categories: Learning methodology
The second article in a series of blogs on Training Myths.
“The Magical Number Seven (Plus or MinusTwo): Some Limits on Our Capacity for Processing Information.”
The training myth that we’re looking at today is likely to be the first one that you encounter when you start learning about how training design should accommodate the way in which human memory is thought to work. You’ve probably seen it presented as follows:
Seven (plus or minus two):
The maximum number of bits of information that can be held in short term memory at one time.
Because of this, you’ve probably been told at some point that you should split your training content up so that learners never encounter more than seven things at once. You may have been given the advice that when you’re writing lists of bullet points, you shouldn’t have more than seven of them. If you’re a web designer, you may have been told that you shouldn’t offer more than seven items in page menus. In more extreme cases, you may even have been told not to use sentences consisting of more than seven words!
In its simplest terms, the assertion is that we can only hold seven things in our heads at any one time – the “plus or minus two” is added because some people are better at memory tasks than others. What would it mean if our Short Term Memory (STM) really could only hold seven things at the same time? It would make complex tasks like shopping, computer programming, driving on the motorway, or playing live music in a band very difficult, if not impossible. Football players, who must continuously track the location of 22 other people on the pitch (don’t forget the referee!), would find games impossibly challenging.
And, as importantly: why has it become so widespread? A quick internet image search for “seven plus or minus two” reveals dozens of slides quoting something like the phrase above. Quite a few of them even reference the research paper that originated the figure, which was published in Psychology Review in 1956 by George Miller of Princeton University. Which is unfair, because what Miller presented in his paper was not a sweeping generalisation but something much more specific. Miller was interested in what psychologists term “absolute judgment.” He wanted to know how quickly our brains become overloaded when we’re asked to distinguish between alternative stimuli that vary in a single dimension of measurement.
In the experiments Miller was reporting, that single dimension was the frequency of a series of audio tones; in effect, he was asking people to arrange a series of musical notes from the lowest to the highest. People had no difficulty completing this task when four notes were involved, but usually started to struggle with eight, and they found arranging fourteen notes in the correct order impossible (unless they were musicians; Miller notes that “there is evidence that a musically sophisticated person with absolute pitch can identify accurately any one of 50 or 60 different pitches. Fortunately, I do not have time to discuss these remarkable exceptions.”)
It’s not wrong to say that STM is limited. One psychologist who has a lively public speaking career bets a member of the audience £100 that he won’t be able to recall a list of ten different objects by the end of his talk; he’s yet to lose that bet. But the emphasis above, that the findings focused on a single dimension of measurement, is very important. Miller is careful to make this point later on in his paper, but perhaps some of his audience had stopped reading by then.
The point is that most of the things that we hold in STM don’t consist of information in a single dimension. And each extra dimension that is added resets the count for our mental storage capacity; the actual maximum number of things we can hold in our heads at the same time is well into the hundreds.
Miller also uses the term “recoding” to describe how we can organise the input sequence into units or chunks. In an experiment by Sidney Smith, this involved taking a sting of binary digits:
1 0 1 0 0 0 1 0 0 1 1 1 0 0 1 1 1 0
And recoding the sequence as chunks of four digits:
1 0 1 0 0 0 1 0 0 1 1 1 0 0 1 1 10
Or even five digits:
1 0 1 0 0 0 1 0 0 1 1 1 0 0 1 1 1 0
It’s reasonably obvious that the chunks are easier to remember, and this is exactly what Sidney Smith found. With practice, it’s possible to remember a string of forty binary digits in this way. Interestingly, individuals with better than average recall appear to achieve it by using larger chunks rather than more chunks. Understandably, you’ll see lots of training presentations that describe this practice as “chunking”, but in modern psychology the word is used for a different purpose: it describes how the brain organises a collection of memories in long-term memory that have strong associations with each other, but weak associations with memories in other chunks. If you’re going to teach people the technique described above, it’s best to refer to it, as Miller did, as recoding.
The point of all this preamble is a simple one: how you organise the information that you want to present in your training material can make an enormous difference to whether it’s retained or not. But you’ll help your learners the most by grouping your content into collections that have clear and obvious associations with each other. That approach will provide more benefit than blindly sticking to the “seven, plus or minus two” rule.
Learning & Development Consultant, ProfitAbility
eLearning specialist, musician, movie nerd, cartoonist, photographer, skier, tech author, blogger, noticer of things