Chunking is the recoding of small units into larger familiar units, so several items use one slot in working memory. Adults hold about seven such chunks in immediate memory, and under strict test conditions the pure limit is closer to four. For study, group three to four related terms under one meaningful label. Learn the label, then recall the parts from it.
- A chunk is the largest meaningful unit you recognize, not a fixed number of letters or words.
- Miller found a span of about seven chunks in 1956, independent of bits per chunk.
- Cowan tightened the pure limit to about four chunks in 2001.
- Thalmann and colleagues showed in 2019 that a chunk cuts the load on working memory. It also helps recall of other items held at the same time.
- For psychology terms, make each chunk familiar, distinct, and small, then test yourself by retrieving parts from the label.
What chunking is
Chunking groups input and gives the group a name, then remembers the name. Miller described the process as grouping input events, applying a new name to the group, and remembering the new name rather than the raw events.
A phone number shows the idea. The string 9849523450 is ten separate digits and exceeds the span. Grouped as 98 495 234 50, it is four groups and fits. A date works the same way. The string 12101946 is eight digits, but grouped as 12 10 1946 it is one month, one day, and one year.
Knowledge decides what counts as a chunk. The twelve letters FBI CBS IBM IRS are twelve items for a reader who does not know the acronyms. For a reader who knows them, they are four familiar units. Cowan defines a chunk as a set with strong links inside and weak links to other sets in use. This example shows why the same string loads memory differently for two people.
Chunking differs from simple repetition. Repetition keeps the same units and repeats them. Chunking changes the units by recoding them into a form that long-term memory already holds. That is why a familiar label helps and a random group does not.
The seven-item limit and why it is now four
In 1956 Miller reviewed span tasks across digits, letters, and words in his 1956 memory span paper and found a consistent limit near seven. Binary digits carry one bit each and decimal digits carry about 3.3 bits each, yet adults recalled about seven of either. Words carry far more bits, yet span stayed near five to seven. Miller concluded that immediate memory is limited by number of chunks, not by bits.
With binary items, the span is about nine and, although it drops to about five with monosyllabic English words, the difference is far less than the hypothesis of constant information would require. The span of immediate memory seems to be almost independent of the number of bits per chunk, at least over the range that has been examined to date.
Miller also warned that no one had fixed what a chunk is. The telegraph learner hears each dit and dah as a separate chunk at first. With practice the sounds form letters, the letters form words, and the words form phrases. The span in chunks stays small while the bits per chunk grow. In one self-experiment Miller cites, a trained listener grouped binary digits by fives, named each group, and repeated forty digits without error. The trick was not a larger span but larger chunks.
Later work tightened the number. Cowan reviewed overload tasks, blocked rehearsal tasks, and discontinuity data in his 2001 four-chunk review. He argued that Miller seven was a rough estimate and a rhetorical device. Under conditions that block recoding into larger units, the pure limit averages about four chunks, usually three to five. This does not replace Miller for study planning. It refines it. Plan for about four strong chunks when the material is new, and expect more when the groups are already familiar.
A common exam trap confuses items with chunks. Seven does not mean seven letters or seven words. It means about seven familiar units, and about four under strict conditions. If a question asks how twelve letters can be recalled, the answer is to recode them into fewer familiar chunks.
How chunking frees working memory
Thalmann, Souza, and Oberauer tested the mechanism in four experiments in 2019. They asked three questions: does a chunk cut the load for other items, does size change the benefit, and does position matter. Their abstract states the result directly. Chunking is the recoding of smaller units into larger familiar units. The benefit appears for chunked items and for other items held with them.
The mechanism they propose has three moves. First, the parts enter working memory and may interfere with items already there. Next, memory finds a matching compact form in long-term memory and retrieves it. Then it drops the separate parts and keeps the compact form, which frees room for what comes next. This account is supported by their finding that other random lists were recalled better when presented with a chunk than with another random list. The full record is available through the 2019 chunking study record.
Two limits matter for study. The first is overlap. In their first experiment, chunks used unique elements, so each chunk could be replaced by its first element, and size did not change the benefit. In later experiments, chunks shared elements across groups. Learners had to encode every part before they could detect the chunk. Larger chunks helped less. The effect also persisted when rehearsal was blocked, so it is not only faster rehearsal. The lesson is direct. Keep chunks distinct and avoid reusing the same words across groups.
The second limit is position. Early chunks helped later items. Late chunks did not repair earlier items. Once the parts of a late chunk have interfered with what came before, replacing them with a compact form repairs little. Order familiar chunks first when you build a list.
