How does age affect working memory capacity according to the 7 +/- 2 rule?

Working memory follows an inverted-U trajectory across the lifespan. Capacity increases through childhood, peaks in the **early to mid-20s**, and then gradually declines. A meta-analysis by Bopp and Verhaeghen (2005) found that adults over 65 show digit span reductions of approximately **1-2 items** compared to young adults. The decline is steepest for tasks requiring manipulation (backward span, n-back) rather than simple storage (forward span). However, cognitive training, physical exercise, and adequate sleep can slow this decline significantly. Notably, crystallized intelligence (vocabulary, general knowledge) continues to rise even as working memory declines, which is why older adults often compensate through greater expertise and more efficient chunking.

Can practicing memory span tests improve overall IQ scores?

Working memory training can improve performance on **working memory-specific subtests** by 0.5-1 standard deviations (Melby-Lervas & Hulme, 2013), which translates to a Working Memory Index increase of roughly **7-15 points**. However, transfer to overall Full-Scale IQ is more modest -- typically **2-5 points** -- because IQ encompasses verbal comprehension, perceptual reasoning, and processing speed in addition to working memory. The most robust improvements come from training programs lasting at least **4 weeks** with daily sessions of **15-25 minutes**. For maximum impact, combine working memory exercises with practice on other cognitive domains using resources like our [practice IQ test](/en/practice-iq-test).

What role does chunking play in overcoming working memory limitations?

Chunking is the **primary mechanism** by which humans exceed the raw 3-5 item capacity limit. By grouping items into meaningful units, you effectively compress information. A chess grandmaster looking at a mid-game board sees ~5-6 familiar configurations (chunks), not 25 individual pieces -- this is why experts can memorize positions that overwhelm novices (Chase & Simon, 1973). The key to effective chunking is **prior knowledge**: the more patterns you have stored in long-term memory, the more efficiently you can chunk new information. This creates a virtuous cycle where learning more makes your working memory *functionally* larger, even though its raw capacity remains constant.

Why do some people remember more than 9 items despite the 7 +/- 2 rule?

Exceptional memory performers use **deliberate strategies**, not larger raw capacity. Rajan Mahadevan, who memorized 31,811 digits of pi, used elaborate mnemonic systems to chunk digits into meaningful groups. K. Anders Ericsson studied a college student ("SF") who expanded his digit span from 7 to **79** over 200+ hours of practice -- not by increasing working memory capacity but by developing a hierarchical retrieval structure using running times (he was a competitive runner). Brain imaging studies confirm that memory athletes do not have structurally different brains; they have developed more efficient *encoding strategies* that allow them to use the same limited working memory more effectively.

How does cognitive load theory influence the design of educational materials?

Cognitive Load Theory (Sweller, 1988) has transformed instructional design by establishing three principles: (1) **reduce extraneous load** by removing unnecessary decorative images, redundant text, and split-attention designs; (2) **manage intrinsic load** by sequencing content from simple to complex and using worked examples before practice problems; (3) **promote germane load** by encouraging learners to self-explain, draw diagrams, and make connections to prior knowledge. The practical impact is significant: Mayer's (2009) *multimedia learning principles* -- derived directly from cognitive load theory -- have been shown to improve learning outcomes by **20-80%** compared to traditional instructional approaches.

Are working memory tests reliable indicators of general intelligence?

Working memory tests are among the **most reliable predictors** of general intelligence, with correlations of *r* = 0.50-0.90 depending on the complexity of the working memory task (Kyllonen & Christal, 1990; Oberauer et al., 2005). Complex span tasks (which require both storage and processing) predict IQ more strongly than simple span tasks (storage only). However, working memory is not the whole story -- it explains roughly **25-50%** of the variance in IQ, leaving substantial room for other factors like processing speed, crystallized knowledge, and perceptual reasoning. For the most accurate picture of your cognitive abilities, a comprehensive assessment like our [full IQ test](/en/full-iq-test) evaluates working memory alongside these other dimensions.

Working Memory Challenge: Understanding the 7 ± 2 Rule

The Mental Bottleneck That Shapes Your Intelligence

Imagine your brain as a high-powered computer. The hard drive (long-term memory) can store a lifetime of experiences. The processor (fluid intelligence) can handle sophisticated logical operations. But the RAM -- the workspace where active thinking happens -- is surprisingly small. That workspace is your working memory, and its capacity is one of the strongest predictors of IQ test performance.

