N-Back (digits)

A diamond with four slots fills the upper screen, and four buttons below are laid out the same way. Before each block, a 2 s cue shows “0-back” or “2-back” over the empty diamond. Then a digit appears every 2 s, for 500 ms each, in its own slot (1 top, 2 left, 3 right, 4 bottom). The stream does not wait for answers: the first press before the next digit counts.

Useful when working-memory load is the variable of interest and a published design is wanted: the default reproduces the 0-back / 2-back task used in alcohol-dependence research (Charlet et al., 2014), which goes back to the fMRI task of Callicott et al. (1999). The 0-back blocks give a perceptual and motor baseline for the 2-back blocks.

Less suitable when onboarding time must be near zero, or when a working-memory-capacity estimate is the goal: n-back scores correlate only modestly with complex-span tasks. For Match / No match responses with signal-detection scores, use the letter N-back.

Keep the defaults to follow the published design: 8 blocks of 14 digits, alternating 0-back and 2-back (four 30 s blocks per condition, about 4:30 min in all; Charlet et al., 2014), each digit for 500 ms followed by a 1500 ms blank, and a 2 s instruction before each block. Practice runs one 14-digit block per condition with feedback; this length is our reading, since the paper reports practice trials without their length. A 0, 2, 0, 2 short version is available; report it as a lab choice.

• Block design: 8 blocks (published), 4 blocks (short version), 2 or 6 blocks
• Digits per block
• Digit duration and blank interval
• Practice block per condition, with feedback

• Hit rate (correct / scored digits), incorrect rate and miss rate per load, as reported by Charlet et al. (2014)
• No responses and wrong buttons per load
• Median and mean RT of correct responses per load
• 2-back − 0-back difference in accuracy and RT, and the performance decrement (2-back minus 0-back % correct; Jansma et al., 2000)
• Per digit: block, load, position, digit, correct answer, response, correctness, RT and how it was timed

Every digit has a correct answer, so chance accuracy is 25%; there are no targets and non-targets and no d′; the hit rate is the share of correct presses, as in Charlet et al. (2014). The first two digits of a 2-back block have no answer yet and are not scored. The 2-back − 0-back difference separates the working-memory demand from perception and motor response. RT runs from the drawn digit to the touch-down on the button; this timing has not yet been validated on real devices. The order of the digits is random; the sources state no further constraint.

• The digit n-back with the digits at the points of a diamond comes from functional imaging research on working-memory capacity (Callicott et al., 1999); every stimulus is both probe and target (Kempf et al., 2008).
• In alcohol-dependent patients, neural activity during the 2-back condition of this task predicted a lower relapse risk (Charlet et al., 2014).
• The same task, with 4 blocks per condition, 500 ms digits and a 1500 ms interval, is part of an RDoC-oriented fMRI battery (Daedelow et al., 2021).
• n-back performance correlates only weakly with complex-span performance, so the two are not interchangeable measures of working memory (Kane et al., 2007; Jaeggi et al., 2010).

• Charlet et al. (2014), working memory load and relapse risk in alcohol dependence
• Charlet et al. (2014), SOBP poster with the task figure
• Charlet (2015), dissertation describing the task (in German)
• Daedelow et al. (2021), the same task in an RDoC fMRI battery
• Callicott et al. (1999), origin of the paradigm
• Kempf et al. (2008), every stimulus both probe and target
• Owen et al. (2005), meta-analysis of n-back neuroimaging studies

• N-Back (letters)
• Rotation Span
• Symmetry Span