Membrane Computing is a theory, which studies formal models of distributed computations with basic notions inspired and abstracted from biology. In a nutshell, membrane computing considers computational models having such basic components as a membrane (separator of an internal region from the external one), membrane structure (nested one or a network), multiset of abstract objects (presenting the content of a region) and evolution/communication rules over the multiset of objects. In other words, models from membrane computing (called also P systems) are formal computational devices, intuitively presented by a totality of membranes organized either in the hierarchical manner (a membrane can enclose other membranes) or as networks (membranes are connected to each other by means of communication channels and can form in this way any topology).
Each membrane has an associated collection of some objects, which can be presented in multiple copies, i.e., as multisets. Objects can be considered to have an atomic structure, hence, presented as letters, or to have complex structure, for instance, to be presented as strings. Each membrane possesses a collection of evolution/communication rules, that dictate how the content of membranes evolves and define the inter-membrane communications.
A number of Turing-universality results is known for various types of P systems, i.e., a P system can be viewed as an abstract universal (i.e., programmable) device. There were reported several applications of P systems in biology, linguistics, computer science, management, etc.
For more details on Membrane Computing we refer to the Scholarpedia article.