In 1956, cognitive psychologist George A. Miller published what became one of the most cited papers in psychology's history: "The Magical Number Seven, Plus or Minus Two." His central finding: the average human can hold approximately 7 items (give or take 2) in working memory at any given moment. Not 20. Not 50. Seven.

"The 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." -- George A. Miller, 1956

This article will challenge you to test your own working memory, teach you evidence-based strategies to stretch that 7-item limit, and show you how working memory capacity connects to your IQ score.

What Working Memory Actually Is (and Is Not)

Working memory is often confused with short-term memory, but they are not identical:

Feature	Short-Term Memory	Working Memory
Function	Passive storage	Active storage and manipulation
Duration	15-30 seconds without rehearsal	Seconds to minutes (task-dependent)
Capacity	~7 items (Miller, 1956)	~4 chunks (Cowan, 2001)
Brain regions	Temporal and parietal lobes	Prefrontal cortex (primary)
Example	Remembering a phone number to dial it	Remembering a phone number while calculating its digit sum
IQ correlation	Moderate (r ~ 0.30)	Strong (r ~ 0.50-0.70)

The critical difference is manipulation. Short-term memory holds information passively; working memory holds it while you do something with it. When you solve 47 x 8 in your head, you need to hold partial products (320 + 56) while combining them. That is working memory at work.

"Working memory is the sketchpad of the mind." -- Alan Baddeley, architect of the modern working memory model, 1986

Baddeley's Working Memory Model

Alan Baddeley's influential model (1986, updated 2000) breaks working memory into four components:

Central Executive -- the attention controller that directs resources
Phonological Loop -- handles verbal and acoustic information (inner voice)
Visuospatial Sketchpad -- processes visual and spatial information (inner eye)
Episodic Buffer -- integrates information from the other components with long-term memory

IQ tests tap into all four components. Digit span tasks engage the phonological loop. Matrix reasoning tasks engage the visuospatial sketchpad. Complex reasoning tasks demand the central executive. Understanding which component a task uses helps you train more effectively.

Challenge 1: Test Your Digit Span Right Now

Here is a classic working memory challenge. Read each row of digits once, then look away and try to recall them in order. Do not re-read or use any aids.

Level	Digit Sequence	Digits to Remember
Level 1	7, 3, 9	3
Level 2	4, 8, 2, 6	4
Level 3	1, 5, 9, 3, 7	5
Level 4	6, 2, 8, 4, 1, 5	6
Level 5	3, 7, 1, 9, 5, 2, 8	7
Level 6	9, 4, 6, 1, 8, 3, 5, 2	8
Level 7	2, 7, 5, 3, 9, 1, 4, 8, 6	9
Level 8	8, 1, 4, 7, 3, 6, 9, 2, 5, 0	10

What Your Score Means

Digit Span	Percentile (approx.)	Interpretation
4 or fewer	Below 10th	Below average; may indicate attention or memory difficulties
5	10th-25th	Low average
6	25th-50th	Average
7	50th-75th	Average to high average
8	75th-90th	High average
9+	Above 90th	Superior working memory

Most adults max out at Level 5 or 6 on a first attempt. If you reached Level 7 or beyond, your working memory is above average. If you found Level 4 challenging, do not worry -- this capacity is highly trainable.

"Everyone complains of his memory, and no one complains of his judgment." -- Francois de La Rochefoucauld, 17th-century moralist

Challenge 2: Reverse Digit Span

This time, read each sequence once and then recall the digits in reverse order. This is harder because it requires both storage and manipulation -- the essence of working memory.

Level	Sequence (Read Forward)	Your Task (Recall Backward)
Level 1	5, 2	2, 5
Level 2	8, 3, 7	7, 3, 8
Level 3	1, 6, 4, 9	9, 4, 6, 1
Level 4	3, 7, 2, 5, 8	8, 5, 2, 7, 3
Level 5	6, 1, 9, 4, 7, 3	3, 7, 4, 9, 1, 6
Level 6	2, 8, 5, 1, 9, 4, 7	7, 4, 9, 1, 5, 8, 2

The average adult achieves a reverse digit span of 5-6 items. Your reverse span is typically 1-2 items shorter than your forward span, because the reversal operation consumes working memory resources that would otherwise be available for storage.

