Dual N-Back Tutorial: How to Train with Locations and Letters

Dual n-back is a working-memory training task where you track two streams at the same time.

In this version, the two streams are:

1. Where the item appears
2. Which letter appears

Your job is to decide whether the current location or letter matches the one from n steps back.

The task is simple to explain, but demanding to do well. It requires attention control, continuous updating, interference resistance and accurate comparison under pressure.

Dual n-back became widely known after Jaeggi and colleagues used a version with squares appearing in eight screen locations and letters/sounds presented in sequence, reporting improvements on fluid-intelligence measures after training. Later research has been more cautious: n-back training often improves the trained task, but evidence for broad far transfer to intelligence is mixed and debated. (PNAS)

Watch the tutorial

Use this video first, then practise the examples below.

What is dual n-back?

In an n-back task, you compare the current item with the item that appeared n trials earlier.

In a single n-back task, you track one stream.

For example:

Letters only:
A → C → A

At 2-back, the third letter matches the first letter.

In a dual n-back task, you track two streams at once.

In this tutorial:

Stream 1: location
Stream 2: letter

So each trial gives you a combined item:

letter + location

Example:

A in top-left
C in centre
A in top-left

At 2-back, the third trial matches the first trial on both:

letter match: yes
location match: yes

The dual n-back task is commonly described as involving a sequence of letters and a sequence of spatial positions, with the user identifying when either stream matches the one presented earlier. (brainworkshop.sourceforge.net)

What does “n-back” mean?

The n tells you how far back to compare.

1-back

Compare with the immediately previous trial.

Trial 1: A top-left
Trial 2: A bottom-right

At 1-back:

letter match: yes
location match: no

2-back

Compare with the item two trials earlier.

Trial 1: A top-left
Trial 2: C bottom-right
Trial 3: A centre

At 2-back:

letter match: yes
location match: no

3-back

Compare with the item three trials earlier.

Trial 1: B top-left
Trial 2: F bottom-right
Trial 3: K centre
Trial 4: B top-left

At 3-back:

letter match: yes
location match: yes

As n increases, the task becomes harder because you must keep a longer moving sequence active and update it continuously.

How to respond

On each trial, ask two questions:

Did this location match n-back?
Did this letter match n-back?

There are four possible outcomes:

The exact buttons depend on the app interface, but the logic is always the same.

Example: 2-back practice

Use this sequence.

The first n trials are usually warm-up trials because there is not yet an item n-back to compare with.

Why dual n-back feels difficult

Dual n-back is not just memory span.

It asks you to:

• hold recent locations
• hold recent letters
• update both streams every trial
• compare the current item with the right past item
• ignore tempting recent-but-wrong matches
• avoid losing the sequence after an error

This is why many people find 2-back challenging at first. The task punishes both distraction and over-rehearsal.

A common mistake is to remember the most recent item instead of the correct n-back item. For example, at 2-back, a 1-back match is a lure, not the correct answer.

How to train properly

1. Start with accuracy, not ego

Do not rush to higher n-levels.

A good training level should feel demanding but still controllable. As a rough rule, you want a level where you make some mistakes but are not constantly guessing.

Too easy:
you are almost perfect and not stretched.

Too hard:
you are guessing and losing the sequence.

Good:
you are working, correcting errors, and staying engaged.

2. Use 1-back as a tutorial, not the goal

1-back teaches the rule.

2-back is where the task starts to require real continuous updating.

Move to 2-back when you understand the response logic and can track both letters and locations without confusion.

3. Do not rely on tricks

Avoid turning the letters into words or memorising fixed location patterns. Those strategies can improve local performance while reducing the value of the task as working-memory training.

The point is not to memorise a clever shortcut.

The point is to practise active updating under load.

4. Reset after errors

When you make a mistake, do not panic and do not abandon the whole block.

Simply restart your tracking from the current trial.

A good mental instruction is:

Drop the mistake.
Re-enter the stream.
Track the next comparison.

5. Keep sessions short enough to stay sharp

Dual n-back is attentionally demanding. Long, tired sessions can become low-quality repetition.

A practical session might be:

• 5-10 minutes for a short session
• 15-20 minutes for a fuller training block

The important signal is not one heroic session. It is repeated, high-quality engagement over time.

6. Track trends, not single scores

One bad block does not mean you are getting worse.

One good block does not mean your working memory has permanently improved.

Look for patterns across sessions:

• higher stable n-level
• fewer false alarms
• better recovery after errors
• less collapse under pressure
• better performance after a break

Does dual n-back improve IQ?

This is the controversial part.

The original 2008 study reported that training on dual n-back improved performance on tests of fluid intelligence, and the study helped make dual n-back famous. (PNAS)

However, later reviews and meta-analyses have been much more cautious. Some studies find near-transfer to related working-memory tasks, but broad transfer to intelligence or real-world cognition is not reliably established. Melby-Lervåg and colleagues concluded that there was no good evidence that working-memory training improves intelligence test scores or real-world cognitive skills, while later reviews continue to debate how much of the apparent benefit reflects true transfer versus task similarity, study design or measurement bias. (PMC)

So the safest conclusion is:

Dual n-back can train performance on a demanding updating task.
It may support related working-memory skills.
It should not be treated as a guaranteed IQ-increase method.

That is why IQ Mindware treats dual n-back as one useful layer inside a broader cognitive-training system, not as the whole intervention.

