Can subliminals help you get a job? What the tests of subliminal tapes found, and what hidden messages can do

What is a subliminal?

A subliminal message is one presented below the threshold of awareness: a word flashed on a screen for a few hundredths of a second, or, in audio, a spoken sentence mixed so far under music or noise that you cannot make it out. “Subliminals” on video platforms today are mostly the audio kind: a track of rain or music said to carry hidden affirmations, often with a title promising a job, a promotion or an interview that goes well.

The idea goes back to the 1950s and to the self-help tapes of the 1980s. Its appeal is easy to see: if the conscious mind is what doubts, a message that bypasses it should meet no resistance. Joseph Murphy’s 1963 book makes a similar argument about the drowsy state before sleep, without hidden audio; the guide “Joseph Murphy’s bedtime technique, and what sleep research says” looks at that one. The question is whether a message you cannot perceive changes what you do, and the tests answer it fairly clearly.

What the tests of subliminal tapes found

Anthony Greenwald and colleagues ran the best-known test, published in Psychological Science in 1991. They recruited 237 people who wanted the effects that commercial tapes promised, either a better memory or higher self-esteem, and had them use the tapes as the manufacturers recommended for a month. The trick was in the labels: the actual content and the label were varied independently, so some people who believed they had a memory tape were listening to a self-esteem tape, and the reverse. The test was double-blind: neither the volunteers nor the people running the study knew who had which tape. It was run three times.

After a month, neither the memory tapes nor the self-esteem tapes produced their claimed effect. Everyone improved a little on both measures, the same whatever they had heard, which the authors call a nonspecific placebo effect. And more than a third of the volunteers, by our arithmetic at least 80 of the 237, had “the illusion of improvement specific to the domain named on the tape’s label”: they felt the tape had worked on what the label said, whatever it actually contained.

Where the idea of subliminals came from

The founding story is a hoax. In 1957 the marketer James Vicary announced that flashing “Eat Popcorn” and “Drink Coca-Cola” during a film had raised popcorn sales by 57.7% and Coca-Cola sales by 18.1%. As a 2009 research paper puts it, “By 1962 at the latest, this was exposed as a publicity hoax.” The story still circulates.

In the 1980s, John Vokey and Don Read tested the other popular claim of the time, messages hidden in pictures or played backwards in songs. In their 1985 review in American Psychologist they found no evidence that such messages affect what people do, across a range of tasks, and argued that the “backward messages” people heard in rock songs owed more to the listeners than to the recordings.

What subliminal priming can do in the lab

Something does get through. In a 2009 meta-analysis in Psychological Bulletin, Eva Van den Bussche and colleagues pooled studies of masked priming, where a word is flashed too briefly to be seen and then hidden by a pattern. They found a significant effect: a hidden word can speed up or slow down how fast people sort a related word that follows, and some of that processing is about meaning. These effects are measured in milliseconds, on a task that follows within a second or so.

One of the best-known cases of a behavioural effect is a 2006 study by Johan Karremans, Wolfgang Stroebe and Jasper Claus. During a visual task, the brand name “Lipton Ice” was flashed 25 times, each time for about 23 milliseconds (about 0.6 seconds in all, by our arithmetic). Afterwards, participants chose between Lipton Ice and a mineral water. The hidden brand name raised the choice of Lipton Ice, but only among participants who were thirsty. A 2009 team found the same pattern with tired participants and dextrose tablets: only the tired ate more of the brand they had been shown.

Read the conditions: a need the person already had, a choice between two products later in the same session, inside the lab. A job is the reverse on every count: weeks of applications, calls and interviews, decided by other people who never hear your recording.

Why subliminals feel like they work

The tape test answers this too. People who choose a subliminal want the change it names, and they expect it. After a month, they notice the moments that fit the label and credit the tape: in 1991, more than a third did, whatever was really on it. Add the base rate of a job search, where many people find work in any given month whatever they listen to, and every job found after a playlist looks like a result; the guide on the technique that “got me my dream job” puts a number on that base rate.

Nothing here says the listening is harmful. A calm track at night can be pleasant. The problem is the hidden part: if the content made no difference in a double-blind test, paying for it, or spending your evening on it instead of on a step you could take, buys a label.

What to listen to instead

If you like hearing something as you fall asleep, make it something you can hear. Dream Architect has one voice option, “Read the moment aloud”, off by default: when the screen goes dark, your phone’s own voice reads your moment once, slowly, and nothing is sent (read in the app’s code and settings on 22 October 2026). It is the opposite of a subliminal: your own words, audible, once.

Here is an illustrative example, not a real person. At “The moment”, write a step you will take tomorrow: “Wednesday, 10 a.m. I open the job post, write two lines on why I fit, and press send.” Pick an emotion at “The feeling”, then let the screen go dark. The night before an interview, the mental rehearsal plan in “Job interview confidence: a mental rehearsal plan for the nerves” goes further.

How we checked

We read the abstract of the 1991 tape study, the 2009 meta-analysis and the 2006 priming study, and the full text of the 2009 replication, which also reports the 1957 claim and its exposure, on 22 October 2026. The “at least 80 of 237” is our arithmetic from the paper’s “more than a third”; the 0.6 seconds is 25 flashes times about 23 milliseconds, as that replication reports the earlier study’s timing. We searched the same day for a controlled study of subliminal audio for jobs or careers and found none. App details come from its code and settings the same day.

Questions people ask next