, Bell hosted Coast to Coast AM — a four-hour nightly conversation about the edges of human experience. UFOs, remote viewing, time travel, consciousness research. Bell treated all of it with the same combination of genuine curiosity and dry humor that made millions of listeners trust him.
One night, Bell had a guest on the program discussing a piece of software unlike anything available to ordinary consumers. It was called ShapeChanger. The premise was extraordinary: you sat in front of your computer, you concentrated on the screen, and the program would measure whether your focused attention could actually influence which of two images the computer displayed.
Not trick the computer. Not click buttons. Influence it with your mind.
Bell was intrigued enough to try it himself. He reportedly found his hit rate — the percentage of random events that went in his chosen direction — consistently running between 85 and 95 percent when he was sitting in front of the screen. When he left the room and tried to concentrate from another location, the effect vanished.
For the listeners who heard that broadcast, something changed. This wasn't a psychic hotline or a carnival trick. This was a piece of software developed by a Princeton University research laboratory, built on decades of rigorous scientific data, sold for $24.95. It was within reach.
The most controversial laboratory in Ivy League history ran a simple experiment for nearly three decades — and never quite got the answer it expected.
The Princeton Engineering Anomalies Research laboratory — known as the PEAR lab — was founded in 1979 by Dr. Robert Jahn, Dean of Princeton's School of Engineering. Jahn was not a fringe figure. He was an aerospace engineer, a respected academic, and a man who took his scientific obligations seriously. When a student brought him a set of anomalous data suggesting human intention might influence physical random processes, Jahn didn't dismiss it. He built a laboratory to investigate it properly.
For 28 years, the PEAR lab ran a deceptively simple experiment. Human subjects sat in front of a random event generator — a device that produced a genuinely random sequence of bits (ones and zeros) using quantum processes — and tried to mentally influence whether the device produced more ones or more zeros. They weren't allowed to touch the device. They didn't press buttons. They simply concentrated.
The results, accumulated across millions of trials and hundreds of operators, showed a small but remarkably persistent deviation from pure chance — in the direction the operators intended.
— Summary of PEAR Lab findings, Princeton UniversityThe effect size was tiny. We are talking about a deviation of roughly one part in ten thousand from what pure randomness would predict. It would be invisible in any individual session. But accumulated across 2.5 million trials, with careful controls for experimenter bias, selective reporting, and statistical artifacts, the deviation persisted. It was never large enough to satisfy the scientific mainstream. It was never small enough to dismiss.
The PEAR lab closed in February 2007 — not because its data was discredited, but because Dr. Jahn and his longtime collaborator Brenda Dunne felt they had asked the question as rigorously as they could, and that the answer the data gave deserved to be absorbed by the wider scientific community before more resources were spent. The question is still being asked. The answer is still contested.
In 1997, the science came home. A Princeton spinoff company built a consumer version of the PEAR experiment that anyone could download for $24.95.
Mindsong Inc. was the commercial arm of PEAR Lab's work, led by Dr. John E. Haaland. ShapeChanger was its flagship product — a consumer software application that translated the lab's abstract random-number experiments into a visceral visual experience.
The interface was simple and striking. A split screen displayed two images — the original version used a wave and an image of the Earth as seen from space. The images slowly morphed into each other based on a hardware random number generator. The user chose one image as their target, concentrated on the screen, and watched to see whether the morph tracked toward their chosen image or away from it.
After a timed session — typically three minutes — the software delivered a score. Not just a hit percentage, but a message with personality: "Nice Try!" or "Resonating!" The first run of 1,000 copies sold out. The program became a fixture in the Coast to Coast AM community, where thousands of Art Bell listeners downloaded it and sat in front of their monitors trying to move pixels with their minds.
Then, quietly, it disappeared. Mindsong Inc. closed. PEAR Lab closed in 2007. The website went dark. The trademark expired. The installer — a Windows executable from the mid-1990s — stopped circulating. Within a decade, ShapeChanger had become a piece of lost software history: something thousands of people remembered but nobody could find.
One Art Bell listener heard that broadcast, downloaded ShapeChanger, played with it briefly — and then lost it. What followed was two decades of intermittent searching that went nowhere.
In a single conversation with Claude (Anthropic), a 20-year search ended — not by finding the original, but by rebuilding it from memory, description, and historical records.
What Claude built wasn't a copy of ShapeChanger. The original source code is gone and the original executable can't be safely recovered. What it built was a faithful recreation — the same concept, the same experiment, implemented with modern tools. Instead of a hardware quantum random number generator, MindMorph uses the browser's cryptographic random number source, seeded from operating system entropy. Instead of a Windows installer, it's a single HTML file that runs in any browser on any device.
The science behind it is identical. The question it asks is the same one the PEAR lab spent 28 years investigating: can focused human intention cause a statistically measurable deviation in a random process? Most sessions will show nothing. That's how it should be. But the data, accumulated honestly across many sessions, might tell you something. Or it might not. That's the experiment.
Choose an image pair. Choose your target. Concentrate. Let the data tell you what it finds.
No data leaves your browser. No account required. No network connection needed after first load. MindMorph runs entirely on your device using your browser's cryptographic random number generator.