Memories encoded in the parent's brain should not be able to transfer to the progeny. This assumption, which is compatible with the tenets of modern neuroscience and genetics, is challenged by new insights regarding inheritance of transgenerational epigenetic responses. Here we reflect on new discoveries regarding "molecular memories" in light of older and scandalous work on "Memory transfer" spearheaded by James V. McConnell and Georges Ungar. While the history of this field is filled with controversies, mechanisms for transmission of information across generations are being elucidated in different organisms. Most strikingly, it is now clear that in Caenorhabditis elegans nematodes, somatic responses can control gene activity in descendants via heritable small RNA molecules, and that this type of inheritance is tightly regulated by dedicated machinery. In this perspective we will focus mostly on studies conducted using C. elegans, and examine recent work on the connection between small RNAs in the nervous system and germline. We will discuss the evidence for the inheritance of brain-orchestrated behavior, and its possible significance.