Recent experimental studies of biological memory demonstrate that principles of its formation, storage and retrieval are noticeably different from the earlier memory theories based on a computer metaphor. Contrary to computer memory its biological counterpart is:

• Non-representational – it is not a direct reflection of the physical world and its events;

• Re-constructive – its reproduction is an active process of self-assembly of functional neuronal system;

• Non-replicative – each new memory recall differs from the previous by recruiting overlapping but non-identical populations of neurons and synapses;

• Re-categorical – each new episode of memory reconstruction, being different in the neuronal composition is classified as belonging to the same instance by categorization on the repertoire of the existing functional systems in the brain.

• Re-consolidational – each new memory reconstruction is reconsolidated by activation of cellular mechanisms of neuronal plasticity that are similar but not identical to the cellular mechanisms for consolidation of a new memory.

Because of these properties biological memory in the neural nets has the qualities of prominent degeneracy (the equivalent subjective item is being encoded by multiple non-identical copies of a neuronal functional systems in the brain), superior auto associativity (different degenerate phenocopies of the same functional system possess different connections to other functional systems in the brain) and remarkable ability for regeneration of the functional system after damage to one of its copies or part of its elements).

Further experimental studies and theoretical modeling of these functions should shed a light on the unique properties of biological memory and open a possibility towards its simulation in architectures of neuromorphic intelligent devices.