Working memory is generally considered to have limited capacity. The earliest quantification of the capacity limit associated with short-term memory was the magical number seven introduced by Miller (1956). He noticed that the memory span of young adults was around seven elements, called 'chunks,' regardless of whether the elements were digits, letters, words, or other units. Later research revealed that memory span does depend on the category of chunks used (e.g., span is around seven for digits, around six for letters, and around five for words), and even on features of the chunks within a category. For instance, span is lower for long than for short words. In general, memory span for verbal contents (digits, letters, words, etc.) strongly depends on the time it takes to speak the contents aloud, and on the lexical status of the contents (i.e., whether the contents are words known to the person or not). Several other factors also affect a person's measured span, and therefore it is difficult to pin down the capacity of short-term or working memory to a number of chunks. Nonetheless, Cowan (2001) has proposed that working memory has a capacity of about four chunks in young adults (and less in children and older adults).
Miller's paper points out that channel capacity on various tasks was around 2.5 bits of information. Measurements of human short term memory capacity also found a 7±2 limit. However, this limit was eventually found to be a result of using subjects who were speakers of English to remember sequences of single digits. It turns out that one component of human working memory, the phonological loop, is capable of holding around 2 seconds of sound. Two seconds is the duration of the English spoken form of 7±2 digits (in Chinese it is around 10 and in Welsh around 6), the variation is highly correlated with the rate at which people speak.
The concept of a limit is illustrated by imagining the patterns on the faces of a dice. It is easy for many people to visualize each of the six faces. Now imagine seven dots, eight dots, nine dots, ten dots, and so on. At some point it becomes impossible to visualize the dots as a single pattern (a process known as subitizing), and one thinks of, say, eight as two groups of four. The upper limit of one's visualization of a number represented as dots is the subitizing limit for that exercise.
The film Rain Man, starring Dustin Hoffman, portrayed an autistic savant, who was able to visualize the number represented by an entire box of toothpicks spilled on the floor. A similar feat was clinically observed by neuropsychologist Oliver Sacks and reported in his book The Man Who Mistook His Wife for a Hat. Therefore one might suppose that this limit is an arbitrary limit imposed by our cognition rather than necessarily being a physical limit.
A number of urban legends have grown up around the number 7±2 and human performance on various cognitive tasks. While Miller's paper is most often cited, by coincidence research into short term memory also threw up a 7±2 finding, which seems to have added impetus to the claims made.
The 7±2 urban legends are various rules specifying the maximum number of items that can occur in a given context (e.g., in software engineering the maximum number of subroutines that should be called from the main program). Whether these 7±2 rules provide the benefits claimed of them can be verified only by experiments. However, neither Miller's paper nor the early short term memory research is likely to provide the primary experimental evidence needed to back up such claims.