What’s the experiment designed to demonstrate?

In our lab today, we’ll be testing the hypothesis that babies can count as early as five months of age. The six babies here are all less than six months old. You’ll be watching them on closed-circuit TV and measuring their responses.
     The experiment is based on the well-established observation that babies stare longer if they don’t see what they expect to see. First, we’re going to let two dolls move slowly in front of the babies. The babies will see the two dolls disappear behind a screen. Your job is to record, in seconds, how long the babies stare at the dolls when the screen is removed.
     In the next stage, two dolls will again move in front of the babies and disappear. But then a third doll will follow. When the screen is removed, the babies will only see two dolls. If we’re right, the babies will now stare longer because they expect three dolls but only see two.
     It seems remarkable to think that such young children can count. My own research has convinced me that they have this ability from birth. But whether they do or not, perhaps we should raise another question -- should we take advantage of this ability by teaching children mathematics at such a young age? They have great untapped potential, but is it good for parents to pressure young children?