THE MAGICAL NUMBER SEVEN

Working as an advertising professional with a focus on digital media and consumer experience design is both a blessing and a curse. For every site that wins critical or popular acclaim there are easily dozens (if not hundreds) of painfully difficult choices that must be made to bring a project to fruition – epic battles waged between brand managers and content marketers, visual designers, programmers, researchers, and other stakeholders – all focused on (hopefully) crafting an engaging balance of vision, design, and usability leading to the achievement of measurable business goals. Unfortunately, nearly every decision used to achieve this balance is predicated upon information that is frequently less than perfect.

Every so often, however, we’re fortunate to encounter a novel concept that offers potentially revolutionary insight into how we might mitigate our imperfect decisions through the application of industry standards or research-based guidelines. Such guidelines frequently come snuggled in the authoritative endorsements of trusted friends and colleagues or industry evangelists. These guidelines usually feel right, comfortably complimenting our own experience as well as the gut feelings we often leverage throughout site design, which helps to explain the ease and frequency with which such guidelines often become dogma within professional circles.

An excellent example of such a guideline being elevated to near-mythic proportions is the 7±2 (seven plus or minus two) rule. Nearly 50 years old, and often recounted as “Seven plus or minus two is all you can remember,” it has been applied as authoritative, scientific fact in countless situations including roadside billboard placement (more than seven items on a billboard near a highway would obviously lead to accidents!), advertising copy-writing (nobody can digest a headline with more than seven words!) to, most recently, the development of websites and navigation elements (it’s been proven that more than seven categories will confuse our users).

Two crucial questions present themselves. First, does this generally-accepted interpretation of 7±2 reflect an accurate understanding of the original research? And second, does 7±2 actually present a valid, beneficial rule that can be successfully applied to contemporary digital media design? To find out, we need to sidestep the briar patch of questionable 7±2 interpretations and instead revisit the original source of what has for nearly a half-century served as an almost universal design standard across a wide range of industries.

 QUESTIONS + METHODS

First published in 1956, George A. Miller’s “The Magical Number Seven, Plus or Minus Two: Some Limits on our Capacity for Processing Information” sought to identify and assess the various factors affecting the human capacity for receiving, processing, and remembering information in the short term.

For Miller, the impetus for such an investigation was something of a self-fulfilling prophecy, due – at least in part – to a perceived unusual recurrence of a particular value range (7±2) throughout his own research as well as in related journals and a variety of other sources:

“Either there really is something unusual about the number or else I am suffering from delusions of persecution.”

In his attempts to establish the actual relevance of 7±2, Miller structured his efforts to further explore whether 7±2 did indeed hold special meaning for three specific facets of human information processing ability, including: ƒ

• Absolute Judgment, the ability to accurately identify items differing on one (unidimensional) or more (multi-dimensional) characteristics. ƒ
• Immediate Memory, the capacity to hold a series of absolutely judgments for later use ƒ
• Recoding, the ability of the mind to reduce complex or difficult chunks of data into more easily manageable forms

Miller relies on his own research as well as both the published and unpublished findings of various colleagues – a seemingly reasonable cross-section of scientific inquiry on which to base his theories – but in any event his investigation was certainly not conducted in a vacuum.

ABSOLUTE JUDGMENT

While investigating the limits of absolute judgment, Miller proffers that

“[T]he observer is considered to be a communications channel…The experimental problem is to increase the amount of input information and to measure the amount of [accurately] transmitted information.”

Miller believes that users will continue to make accurate judgments of various stimuli until they are overwhelmed by the information flow, and that the error frequency will increase as the observer reaches his “channel capacity.”

Miller cites four separate, published studies into the human ability to make absolute judgments on uni-dimensional variables, with each study focusing on individual stimuli including tones, loudness, taste intensity of a saltwater solution, and visual position. He also references a number of unpublished studies on channel capacities for other single-variable judgments including size, hue, and brightness. Throughout all these experiments, the observer’s channel capacity ranged from 4 to 10 accurate absolute judgments per stimulus variable, surprisingly consistent with the 7±2 range.

But the observed performance on uni-dimensional stimuli falls far short of what humans do on a daily basis, including our ability to “…identify accurately any one of several hundred faces, any one of several thousand words, any one of several thousand objects, etc.” To develop a more complete understanding of absolute judgments, Miller also investigated the human capacity to accurately judge stimuli that differ from one another in more than one aspect.

