How many conformations can a protein remember?

We show that a protein can be trained to recognize multiple conformations, analogous to an associative memory, and provide capacity calculations based on energy fluctuations and information theory. Unlike the linear capacity of a Hopfield network, the number of conformations which can be remembered by a protein sequence depends on the size of the amino acid alphabet as lnA, independent of protein length. This admits the possibility of certain proteins, such as prions, evolving to fold to independent stable conformations, as well as novel possibilities for protein and heteropolymer design.