On the WAIS-IV, digit span forward and backward are combined into a single Digit Span subtest score that contributes to the Working Memory Index (WMI).

The 7 +/- 2 Rule: What Modern Research Says

Miller's 1956 paper has been enormously influential, but subsequent research has refined his estimate. Nelson Cowan's 2001 review, "The Magical Number 4," argued that the true capacity of working memory -- when chunking strategies are controlled for -- is closer to 3-5 items, not 7.

Researcher	Year	Estimated Capacity	Key Insight
George Miller	1956	7 +/- 2 items	Introduced the concept of "chunks"
Alan Baddeley	1986	~2 seconds of speech	Capacity limited by rehearsal time
Nelson Cowan	2001	3-5 items (pure capacity)	Miller's 7 inflated by chunking
Luck & Vogel	1997	~4 visual objects	Visual working memory has its own limit
Oberauer et al.	2018	3-5 items (confirmed)	Meta-analysis of 26 studies

The discrepancy between Miller's 7 and Cowan's 4 arises because Miller's participants were unconsciously chunking -- grouping items into larger meaningful units. When chunking is prevented (e.g., by using random, unrelated stimuli presented rapidly), capacity drops to about 4.

"The magical number four in short-term memory: A reconsideration of mental storage capacity." -- Nelson Cowan, title of his 2001 landmark paper

Chunking: The Strategy That Breaks the Bottleneck

Chunking is the single most powerful strategy for expanding effective working memory capacity. It works by compressing multiple items into a single meaningful unit, reducing the number of "slots" needed.

Chunking in Action

Consider this 12-digit sequence: 1-9-6-9-1-9-8-9-2-0-0-1

Without chunking: 12 items -- far beyond anyone's capacity.

With chunking: 1969 (Moon landing) + 1989 (Berlin Wall falls) + 2001 (September 11) = 3 chunks. Easily within capacity.

Types of Chunking Strategies

Strategy	How It Works	Example	Effective For
Temporal grouping	Group by rhythm/pauses	485-293-7610 (phone number format)	Numbers, sequences
Semantic grouping	Group by meaning	FBI, CIA, NSA = 3 chunks, not 9 letters	Acronyms, abbreviations
Pattern detection	Find mathematical relationships	2, 4, 8, 16 = "doubling" = 1 rule, not 4 numbers	Number series
Narrative linking	Create a story connecting items	"The CAT sat on the MAT near the HAT"	Word lists
Spatial grouping	Organize visually in mental space	Method of loci (memory palace)	Complex information

Real-World Chunking Examples

Chess grandmasters can memorize an entire board position (~25 pieces) in seconds because they see familiar configurations, not individual pieces. De Groot (1965) showed that this advantage disappears with random piece placement -- proving it is chunking of meaningful patterns, not raw memory capacity, that gives experts their edge.

Musicians read sheet music in phrases, not individual notes. A concert pianist does not process each of the ~1,500 notes in a Chopin etude separately; they recognize chord progressions, arpeggios, and melodic patterns as chunks.

Experienced radiologists can spot a tumor in a chest X-ray in under 200 milliseconds because they have chunked thousands of normal images into a "template" -- anything deviating from the template pops out immediately.

"Expertise is not about having a bigger working memory; it is about having better chunks." -- Herbert Simon, Nobel laureate in economics and pioneer of cognitive science

Challenge 3: The Dual-Task Challenge

This exercise tests the central executive component of working memory -- your ability to manage two tasks simultaneously.

Instructions: Read this list of words. While reading, count how many words contain the letter "A":

BRIDGE, CASTLE, HAMMER, PENCIL, BASKET, FOREST, GARDEN, PLANET, BOTTLE, ANCHOR

Now, without looking back:

How many words contained the letter "A"?
Can you recall the first three words in order?
Can you recall the last three words in order?

Answers: Six words contain "A" (CASTLE, HAMMER, BASKET, GARDEN, PLANET, ANCHOR). The first three are BRIDGE, CASTLE, HAMMER. The last three are BOTTLE, ANCHOR... wait -- that is only two if you excluded PLANET. This exercise illustrates divided attention cost: monitoring for a letter while encoding word order degrades performance on both tasks.

Research by Navon and Gopher (1979) established that dual-task performance follows a resource-sharing model -- when two tasks draw on the same working memory subsystem (in this case, both are verbal), interference is greatest.

