Something wasn’t right. Every time I pressed the button, the computer responded twice, spitting out two lines of text instead of one. I triple-checked my code, confirming that I had only written one command for the computer to display the words “button pressed.” So, what then? Was I accidentally pressing the button twice, somehow? The four-by-five foot experiment room suddenly felt very quiet and very dark.
It was my third month at the University of Sydney, where I had come on a fellowship to study visual perception. With help from a professor there, I had designed an experiment to measure how well people can synchronize finger movements with visual events. The idea was to have subjects watch a white bar slide across a computer screen and press a button as soon as it became aligned with a stationary bar. I wanted to measure how early or late their finger movements would be, depending on the brightness of the bar. To help me do that, another lab had lent me a special button-box that precisely records the time its button gets pressed.
After a week of trial-and-error computer programming, I had cajoled the button-box into talking to my computer and started collecting data. The pattern of results from six subjects was fascinating and unexpected. Just an hour earlier, at lunch, I had been talking with my colleagues about what the results meant, what to do next, and to which high-profile journal I should send the manuscript. After that, I had walked casually back to the lab to modify the code for a new experiment, which was why I had asked the computer to display a message whenever the button was pressed.
Now, sweating in the tiny windowless room, I could hear faint murmurs of people talking in the office outside. Do they know I’m in here? Can they hear my heart pounding through this insulated door? I forced myself to focus on the problem at hand. What if…I pressed the button and held it down, and the computer said, “button pressed.” When I released the button, it said again, “button pressed.” Oh no.
The box was sending a signal when the button was pressed down and when it was released. A few more minutes of frantic work revealed that the computer could not tell these two signals apart, so it was overwriting the first with the second. That meant that all my data were not measurements of when subjects pressed the button relative to the visual event, but of when they happened to release pressure on the button.
I stayed hidden for as long as I could, listening to people come and go outside. The thought of telling them that it was all a mistake and the results were not accurate made me ill. We would have to start over from scratch. I was already worried that I had over-sold myself to win this fellowship. And the results were too good to be true. Now they’ll know.
But…do I have to tell them? Couldn’t I just quietly fix the bug and go on collecting data? Surely the timing of button presses is correlated with the timing of button releases. The general pattern would be the same, and that’s what matters. Right?
In my brief, delirious dance with the devil of scientific dishonesty, I couldn’t decide whether the possibly inconsequential nature of the mistake was more or less of a reason to hide it.
I suddenly snapped out of it and jogged down to my professor’s office. “I have some bad news,” I began, before he even looked up from the papers on his desk. He was grave at first, but then he managed to laugh at the panic in my eyes and calmly instructed me to fix the code and start collecting new data right away.
Two months later I had a fresh set of data showing the same interesting pattern as the first, flawed set. This internal replication of the finding ended up making me feel more confident about submitting it for publication. I had discovered something about the brain that was, at least in some limited sense, true, independent of my youthful eagerness to succeed. The day the results were published I felt that science was righteous.
But the day I discovered my mistake, the professor had assured me that all scientists make that kind of goof at some point. Yeah, I remember thinking, but how many feel the temptation to cover it up, and how many actually do?
Alex White got his Ph.D. in Psychology from New York University in 2013. He is now a postdoctoral researcher at the University of Washington. He still studies visual perception, and still watches carefully for his own inevitable mistakes.