Indeed, Miller cites a variety of data to establish a range of human performance in judging multi-dimensional stimuli – from assessing two-dimensional positioning of dots on a square, taste intensities of a fluid solution including salty and sweet components, tones that differ in both loudness and pitch, or – in an extreme case – an auditory study that presented observers with stimuli that differed in six respects. In each case, the observer’s channel capacity for judging multi-dimensional stimuli generally exceeded their performance on uni-dimensional stimuli, ranging from 24 to effectively 150 uniquely identifiable absolute judgments.

IMMEDIATE MEMORY

Miller’s discussion of immediate memory is somewhat more complicated than the concept of absolute judgment, but for our purposes we can probably replace “immediate memory” with “memory span,” which is to say that assessing immediate memory is to determine the amount of information an observer can recall almost immediately after initial perception (numbers in a range, items on a list, collections of synonyms, and so on.) Memory span experiments were conducted using five types of variables, including numbers (binary and decimal), letters of the alphabet, a combination of letters and decimal digits, and a series of monosyllabic words, to determine whether immediate memory functioned in a fashion similar to absolute judgment, or if there was a discernable distinction in its operation.

The results of the tests, in which users were presented with various lists (numbers, letters, etc.) and prompted to recall the lists in the order given, indicated that while the channel capacity for absolute judgment is limited by the total amount of information to which the observer is exposed, the capacity of immediate memory is limited instead by the total number of items (not surprisingly, somewhere in the neighborhood of seven), regardless of the amount of information per item. More specifically, “…I can say that the number of bits of information is constant for absolute judgment and the number of chunks [collections of bits, or individual absolute judgments] of information is constant for immediate memory.” And while immediate memory is itself limited not unlike absolute judgment, the ‘chunks’ of information stored by immediate memory far exceed the informational “information weight” of absolute judgment. But how is this possible?

RECODING

Briefly, Miller identifies a process, ‘recoding,’ that functions essentially as a mental shorthand – or more accurately an internal interpreter – for allowing our memory to translate complicated or meaningless chunks of information into recoded values that are more easily recalled. Miller shares that a perfect example of a recoding process is when an observer recounts a past event “in their own words,” recoding a complicated multi-sensory event into a more basic verbalization, which is then committed to memory. Recoding is, therefore, something of a mental compression technique allowing an observer to pack more data into a smaller and more easily managed amount of mental storage space.

CONCLUSIONS

Through this review of Miller’s original work on 7±2, we’re able to develop a slightly better understanding of his theory, research, and conclusions than many 7±2 proponents or skeptics, who themselves promote or denigrate Miller solely through casual readings or fifth-hand retellings.

But where does this leave us? How do we more accurately evaluate 7±2 and determine how (or whether) any of Miller’s findings in “Magical Number Seven” should be applied to the design of web, mobile, and other digital properties?

We can begin by answering our earliest question – is the generally-accepted interpretation (that 7±2 means “seven plus or minus two is all you can remember”) an accurate shorthand of Miller’s findings? If we were to limit our discussion only to that portion of “Magical Number Seven” dealing with absolute judgments of uni-dimensional stimuli, then the answer would have to be yes.

Common sense and observed behavior seem to indicate otherwise, however, and even the balance of “Magical Number Seven” establishes that the limits of human short-term memory can be – as is evidenced in everyday situations of recognizing friends or managing sets of more complicated data – enhanced well in excess of 7±2. So as a standalone theorem, the commonly held interpretation of Miller’s 7±2 is misleading at best.

But what if we expand our frame of reference to the totality of “Magical Number Seven,” specifically to the direct conclusions drawn by Miller himself? Would a more complete understanding of 7±2, encompassing what are (essentially) the verbatim findings of the author, augment its relevance specifically to web development? Miller concludes that: ƒ

• [T]he span of absolute judgment and the span of immediate memory impose severe limitations on the amount of information that we are able to receive, process, and remember.
• By organizing the stimulus input simultaneously into several dimensions and successively into a sequence of chunks, we manage to break (or at least stretch) this informational bottleneck. ƒ
• [T]he process of recoding is a very important one in human psychology […and] the kind of linguistic recoding that people do seems to me to be the very lifeblood of the thought processes.

Does 7±2, then, in light of this more comprehensive perspective, enhance our understanding of how to craft top-flight web sites?

IMPLICATIONS FOR DIGITAL MEDIA DESIGN

The impact that an even more comprehensive understanding of 7±2 should hold on digital media design and usability is certainly more apparent than ever before, though I’d caution against reading more into the findings of only one of a multitude of papers relevant to our understanding of these topics. To fully explore the implications of 7±2 on web design, we must review the perceived implications, the potential problems in applying 7±2 in such a context, and the likely implications of 7±2 thanks to our refreshed understanding of Miller’s actual methods and conclusions.

