Skip to content

N-Back Task

Judge each item in an ongoing stimulus sequence against the item a fixed number of positions earlier.

Version
v1 · 2026-10-04 · History
Domain-specific #
13753
Domain group
Social Sciences
Origin domain
Psychology & Behavioral Sciences
Subdomains
Cognitive Psychology, Working Memory Research → Psychology & Behavioral Sciences
Aliases
N-back, N-back task, N-back paradigm

Core Idea

The n-back task presents an ordered stream of stimuli and asks for a repeated comparison: does the current item match the one n positions earlier in that same stream? A constructed visual 2-back could show letters and require a response when a letter repeats after one intervening item; Jaeggi and colleagues' visual condition instead used abstract shapes. The positive lag n sets which past position is relevant, while the stream keeps moving. The identity is the sliding, fixed-lag comparison rule, not a particular kind of stimulus, response button, brain scanner, or performance statistic.[1]

Researchers use n-back conditions to manipulate a task demand associated with working-memory maintenance and updating. This does not make performance a pure measure of working memory capacity. Perception, attention, response bias, strategies, timing and modality can also affect results. A 0-back condition often asks for a prespecified target and serves as a comparison block, but its fixed-target rule lacks the defining positive-lag comparison.[1][2][3]

Structural Signature

Sig role-phrases:

  • Sequential stimulus stream — Items appear in a known order, one after another, so “n positions back” has a determinate referent.[1]
  • Fixed positive lag n — The current item is compared with the item exactly n earlier presentations, not merely with any previously seen item.
  • Rolling judgment — Each eligible new position creates another match/nonmatch opportunity; this is not one final delayed-recognition question.[1]
  • Task response — A participant or system signals the defined match, permitting hits, misses, false alarms and timing to be assessed.[2]
  • Interpretive qualification — The observed result measures performance on this task under its chosen conditions; general claims about memory capacity or training transfer require additional evidence.[3]

What It Is Not

  • Not a 0-back target-detection condition. In that common control, a stimulus is compared with a fixed announced target rather than a moving item in the stream.[1]
  • Not ordinary serial recall. The participant need not reproduce the whole sequence; the task calls for local fixed-lag match judgments.
  • Not Working Memory itself. The live node names a cognitive function; n-back is one experimental paradigm often used to study it.
  • Not proof of general intelligence improvement. Practicing a task and transferring to broad cognitive abilities are separate empirical claims outside the defining rule.
  • Not a literal buffer algorithm. A person may use chunking or other strategies; the task specifies what comparison must be made, not a unique internal implementation.

Scope of Application

Positive-lag n-back tasks can use letters, shapes, spatial locations, sounds or simultaneous streams. Jaeggi and colleagues' original research contrasted 0-, 1-, 2-, and 3-back conditions and used auditory and visual material, including dual-task combinations.[1] The structural rule remains the same when the item type changes: compare the current item with the item a specified number of positions behind in the relevant stream. In a dual variant, separate channels may each have a lagged match decision, adding divided-attention demands.

The paradigm appears in cognitive psychology and neuroscience because it supplies a repeatable sequence of trialwise judgments at controlled lags. Its portability across stimuli does not mean the measured cognitive construct is identical across modalities, timings, or participant groups. An empirical validity study specifically tested whether n-back scores converge with another working-memory measure; such a question would be unnecessary if the construct mapping were automatic.[3]

Clarity

State the stimulus modality, lag, match rule, timing, response rule, target frequency, and scoring method. “2-back accuracy” without these conditions may hide important differences between experiments. A match opportunity begins only once enough earlier items have appeared. A match response to a nonmatch is a false alarm; failure to respond to a match is a miss in a target-response version. Accuracy, reaction time, and discrimination indices answer different questions and should not be silently substituted for one another.[2]

The term load is also operational. Raising n usually changes the distance of the relevant comparator, but may change difficulty through several processes besides storage. Thus the structural parameter n is controlled; the cognitive explanation of a performance change remains an empirical inference.[1][3]

Manages Complexity

The task turns an open-ended question—how a person handles recent information—into a repeated, parameterized decision. A fixed lag allows comparable blocks and separates positive-lag judgments from a simple target-detection control. This helps organize stimulus presentation and scoring. It does not eliminate interpretation: a headline percent correct can combine low hit rate with low false-alarm rate, or a speed–accuracy tradeoff with response bias. Meule's methodological analysis argues for reporting omissions and commissions separately and considering reaction time and signal-detection-style indices where appropriate.[2]

Abstract Reasoning

As an author-constructed illustration of Jaeggi and colleagues' lag rule, consider the letter stream A, B, A, C, A. In a 2-back task, the third item matches the first; the fourth does not match the second; the fifth matches the third. The relevant comparator shifts with every new item. In a 0-back block with fixed target A, all three A presentations would be targets, a different decision rule. The distinction is structural even if both tasks use the same letters and response button; the original study presented consonants auditorily and abstract shapes visually.[1]

Suppose two participants each answer 80% of trials correctly, but one misses many true matches and rarely responds to nonmatches while another responds often and incurs many false alarms. Their single accuracy percentages conceal different response profiles. The task's definition stays fixed, but the inference about working memory, attention or decision bias changes.[2]

