Flashback

What you see

A trial is one stream of 18 to 30 symbols. The question above the board states the look-back distance: "the one before it" for 1-back, "the one 2 symbols back" or "the one 3 symbols back" at higher levels. Press Start (or Space) and the stream begins after a short 3, 2, 1 countdown. Once started, it never waits.

• The symbol. Each symbol is shown for its exposure time (1.5 s at level 1, down to 0.7 s at level 9), then the screen goes blank for a gap of about a second before the next one.
• Your answer. Press Match (or the Left arrow) if the symbol matches the one n symbols back, No match (or the Right arrow) if it does not. You can answer while the symbol is up or during the blank that follows. One answer per symbol; it cannot be changed. If you do not answer before the next symbol appears, the symbol is recorded as unanswered.
• The warm-up. The first n symbols have nothing to compare with, so the answer buttons are disabled and the line under them reads memorise only. Just take them in.
• The stream strip. Under the symbol is a row of small cells, one per symbol in the stream. The current cell is ringed, and a dashed ring marks the cell n symbols back, reminding you which symbol you are comparing with (never what it was). Cells you have answered turn grey. After you check your answer, each answered cell turns green if you were right or red if you were wrong, and a cell you left unanswered gets a red dashed outline.

The symbols

Symbols are six shapes: circle, square, triangle, diamond, star and hexagon, drawn in one of three fills (outline, striped or solid). On most levels only the shape counts: every symbol in the stream uses the same fill, so fill cannot distract you. On levels 7 and 10 the fill varies and a match needs both shape and fill to be the same; a striped star does not match a solid star.

How the stream is built

• Matches are rationed. Each level sets an exact number of matches, which works out to roughly three in every ten scored symbols.
• Lures from level 4. A lure is a symbol that matches the symbol one step nearer or one step further back than n, but not n itself. In a 2-back stream, a symbol that repeats from three symbols ago is a lure. Lures never sit right next to a real match.
• No accidental matches. Every other symbol is chosen so it does not match anything within n + 1 steps, so the only repeats you see are real matches and deliberate lures.

The ten levels

Practice starts at level 3. A perfect stream (every match caught, no false alarms, nothing left unanswered) moves you up one level. A stream with some errors but better than guessing keeps you where you are. A stream no better than guessing moves you down.

A stream lasts under a minute at most levels: 30 symbols of two seconds each at level 10 is exactly one minute.

What it measures

The n-back task was first reported by Wayne Kirchner in 1958 and became one of the most used working memory tasks in brain imaging. A meta-analysis of 24 neuroimaging studies found that n-back variants reliably activate a network of frontal and parietal regions, including the dorsolateral and ventrolateral prefrontal cortex. In Baddeley's terms it leans on the central executive: the job is less about storing and more about updating, replacing the oldest item in your window every time a new one arrives, and comparing on every symbol.

In our taxonomy Flashback is the only game that feeds the memory updating facet. Its competency tags are 75% working memory and 25% processing speed, the second reflecting that the stream sets the pace.

Lures are what make the task demanding. Kane and colleagues added lures to 2-back and 3-back tasks and found they drew more false alarms than ordinary non-matches, because a symbol you have seen recently feels familiar even when it is in the wrong position. Gray, Chabris and Braver found that on high-interference 3-back trials of this kind, people with higher fluid intelligence were more accurate.

an n-back procedure with and one without lure trials are almost two different tasks in terms of what they measure

How it is scored here

Only symbols after the warm-up count. Every scored symbol falls into one of five outcomes:

The trial score is hit rate + correct-rejection rate − 1, a measure called Youden's J. It is the same as your hit rate minus your false-alarm rate when you answer every symbol, and it is a close cousin of the "hits minus false alarms" score used in n-back research. It runs from 0 (no better than guessing) to 1 (perfect), and is scaled to 10,000 points. A perfect stream scores 10,000; anything short of perfect is capped at 9,999.

Worked through on a level 4 stream (20 scored symbols: 6 matches, 14 non-matches):

• One miss and one false alarm: 5/6 + 13/14 − 1 = 0.762, so 7,619 points. Partial: your level holds.
• Cautious, no false alarms but half the matches missed: 3/6 + 14/14 − 1 = 5,000 points.
• Every match caught but three non-matches left unanswered: 6/6 + 11/14 − 1 = 7,857 points. Unanswered symbols cost you even when nothing was there.
• Match on everything: 6/6 + 0/14 − 1 = 0 points, and you drop a level.

From trials to your 0–100 score

Speed is not scored in Flashback beyond the stream's own pace: a fast correct answer earns the same as a slow correct answer. Each trial is weighted by its measured difficulty, which rises steeply with the look-back distance (each extra step counts for roughly two levels), plus extra weight when shape and fill must both match and when more of the stream is lures. On our 0–100 scale, 50 means you scored what the difficulty you work at predicts.

