In a video that's circulating (https://www.youtube.com/watch?v=4o-VplYrqBs), a 10-month baby responds to her mother's sad song with tears. Tears collecting in the eyes, then sliding down the cheeks. Sadness isn't one of the 7 pure affects that Sylvan Tomkins recorded in 6-weeks-old babies, but this sure looks pure to me. Enjoyment, one of Tomkins' affects, is visible on the baby's face at the beginning of the song. She attempts to imitate 3 or 4 of her mother's words with her lips. Then the tears begin to collect.
We don't think of babies being sad. Red-faced crying when they're hungry, shrieks when something hurts, but not tears of sadness. But our emotions are sensory responses to our experiences, not cognitive responses. The mother sings the song's sadness, wraps the baby in her own emotional field, the baby responds. As adults, we don't always know why we cry. An image, a song, a few words of kindness, and the tears appear. We don't decide to cry. We don't "think" about the reasons for our crying unless we're asked.
We don't think of babies being lonely. One of my grandsons was adopted when he was nine-months old. When he was four, his mother was pushing him on a swing, higher and higher. They were still laughing as they left the playground, hand in hand. He said, "I was so lonely until you found me."
That emotional responses can't be tested in pre-verbal children doesn't mean that they're not being felt and remembered. That we don't have reasons for our own emotional responses doesn't mean that they're not real and valid.