The Number Seven Is Not Magical, but Cognitive Capacity Limitations Are Real and Relevant (Part 1)

René Liesefeld
January 31st, 2012

There are several apparently axiomatic design principles that purport to be perfectly adjusted to the human cognitive system. Their prominent characteristics are that they are broadly applicable and easy to grasp for the psychological layperson. Unfortunately, however, they are usually false. One of these principles is the “magical number seven”. Very loosely based on an influential article by Miller (1956), this “magical” number provides designers with an easy guideline to estimate how many elements their products can maximally contain without overcharging the cognitive capabilities of their users. Generations of designers were forced to limit, for example, steps in a workflow, tabs, items in dropdown lists, links, choices, bulleted lists, radio buttons and checkboxes, to this apparently magical number (cf., e.g., Eisenberg, 2004). As every myth, there is also some truth to the “magical number seven”. I here give a brief overview of some aspects of research on cognitive capacity limitations from a basic experimental psychological perspective. Although the here discussed insights are not as magically applicable as some would like, the present overview might be of use for the interested UI designer.

Failure to read Miller

Those who have actually read Miller’s (1956) article know that he did surely not promote any magical number (but the idea of recoding in order to chunk information as briefly introduced below). Additionally, Miller (1989) later clarified that he had employed the seven only as a rhetorical device to tie together two otherwise apparently unrelated topics of his research (cf. Cowan, 2001). Possibly having foreseen the misinterpretation of his “magical” number, Miller (1956) repeatedly cautions not to take this rhetorical device seriously. Accordingly, the final sentence of his article reads: “Perhaps there is something deep and profound behind all these sevens, but I suspect that it is only a pernicious, Pythagorean coincidence.” As things turned out, Miller’s efforts to prevent misinterpretation of his paper were in vain; it would be interesting to know what might have happened if he had not omitted the quotation marks around “magical”.

Chunking and the “magical” number 4±1

Humans can only hold a very restricted amount of information active in mind, or more specifically in working memory1. A common application of this working memory function in daily life is when people silently repeat a telephone number in order to dial it immediately. One topic of Miller’s (1956) paper is how people manage to hold much more information in working memory than their restricted capacity would usually allow them to. He assumes that they combine small units of information into larger and more meaningful units. This process became known as chunking. For example, remembering France as one chunk for the color sequence blue, white, red costs much less storage capacity than actually remembering blue, white, red. This recoding strategy, of course, only works if appropriate knowledge is available. Stated otherwise, how well chunking works depends on what a person already knows. Those who do not know the banner of France will not benefit from the described chunking strategy. If you pay close attention you will recognize that you are chunking information all the time, for example, when instead of reading and remembering each letter, you read and remember words or even larger chunks at once. Asked to recall the exact words or letters, you can fairly well “unpack” this information based on your knowledge on the spelling of words and on grammatical structure. People are, of course, usually not aware of either the initial chunking or the later “unpacking”. As in many other highly trained skills2, “bypassing” consciousness strongly increases the efficiency of chunking. Chunking also happens when a faithful Roman Catholic is instructed to pray a rosary instead of the exact number of Glory Be to the Fathers, Hail Marys and Lord’s Prayers (imagine the priest would instead each time have to tell the believer the exact words he has to pray), or when a UI designer considers using Metro style which specifies a set of design principles.
Obviously, school, work and life in general are in large part training camps for information chunking, because each piece of accumulated knowledge can in principle assist chunking. As people are highly trained in this skill and usually perform it unconsciously, it would be surprising if they would not also employ chunking in cognitive experiments (and while using UIs). It follows that chunking was probably also employed in the studies which Miller (1956) has reviewed more than 55 years ago. Miller consequently markedly overestimated capacity limits. For deriving a more valid estimate of capacity limits, steps must be taken to assure that no chunking occurs or that the amount of chunking is brought under experimental control. These more advanced techniques have led to current estimates of the “magical” number at around 4±1 (Cowan, 2001). I hasten to add that this number, too, might not be the answer to everything. It is rather likely that even the more advanced techniques for capacity estimation still have some flaws. Evidence for capacity limitations close to four items is, however, accumulating. Furthermore, the existence of some limitation of cognitive capacity (whatever its exact extent or reason) is beyond dispute in experimental psychology.
The take-home messages so far are that (a) the broad application of the “magical number seven” is based on a misinterpretation of Miller (1956), (b) chunking is a ubiquitous strategy for remembering information, and (c) cognitive capacity is more probably limited to around four items. In Part 2 of this article, I will introduce some evidence for the relevance of capacity limitations beyond working memory tasks. Furthermore, I will discuss some implications of capacity limitations for UI design.

1Working memory is a hypothetical cognitive construct whose function is to hold and process small amounts of information. Please note that Miller uses the term immediate memory. Like short-term memory, immediate memory, roughly speaking, is the passive storage function of working memory. Working memory, however, is the theoretically more developed and in current scientific discourses more commonly employed construct.
2In UI design, e.g., the concept of muscle memory has become popular (see, e.g., Kiermasch, 2010).


  • Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences, 24(1), 87-185. doi: 10.1017/S0140525X01003922
  • Eisenberg, B. (2004, October 29). Debunking Miller’s magic 7. Retrieved from
  • Kiermasch, J. (2010, June 16). The comeback of the pie menu. Retrieved from
  • Miller, G.A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63(2), 81-97. doi: 10.1037/h0043158
  • Miller, G.A. (1989). George A. Miller. In G. Lindzey (Ed.), A history of psychology in autobiography (Vol. VIII, pp. 391-418). Stanford, CA: Stanford University Press.

Want to know more about our services, products or our UX process?
We are looking forward to hearing from you.

Senior UX Manager
+49 681 959 3110

Before sending your request, please confirm that we may contact you by clicking in the checkbox above.