One of the favorite
old saws of
bad science fiction (and some good stuff too,
Larry Niven has used the device) is the idea of an injection that will transfer memories between humans. Memory
RNA is usually the vector described, as its existence was postulated all the way back in the fifties by
J. McConnell in his experiments with planarian flatworms. RNA's ubiquity in cells along with the complexity of information it can contain are likely what made seem so attractive as a storage site for
memory engrams, but
extraordinary claims demand extraordinary evidence, and the evidence for Memory RNA is anything but.
Early experiments found that there were different concentrations of RNA in the brains of rats who had been conditioned to learn something versus a control group. Later, McConnell "found", using fairly suspect scientific methods, that planarian flatworms fed either the pulverized bodies or purified RNA of trained planarians exhibited the behavior sought in the trained group; see "the ability of planarian worms to run a maze more successfully after being fed the remains of a successful worm" for more about them. Later experiments by A. L. Jacobsen confirmed some of the same results with purified trained rat RNA injected into the abdominal cavities of untrained rats, after which sensitivity to a light stimulus was specifically increased.
Unfortunately for the researchers, successful replications of their experiments were sometimes impossible, and success rate was often found to vary with the methods used to extract the RNA in question. Worse still, one experiment by J. Lutthes et al. used radioactive tracers to follow RNA introduced to the digestive system, and found that virtually all of it was excreted and came nowhere near the subject's brain!
The same types of studies were done transferring fully formed proteins rather than RNA, to determine if differently-generated proteins might be the seat of memory engram storage. Again, the results were at best inconclusive -- even experiments inhibiting the formation of proteins entirely were unable to stop some degree of long-term memory retention.
An examination of the idea of a chemical engram reveals flaws in it, regardless of the failure of replicating the experiments. For one, RNA is known to have important functions in the metabolism of all cells, functions that continue throughout the cell's life. While nature does love to use specific components for wildly different purposes depending on context, a strand of RNA coding an engram of high-level information rather than low-level protein construction details would probably hopelessly confuse the ribosomes and other cellular hardware responsible for protein transcription. Also, there seems to be no evidence for any kind of "tape head" for engrammatic information -- that is, no cellular organelle that can not only read information stored in RNA or proteins, but read it back at the insanely fast speed that animals can recall a given memory.
This writeup probably wouldn't have been possible without the info at http://clawww.lmu.edu/faculty/lswenson/Learning511/L15NPSY.html