A three-letter chunk still loads more than a single letter in their data, so capacity is not a fixed chunk count regardless of size. Treat four as a planning average, not a guarantee. Small, familiar, distinct groups behave like one unit. Large or overlapping groups do not.
How to chunk psychology terms in three steps
Use chunking for lists that share a theme, such as biases, defense mechanisms, or schedules of reinforcement. Do not use it for unrelated terms. The links inside each group must be stronger than the links between groups.
- 1.Group three to four related terms by meaning, not by alphabet. Keep groups distinct and avoid repeating a word across groups.
- 2.Name each group with a familiar label you already know. Learn the label first, so long-term memory can supply the compact form.
- 3.Retrieve the parts from the label without looking, then check. Drill the label to parts direction, because that is the direction the test requires.
A worked example uses four biases. Instead of learning confirmation bias, hindsight bias, self-serving bias, and availability heuristic as four separate definitions, form two chunks. Label one Errors about evidence and put confirmation bias and hindsight bias in it. Label one Shortcuts about self and memory and put self-serving bias and availability heuristic in it. Learn the two labels, then say the two members of each from memory. That is two chunks, not four items.
Then connect the routine to a full study plan. Collect the terms for one exam into a small set of such groups, and keep each group to three or four terms. A bookmark list in Psychology World can hold one chunk set, for example Exam 1 biases. Drilling it with flashcards or typing tests the label to parts retrieval directly. For the rest of the routine, see the guide to memorise psychology terms, which shows how to build lists and space their retrieval.
When chunking fails
Chunking fails when the group has no meaning for you. An acronym you do not know is not a chunk. A category you cannot explain is not a chunk. If you cannot state the link in one sentence, split the group or pick a more familiar label.
It also fails when groups overlap. If two groups share the same cue word, memory must hold every part to tell them apart, and the benefit shrinks. This matches the overlapping-element result in the 2019 experiments. Rename one group or move the shared term.
It fails when groups grow too large. Six or seven terms under one label behave like several units, not one. Split the group into two smaller groups with separate labels.
It fails when you only reread. Reading the label and nodding does not install the compact form. Close the list and write the members from the label. If you miss one, restudy that link only, then test again.
How chunking differs from related methods
Students often mix methods that solve different problems. Chunking sets the grouping. Other methods set the timing, the order, or the cue. Use them together, but do not substitute one for another.
| Method | What it changes | When to use it |
|---|---|---|
| Chunking | Grouping of items under a familiar label | When a list is too long to hold at once |
| Spaced repetition | Timing of retrieval across days | After grouping, to keep each chunk over weeks |
| Interleaving | Order of practice across confusable topics | After chunking, to practise telling similar chunks apart |
| Method of loci | Cue route for ordered recall | When order matters and labels alone slip |
For timing, use spaced repetition after you form the chunks, so each retrieval takes effort but succeeds. For order, use interleaving when two chunks are easy to confuse, such as two similar biases, and practise sorting examples. For routes, use the method of loci when you must keep a fixed sequence. For depth of processing, see the contrast in maintenance versus elaborative rehearsal. It explains why linking a term to meaning beats repeating its sound.
Frequently asked questions
What is chunking in working memory?
Chunking is recoding small units into a larger familiar unit, so several items take one slot in working memory. A familiar group such as FBI counts as one chunk. The same three letters count as three items for a reader who does not know the acronym.
What is a good example of chunking in memory?
A phone number grouped as 98 495 234 50 is a good example, because ten digits become four groups. The twelve letters FBI CBS IBM IRS recalled as four acronyms is another, because knowledge turns twelve items into four chunks.
Does chunking increase memory capacity?
Chunking does not raise the number of slots. It raises the amount each slot holds by replacing parts with a compact form from long-term memory. Under strict test conditions the pure limit stays near four chunks, but each chunk can carry more when it is familiar and distinct.
How many items should be in a chunk?
Plan for three to four related items under one familiar label. Smaller distinct groups behave like one unit, while large or overlapping groups force memory to hold the parts and the benefit shrinks.
How is chunking different from the method of loci?
Chunking changes grouping by giving several items one label. The method of loci changes cues by placing items along a familiar route. Use chunking when a list is too long, and add a route when order also matters.
Sources
- 1.George A. Miller (1956). The magical number seven, plus or minus two.
- 2.Nelson Cowan (2001). The magical number 4 in short-term memory.
- 3.Mirko Thalmann, Alessandra S. Souza, Klaus Oberauer (2019). How does chunking help working memory.
- 4.Thalmann, Souza, Oberauer (2019). Full record and text via ZORA.
- 5.Chunking background and everyday examples cross-checked.