How Working Memory Predicts IQ

The correlation between working memory capacity and general intelligence (g) is one of the strongest and most replicated findings in cognitive psychology.

Study	Sample Size	WM-IQ Correlation (r)	Key Finding
Kyllonen & Christal (1990)	2,000+	0.80-0.90	"Working memory capacity is (almost) nothing more than g"
Engle et al. (1999)	133	0.59	WM predicts fluid Gf after controlling for short-term memory
Ackerman et al. (2005)	Meta-analysis	0.50 (average)	Correlation varies by task complexity
Oberauer et al. (2005)	261	0.72	Complex span tasks show highest correlation

Why is the correlation so strong? Because most IQ test items require holding information in mind while performing operations on it. Consider a matrix reasoning problem: you must hold the patterns from Row 1, compare them to Row 2, extract the rule, and apply it to Row 3 -- all simultaneously. Every step demands working memory.

"The ability to hold in mind and manipulate multiple pieces of information is at the very heart of what we mean by intelligence." -- Randall Engle, Georgia Institute of Technology

The Working Memory Index on IQ Tests

On the WAIS-IV, the Working Memory Index (WMI) is composed of:

Subtest	Task Description	Skills Assessed
Digit Span	Repeat digits forward, backward, and in sequence	Phonological loop, central executive
Arithmetic	Solve math problems mentally (no pencil/paper)	Numerical WM, central executive
Letter-Number Sequencing (supplemental)	Hear mixed letters/numbers, recall numbers ascending then letters alphabetically	Reordering, central executive

A person with a WMI of 115 (84th percentile) can typically handle more complex reasoning tasks, learn new material faster, and perform better under time pressure than someone with a WMI of 85 (16th percentile).

Five Evidence-Based Exercises to Train Working Memory

Exercise 1: N-Back Training

The dual n-back task is the most studied working memory training paradigm. You see a sequence of positions on a grid while hearing a sequence of letters. You must press a button whenever the current stimulus matches the one presented n steps back.

1-back: Does the current letter match the previous letter?
2-back: Does the current letter match the one two steps ago?
3-back: Does the current letter match the one three steps ago?

Jaeggi et al. (2008) reported that 4 weeks of daily n-back training improved fluid intelligence on transfer tasks, though these findings have been debated (Melby-Lervas & Hulme, 2013; Redick et al., 2013). Regardless, n-back training reliably improves working memory capacity itself.

Exercise 2: Mental Arithmetic Chains

Perform multi-step calculations without pen or paper:

Beginner: 7 + 5 - 3 + 8 = ? (Answer: 17)
Intermediate: 12 x 3 + 7 - 15 / 5 = ? (Answer: 40, following left-to-right order without operator precedence; with standard precedence: 12 x 3 + 7 - 3 = 40)
Advanced: Start at 100. Subtract 7 repeatedly. How many steps to reach below 0? (Answer: 15 steps; the sequence goes 100, 93, 86, 79... reaching -5 at step 15)

The serial sevens task (subtracting 7 from 100) is actually used as a clinical screening tool for cognitive impairment in the Mini-Mental State Examination (MMSE).

Exercise 3: Sentence Span Task

Read each sentence, determine if it is true or false, and remember the last word of each sentence. After a set of sentences, recall all the final words in order.

"Dogs are a type of bird." (False) -- bird
"Paris is the capital of France." (True) -- France
"Water boils at 50 degrees Celsius." (False) -- Celsius

Now recall: bird, France, Celsius. Start with sets of 2-3 sentences and increase to 5-6 as your capacity grows. This task, developed by Daneman and Carpenter (1980), is one of the strongest predictors of reading comprehension ability.

Exercise 4: Spatial Working Memory Grid

Visualize a 4x4 grid. A sequence of cells is highlighted one by one. After the sequence, reproduce the order from memory.

Level 1: 3 cells
Level 2: 5 cells
Level 3: 7 cells
Level 4: 9 cells

This trains the visuospatial sketchpad -- the component most relevant to matrix reasoning and pattern recognition on IQ tests.

Exercise 5: Updating Task

You are given a running list of items, but you must only remember the last 4 items at any time. As each new item appears, drop the oldest:

Stream: Apple, Chair, Moon, River, Pencil, Clock, Tiger...