PERCEIVED IMPLICATIONS

The perceived implications of 7±2 include a fairly substantive set of fallacious conclusions based upon misinterpretations of “Magical Number Seven” that have made their way into web design theory. Often with the best of intentions, 7±2 is often cited to “claim that a web page should have no more than 7±2 links on it,”10 or that menu bars, pull-down menus, and bulleted lists similarly feature no more than 7±2 items. Given that even Miller has frequently disavowed such applications, it is probably safe to discard these implications as invalid.

POTENTIAL PROBLEMS IN APPLYING 7±2

In fact, over the past few years a movement has sprung up to protest such ill-informed and reckless applications of 7±2 to web design. This movement has also rightly pointed out certain core problems with the general use of 7±2, including: ƒ

• That Miller’s “Magical Number Seven” is often either fundamentally misunderstood or selectively quoted, leaving out qualifiers that the author includes to limit the breadth of his findings ƒ
• That the author has frequently been critical of the manner in which 7±2 has been applied to information design, going so far as to state that

“…7 [is] a limit for the discrimination of unidimensional stimuli (pitches, loudness, brightness, etc.) and also a limit for immediate recall, neither of which has anything to do with a person’s capacity to comprehend printed text.”

• A persistent willingness to ignore the wealth of additional, possibly more relevant scholarly and scientific research conducted in the years since “Magical Number Seven” was published, specifically as it relates to human memory, interactive technologies, and web designƒ
• That Miller’s work appeared approximately 40+ years before the rise of the commercial Internet/Web, and could hardly envision the unique nature and demands of web navigation

LIKELY IMPLICATIONS

We now know how people have routinely misapplied 7±2 to many information design projects, including for the web and other digital platforms, and we’ve established some of the factors that facilitate the misapplication. Are there any helpful lessons, theoretical or applied, to be taken from “Magical Number Seven”?

At a very high level, there are three concepts to keep in mind when working to enhance how you approach visual and information development for the web relative to how such designs will affect user experience, including: ƒ

• Absolute Judgment – That humans are generally very limited in their ability to distinguish between similar items that differ only in one aspect, but their capacity to distinguish between similar items increases dramatically when the items differ in more than one respect ƒ
• Immediate Memory – That while there are only so many items that you can reasonably expect a person to recall via short term memory, the brain is amazingly resourceful and – when properly cued through proper segmentation, organization, and categorization of data or visual information – the brain can be made to deal successfully with larger amounts of information ƒ
• Recoding / Chunking – That in certain situations, the brain is capable of various memory shortcuts that allow humans to process and store large amounts of information and/or complicated amounts of data that might otherwise be expected to be beyond our capacity (based on what absolute judgment or immediate memory might predict)

Of course, every designer, project manager, or usability specialist will interpret these concepts in their own way, bringing their personal experience, bias, and worldview to bear. Perhaps the most important theoretical observation, therefore, is to keep in mind that while absolute judgment and immediate memory work to limit human memory capacity, each person’s unique mix of experience, knowledge, and ingenuity can – to varying degrees – counter these limitations and requires that we as information professionals work to properly prepare information elements for our users.

As far as applied web development is concerned, there are a few “potentially useful” implications, though it’s important to acknowledge that there remains nearly 50 years of additional research that have not been included in the scope of this review, and there could easily be two, seven, or 113 theories that are better fits for understanding how to more effectively design for the web. In truth, some of these “potentially useful” implications may be stretching Miller’s basic conclusions somewhat. But this initial review of “Magical Number Seven,” when coupled with a bit of common sense and professional know-how, allows us to extend the theoretical implications into useful rules-of-thumb, including that:

• Designers should provide multi-dimensional cues to users within web sites to support and enhance navigation and content comprehension, including cues based on color, texture, font face, font size, absolute and relative position, highlighting, rollovers, and other visual feedback. This will likely decrease the effort required to acclimate users to the specific design of an individual web site and should promote a more efficient, pleasurable, and successful experience for user and client alike. ƒ
• When developing a site, be attentive to the placement of critical content, navigation, and functional components. Depending on their familiarity with a site, users may “chunk” or visually parse various segments of web pages and discard objects that do not demonstrably enhance their ability to fulfill goals (branding, advertising, etc.) ƒ
• When selecting the specific taxonomy for site sections, navigation elements, and other site components, both immediate memory and recoding theory suggest that careful selection of words and grouping of links and tools may – MAY – increase a user’s ability and efficiency in fulfilling goals and tasks.