Many people believe that listening to Mozart makes babies smarter. But the first study in 1993 did not say this. It tested 36 college students, not babies. After they listened to Mozart for 10 minutes, they did a little better on a puzzle about shapes.

But the effect went away after about 15 minutes. Later studies found that the effect came from a good mood, not from Mozart's music itself.

In the 1990s, many parents played Mozart to their babies, hoping to make them smarter. The idea came from a one-page paper published in Nature in 1993. That study tested 36 college students and no babies at all.

After ten minutes of a Mozart sonata, the students did slightly better on spatial reasoning tasks than after silence or relaxation instructions. The effect disappeared within about 10 to 15 minutes. As the news spread, the story grew. The students turned into babies, and a short effect turned into lasting intelligence.

Later reviews found that the effect was small and not special to Mozart. A 2001 study offered a likely reason. Cheerful, fast music improved scores, but slow, sad music did not, so the boost seemed to come from a better mood.

By the late 1990s, shops were full of recordings that promised smarter babies, and one US state even proposed giving classical CDs to the families of newborns. The source of all this hype was a 1993 paper by Rauscher, Shaw and Ky, and it claimed very little. Thirty-six college students listened to a Mozart sonata for ten minutes and then did better on spatial reasoning tasks than they did after silence or relaxation instructions.

The advantage faded within about fifteen minutes. The paper said nothing about infants or general intelligence. Other labs had mixed results when they repeated the experiment. In 1999, Chabris pooled the available studies and concluded that any effect was small, worth perhaps one or two IQ points, and a 2010 meta-analysis found nothing unique about Mozart. The design had a flaw.

Because the comparison conditions were silence and relaxation, the study could not separate a specific Mozart effect from a general lift in arousal and mood. Thompson and colleagues tested exactly that in 2001. Upbeat Mozart improved performance and slow, sad music did not, and once mood and arousal were taken into account, most of the difference disappeared.