A measure for the capacity of short-term (primary) memory was first introduced even a bit earlier, in 1887, by Joseph Jacobs, who tested the span of digits his students were able to remember.2) Under the criterion that at least 50% of the digits need to be remembered correctly, subjects he tested mostly remembered about 7 digits. What he also noticed is that this result can be improved by grouping the digits (for example in groups of three), or reading aloud.
Similar results were reported by George Miller in 19563) . Based on his research, he suggested human short-term capacity was determined by the number of chunks or cognitive wholes one can remember, no matter if it is a letter, digit or word. This number on average equals seven, and may vary usually between five and nine.
This research increased interest in human memory research. Until then, it was commonly accepted that forgetting occurs due to interference of target information with new acquired information. This idea known as retroactive interference 4) was introduced in 1900 by Georg Elias Müller and Alfons Pilzecker 5) who concluded that learning does not cause instantaneous and long lasting memories, but that memory takes time to consolidate (Konsolidierung). This also means that during the consolidation period the memory is vulnerable.
During the 1950s John Brown 6) and Lloyd and Margaret Peterson 7) conducted research on information forgetting, offering other explanations of forgetting. Their hypothesis was that forgetting occurs due to the rapid decay of the memory trace (the neural substrate of retention) when it is first established. Both reasons will later be accepted as correct.
Another issue discussed during 1960s was weather human memory system has one or two components. Some authors like Arthur Melton 8) argued that both short term memory (STM) and long term memory (LTM) are just two subcomponents dependent on the same system. He justified his views by proofs of activation of the LTM in STM experiments. His work was very influential, yet during the years more and more evidence of at least two separate memory systems have accumulated.
The first more influential two component memory model was introduced in 1968 by Richard Atkinson and Richard Shiffrni 9) . Their model called Multi-store model consisted of long-term and working or short-term memory model and was later improved by an additional component, the sensory memory. Sensory memory contains one register for each sense and serves as an short lasting buffer-zone before the information can enter short-term memory. Short-term memory is a temporal storage for new information before it enters long-term memory, but is also used for cognitive tasks, understanding and learning.
Another term should be clarified here: the working memory, which is often mistaken for the short-term memory. The main difference between these two is that working memory usually includes the structure and processes performed by a system in control of the short-term memory.