Knowledge Transfer

To recognize n-back across modalities, find the ordered stream, positive integer lag, repeated current-versus-lagged comparison, and response to matches. A sound sequence and a spatial-location sequence can instantiate the same task structure. A dual-channel version preserves it separately within each channel while adding coordination demands. Transfer fails if the “earlier” item is a fixed memorized target, if the sequence is unordered, or if the output is merely a final list-recall score.[1]

Examples

Constructed visual letter 2-back

In this author-constructed case, a participant sees A, B, A, C, A and indicates when the current letter equals the one two positions earlier: positions three and five are matches, while position four is not. It applies the 2-back comparison rule documented by Jaeggi and colleagues, but is not their visual stimulus condition, which used abstract shapes; their consonants were spoken.[1]

Mapped back: Stream → ordered letters; lag → two; rolling comparison → each current letter versus two positions earlier; response → match/nonmatch signal; interpretation boundary → task accuracy is not a complete estimate of working-memory capacity.

Dual visual-auditory n-back

Jaeggi and colleagues compared single-channel auditory or visual conditions with dual-task conditions involving both modalities. Each relevant sequence still asks for its own lagged match judgment, while the simultaneous streams add divided-attention demands.[1]

Mapped back: Stream → visual and auditory sequences; lag → fixed positive offset per selected condition; rolling comparison → within-channel current versus prior item; response → target detection by channel; interpretation boundary → dual-task cost need not be a pure memory-storage effect.

Structural Tensions

  • Parametric demand versus construct specificity. Raising n changes the comparison distance and often task difficulty, but performance can reflect strategy, attention, perception and response as well as memory. Diagnostic: Which other task demands change with lag, and what control condition separates them?[1][3]
  • Compact score versus informative response profile. One accuracy number is easy to compare; separate hit, miss, false-alarm and reaction-time measures can reveal different mechanisms. Diagnostic: Are omissions and commissions distinguished, and is speed interpreted separately from accuracy?[2]

Structural–Framed Character

N-back is structural but experimentally framed: serial order, a fixed positive lag, repeated comparison and response jointly define it. Evaluative weight enters test design and interpretation—performance is not automatically a judgment of overall cognition. Human practice controls stimulus modality, timing, target frequency and scoring rules; laboratory conventions stabilize versions but no single institution defines the task. Vocabulary travels literally across letters, sounds and spatial positions when the lagged comparison survives. Importing “n-back” to a software rolling comparison without participants and trial responses is analogy; recognizing the same comparator is only the underlying skeleton. Its character: a repeatable experimental comparison task with context-dependent cognitive interpretation.

Structural Core vs. Domain Accent

Skeletal relation. A moving stimulus stream repeatedly asks whether the present item equals the item at a specified earlier offset.

Domain-bound condition. In cognitive experiments, the carrier is a timed stimulus sequence and the result is a participant's trialwise behavior. A positive lag distinguishes the task from a 0-back fixed-target control. Remove serial order or the fixed-lag comparator and the identity disappears.

Prime bar. The sliding-comparison skeleton could motivate a future-prime question, but n-back names a cognitive task paradigm with experimental presentation and performance interpretation. General resemblance does not make every rolling comparison this task.

Working Memory and Working Memory Updating are relevant cognitive targets, but the task is not a strict kind of either mental function. Working Memory Training is an intervention family that may use n-back, not a parent of this experimental rule. No strict whole-identity parent is established; this task remains a provisional unparented root unless a defensible experimental-task genus is later identified.

Neighborhood in Abstraction Space

N-Back Task sits in a moderately populated region (54th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Named Cognitive & Behavioral Effects (32 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With

0-back commonly compares each item with a constant target. Delayed matching can present a sample and later probe it once rather than sliding over an ongoing stream. Working-memory capacity is a hypothesized cognitive property, not synonymous with one task score. Dual n-back is a multi-stream variant of the same lag rule, with additional coordination demands. These distinctions explain why task validity and training-transfer claims require independent evidence.[1][3]

References

[1] S. M. Jaeggi and colleagues, original 2003 NeuroImage study of sequential single- and dual-task n-back conditions, NeuroImage 19 (2003), 210–225. Original full PDF checked, especially task design pp. 212–213 and Figure 1: spoken consonants in the auditory condition and abstract shapes in the visual condition. Letter strings in this entry are constructed illustrations of its lag rule, not reported visual stimuli. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m

[2] Adrian Meule, “Reporting and Interpreting Working Memory Performance in n-back Tasks”, Frontiers in Psychology 8 (2017), article 352. Original methodological article checked for distinct performance measures. registry ↩a ↩b ↩c ↩d ↩e ↩f

[3] K. M. Miller and colleagues, “Is the n-back task a valid neuropsychological measure for assessing working memory?”, Archives of Clinical Neuropsychology 24 (2009), 711–717. Original empirical abstract indexed by PubMed checked; full paper not checked here. registry ↩a ↩b ↩c ↩d ↩e ↩f