How IQ Mindware uses n-back differently

Standard dual n-back usually trains you to track locations and letters.

That is useful. It trains continuous updating under load:

Where was it?
What letter was it?
Did either one match n-back?

But IQ Mindware extends this into relational n-back.

Instead of only asking whether a location or letter repeated, the task asks whether a relation repeated.

For example:

• Was the pattern moving out or in?
• Was it rotating clockwise or anticlockwise?
• Was this the same change as before?
• Did this relation match the one n trials back?
• Did this relation-colour combination match n-back?
• Did this graph transition match a previous transition?

This changes the training target.

Standard dual n-back mostly trains item updating:

location + letter

IQ Mindware relational n-back trains active relation tracking:

variable → relation → binding → prediction → reasoning

That matters because fluid reasoning is not just remembering items. It depends heavily on holding relations in mind, integrating them, resisting tempting near-matches, and using them to infer what follows.

Research on fluid intelligence and relational integration supports this emphasis: relational integration tasks predict fluid reasoning, and fluid intelligence has been linked to cognitive focus and integration across distributed brain systems. (PMC) The hippocampus is also strongly implicated in relational binding and comparison, while hippocampal-prefrontal interactions are important for memory integration and flexible use of past structure. (Frontiers)

So the IQ Mindware approach is not simply:

Train harder n-back.

It is:

Train the active workspace for relations.

A relational n-back game may ask you to hold:

• outwardness
• clockwise rotation
• same-change
• relation + colour
• state → transition
• valid graph edge

and then compare that relation across delay.

This is closer to the kind of structure used in matrix reasoning, analogy, rule learning and problem solving:

• What changed?
• What stayed the same?
• Which relation is being held?
• Does the current case match the earlier relation?
• Is this a true match or a near-miss?

In IQ Mindware, n-back is therefore one layer in a broader adaptive-intelligence stack:

Attention Control
→ Relational Memory
→ Binding Memory
→ Path Prediction
→ Reasoning

The goal is not to chase a high n-back score.

The goal is to train whether useful relations can be held, updated, bound to context, recovered after interference, and later used in reasoning.

This is why IQ Mindware treats n-back as relational memory training, not just memory-span practice.

A careful claims boundary is important: relational n-back is designed to target cognitive operations that are more directly relevant to fluid reasoning, but it should not be described as a guaranteed IQ-increase method. The stronger claim is that it trains and tests relation handling under load - the kind of operation that broader reasoning depends on.

Common dual n-back mistakes

Mistake 1: Comparing with the wrong trial

At 2-back, compare with two trials ago, not the previous trial.

2-back means:

current trial ↔ two trials earlier

Mistake 2: Tracking only one stream

Some people become good at locations but ignore letters, or vice versa.

Dual n-back requires both.

Mistake 3: Guessing too often

Guessing can create noisy progress.

When unsure, respond only when you genuinely detect a match.

Mistake 4: Over-rehearsing letters

Repeating long letter strings can overload the task.

Keep the letter stream active, but do not turn the task into a verbal memorisation drill.

Mistake 5: Training when exhausted

If you are tired, distracted or frustrated, performance may collapse.

That does not necessarily reflect your true capacity. It may reflect state.

Best beginner training plan

Session 1

Learn the rules.

Practise 1-back.

Do not worry about score.

Sessions 2-3

Move into 2-back.

Focus on clean responses.

Notice whether errors are mostly letter errors, location errors or wrong-lag errors.

Sessions 4-7

Train at the level where accuracy is challenging but not chaotic.

Aim for stable tracking, not maximum n-level.

After week 1

Review trend:

• Are you recovering faster after errors?
• Are false alarms dropping?
• Is your stable n-level improving?
• Can you maintain performance across several sessions?

What is a good dual n-back score?

There is no universal “good” score because apps differ in timing, stimulus design, scoring rules, adaptive logic and response method.

As a practical interpretation:

But do not compare your score too literally across different apps.

A 3-back score in one app may not mean the same thing as 3-back in another app.

Dual n-back FAQ

What is dual n-back?

Dual n-back is a working-memory task where you track two streams at once, such as letters and locations, and respond when either stream matches the item from n trials earlier.

What is 2-back?

2-back means you compare the current trial with the item two trials earlier.

What should I press if both the letter and location match?

Press both match responses, or select the combined match response, depending on the app.

Should I train every day?

Daily practice can help consistency, but quality matters more than forcing long sessions. Short, focused sessions are usually better than tired repetition.

Does dual n-back increase IQ?

It is not safe to claim that dual n-back reliably increases IQ. The evidence is mixed: the original study was promising, but later reviews have been more cautious about far transfer. (PNAS)

Is dual n-back better than single n-back?

Dual n-back is usually more demanding because it requires tracking two streams at once. However, harder is not automatically better. The best task is the one that keeps you in a useful training band without constant guessing.

Why do I suddenly drop after doing well?

That is common. Dual n-back is sensitive to fatigue, distraction, frustration and loss of sequence. Treat sudden drops as state information, not proof that you have lost ability.

Final takeaway

Dual n-back trains the ability to hold and update information under pressure.

In the locations-and-letters version, you are always asking:

Did this location match n-back?
Did this letter match n-back?

Train for clean tracking, good recovery and steady improvement.

Do not chase a number.

Build the active workspace that supports better reasoning under changing demands.