After "Tiger," you should be holding: River, Pencil, Clock, Tiger (having dropped Apple, Chair, and Moon).

This trains the updating function of the central executive -- the ability to revise the contents of working memory in real time.

Training Schedule

Week	Exercise Focus	Daily Duration	Expected Improvement
1-2	Digit span (forward + backward)	10-15 minutes	+0.5-1 digit on span
3-4	N-back (start at 2-back)	15-20 minutes	Advance to 3-back
5-6	Sentence span + mental arithmetic	15-20 minutes	+1 sentence in span
7-8	Mixed exercises + timed IQ practice	20-25 minutes	Consolidation

"The brain is like a muscle. When it is in use, we feel very good. Understanding is joyous." -- Carl Sagan

Managing Cognitive Load: The Practical Application

Understanding your working memory limits has immediate practical value. Cognitive Load Theory (Sweller, 1988) provides a framework:

The Three Types of Cognitive Load

Load Type	Definition	Example	How to Manage
Intrinsic	Difficulty inherent to the material	Calculus is harder than arithmetic	Break complex material into steps
Extraneous	Difficulty caused by poor presentation	Confusing diagrams, cluttered slides	Simplify and declutter
Germane	Effort devoted to understanding	Building mental models, making connections	Maximize this by reducing the other two

Practical Applications

Studying: Do not try to learn 50 vocabulary words in one session. Your working memory can only process 4-7 new items deeply at a time. Learn in batches of 5-7, with breaks between batches.
Test-taking: On IQ tests, if a question overwhelms your working memory, use external strategies -- write down intermediate results (if permitted), break the problem into sub-problems, or skip and return to it.
Work presentations: Slides with 7+ bullet points exceed most audience members' working memory capacity. The most effective presentations use 3-4 key points per slide.
Cooking: Following a complex recipe is a working memory task. Experienced cooks chunk steps into phases (prep, cook, plate), reducing cognitive load.

Conclusion: Embrace the Bottleneck, Then Stretch It

Working memory is the cognitive bottleneck through which all conscious thought must pass. The 7 +/- 2 rule (or more precisely, Cowan's 3-5 item pure capacity limit) defines the boundary, but chunking, training, and cognitive load management let you accomplish far more within those boundaries than raw capacity alone would suggest.

The exercises in this article are not just academic curiosities -- they train the same cognitive systems that IQ tests measure. Digit span, n-back, sentence span, and updating tasks all exercise the central executive, the phonological loop, and the visuospatial sketchpad that underpin your Working Memory Index score.

Start with the digit span challenges above. If you scored at Level 5, aim for Level 6 within two weeks. If you are already at Level 7, try the reverse span challenge and the dual-task exercise. Consistent daily practice of 15-20 minutes has been shown to produce measurable gains within 4-6 weeks.

For a comprehensive assessment of your working memory and other cognitive abilities, take our full IQ test. To practice under timed conditions, try our timed IQ test. And for ongoing cognitive training, our practice IQ test offers a variety of challenges that target working memory alongside other domains.

"The mind is not a vessel to be filled, but a fire to be kindled." -- Plutarch

References

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.
Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences, 24(1), 87-114.
Baddeley, A. D. (1986). Working Memory. Oxford University Press.
Baddeley, A. D. (2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 4(11), 417-423.
Jaeggi, S. M., et al. (2008). Improving fluid intelligence with training on working memory. Proceedings of the National Academy of Sciences, 105(19), 6829-6833.
Kyllonen, P. C., & Christal, R. E. (1990). Reasoning ability is (little more than) working-memory capacity?! Intelligence, 14(4), 389-433.
Engle, R. W., et al. (1999). Working memory, short-term memory, and general fluid intelligence: A latent-variable approach. Journal of Experimental Psychology: General, 128(3), 309-331.
Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257-285.
Daneman, M., & Carpenter, P. A. (1980). Individual differences in working memory and reading. Journal of Verbal Learning and Verbal Behavior, 19(4), 450-466.
De Groot, A. D. (1965). Thought and Choice in Chess. Mouton.
Luck, S. J., & Vogel, E. K. (1997). The capacity of visual working memory for features and conjunctions. Nature, 390, 279-281.

Working Memory Challenge: Understanding the 7 ± 2 Rule

The Mental Bottleneck That Shapes Your Intelligence