N-Back

The N-Back task is a continuous performance measure of working memory updating (Owen et al., 2005). Participants view a sequence of letters and must identify when the current letter matches the one shown N items ago. Performance is quantified using signal detection theory (d-prime and criterion), providing precise measures of working memory capacity that separate sensitivity from response bias.

ActivitySpec: lamp.nback

Cognitive domain: Working memory, working memory updating

Configuration

Setting	Description
Levels	N-back levels to run (supports 1, 2, 3)
Trials per Level	Number of trials per level (clamped 10-100)
Stimulus Duration	Letter display duration (ms)
ISI	Inter-stimulus interval (ms)
Target Proportion	Proportion of trials that are targets

API settings fields

Dashboard Setting	API Field	Type	Default
Levels	levels	number[]	[1, 2]
Trials per Level	trials_per_level	number	20
Stimulus Duration	stimulus_duration_ms	number	1500
ISI	isi_ms	number	2000
Target Proportion	target_proportion	number	0.3

Sample Instructions

"Letters will appear one at a time. Tap MATCH if the current letter is the same as the one shown 1 letter ago."

Usage

Consonants are displayed one at a time (vowels are excluded to prevent word-forming strategies). Each letter is shown for a configurable duration, followed by a blank inter-stimulus interval. The participant taps "Match" if the current letter is the same as the letter shown N items ago.

Trial outcomes are classified as: Hit (correct match), Miss (missed match), False Alarm (incorrect match), or Correct Rejection (correctly withheld response). d-prime is computed with log-linear correction (Hautus, 1995) to handle extreme proportions.

Multi-level support is built in: by default the task runs 1-back then 2-back sequentially, with a transition screen between levels.

Scoring

d-prime (sensitivity) is the primary metric -- higher values indicate better discrimination between targets and non-targets. Criterion measures response bias (negative = liberal, positive = conservative).

References

1. Owen, A. M. et al. (2005). N-back working memory paradigm: a meta-analysis of normative functional neuroimaging studies. Human Brain Mapping, 25(1), 46-59. DOI: 10.1002/hbm.20131
2. Hautus, M. J. (1995). Corrections for extreme proportions and their biasing effects on estimated values of d'. Behavior Research Methods, Instruments, & Computers, 27(1), 46-51.

Data

static_data

Field	Type	Description
n_level	number	N-back level
d_prime	number	Signal detection sensitivity
criterion	number	Response bias
hit_rate	number	Proportion of targets correctly identified
false_alarm_rate	number	Proportion of non-targets incorrectly identified
hits	number	Correct match responses
misses	number	Missed targets
false_alarms	number	Incorrect match responses
correct_rejections	number	Correctly withheld responses
hit_rt_mean	number	Mean RT on hits (ms)
hit_rt_median	number	Median RT on hits (ms)
hit_rt_sd	number	RT standard deviation on hits (ms)
score	number	Overall score
correct_answers	number	Total correct (hits + correct rejections)
wrong_answers	number	Total errors (misses + false alarms)
total_questions	number	Total trials
questionnaire	object	Post-game ratings: clarity (1-5), happiness (1-5)

temporal_slices

One entry per trial.

Field	Type	Description
item	number	Trial index
level	number	N-back level (1, 2, or 3)
type	boolean	Whether the response was correct
value	null	Unused
duration	number	Reaction time (ms)
letter	string	Displayed letter
is_target	boolean	Whether this was a target trial
responded	boolean	Whether the participant responded
outcome	string	"hit", "miss", "false_alarm", or "correct_rejection"

Cortex Features

No Cortex features are currently available for this activity.

View in Portal | Python SDK | API Reference