Your report adds signal-detection measures from all your streams:

Where you will meet it

The n-back is a laboratory and brain-imaging workhorse, and a dual version of it was the training task in the 2008 study that started the brain-training debate and in many studies since. In hiring, span-type tasks are much more common: we have not found a major game-based assessment vendor that publicly documents an n-back game. The skill it isolates still matters for tests you will sit. Updating is what you do when you keep a running total through a data interpretation set, track which conditions of a puzzle are still live, or follow a changing number through a case calculation.

Practise Flashback as the updating component of your memory preparation, and do not expect a vendor's memory game to look like it.

The method

1. Name each symbol in one short word

   Say the shape to yourself: "star", "hex", "tri". On shape-and-fill levels, add the fill as one syllable in front: "stripe-star", "solid-hex", "line-tri" for outline. Words are faster to rehearse than pictures.

2. Hold exactly n names as a rolling list

   For 2-back, keep a two-word list: the symbol before last, then the last symbol. For 3-back, three words. Say the list in order, oldest first.

3. Compare the new symbol with the oldest name

   When a symbol appears, compare it with the first word of your list only. Same: Match. Different: No match. Answer while the symbol is still up if you can; you have until the blank ends.

4. Drop the oldest, add the newest

   Remove the first word and put the new symbol's name at the end. Your list is now ready for the next symbol. Say it once during the blank: "hex, tri".

5. Answer every symbol

   Unanswered symbols count against you. If you lose the thread, answer No match while you rebuild the list from the next n symbols. No match is right about seven times in ten, so it is the least costly default.

The first column is worth a second look. The circle there repeats the symbol from one step back, not two, so it is a lure, not a match, even though it feels like one. Only the older name in the list counts.

Worked example 1: a 2-back stream with a lure

Here is the stream from the first illustration, as words. Answer each symbol from symbol 3 on.

circle, star, circle, square, hexagon, square, triangle, hexagon, diamond, hexagon

Show the solution

Symbol	Symbol	Two back	Answer
3	circle	circle	Match
4	square	star	No match
5	hexagon	circle	No match
6	square	square	Match
7	triangle	hexagon	No match
8	hexagon	square	No match (lure: hexagon was three back)
9	diamond	triangle	No match
10	hexagon	hexagon	Match

Three matches out of eight scored symbols. Symbol 8 is the trap: the hexagon from symbol 5 is still fresh, but it sits three steps back.

Worked example 2: 3-back with lures on both sides

Level 8 and above use 3-back. Answer from symbol 4 on.

star, square, triangle, star, circle, square, diamond, circle, hexagon, circle

Show the solution

Symbol	Symbol	Three back	Answer
4	star	star	Match
5	circle	square	No match
6	square	triangle	No match (lure: square was four back)
7	diamond	star	No match
8	circle	circle	Match
9	hexagon	square	No match
10	circle	diamond	No match (lure: circle was two back)

Symbol 10 is the harder lure. The circle from symbol 8 is the freshest thing in your memory, which is exactly why it tempts you. Your three-word list at symbol 10 should read "diamond, circle, hexagon": the oldest word is diamond, so No match.

Now try one live. Letters appear one at a time; count the 2-back matches as you go, then check.

Ordering exerciseSign in to try it

Check your count

Three matches: symbol 3 (B, as in symbol 1), symbol 6 (H, as in symbol 4) and symbol 10 (L, as in symbol 8). Symbol 8 (L) is a lure: L last appeared three symbols earlier, at symbol 5.

Tips and traps

Levelling up

• Levels 1 to 2 (1-back). Compare with the previous symbol. Use them to make the answer keys automatic.
• Level 3 (2-back). The real task begins. Build the two-word list habit here.
• Levels 4 to 6. Lures arrive and the symbol time shortens to 0.8 s. Answer during the blank if you need the time.
• Level 7. Shape and fill must both match. Lengthen your labels by one syllable, not more.
• Levels 8 to 10 (3-back). A three-word list, lures on both sides of n, and at level 10 a 30-symbol stream with fill. The list method is the difference between guessing and knowing.

1. Week 1
   • Two or three short blocks per session at your level
   • Say the rolling list aloud if you are alone; it builds the habit faster
   • After each stream, check your near-miss traps count
2. Week 2
   • Aim for one perfect stream per session at 2-back
   • Practise the shape-and-fill labels before you reach level 7
   • Try the 2-back drill below as a warm-up
3. Week 3
   • Work at 3-back: three-word list, oldest first
   • If you lose the thread, default to No match and rebuild
   • Alternate Flashback with GridRecall to train updating and holding under load

Practice drillSign in to try it

Check yourself

5 practice questions

1. You are on a 2-back stream. The last four symbols were star, circle, square, circle. Is the second circle a match?
2. On a level 4 stream with 6 matches and 14 non-matches, you press Match on every symbol. What is the trial score?
3. In 3-back, the current symbol matches the one two symbols back but not the one three back. What is it, and what should you answer?

and 2 more

Create a free account to answerSign in

Sources

1. Kirchner, W. K. (1958). Age differences in short-term retention of rapidly changing information. Journal of Experimental Psychology, 55(4), 352–358. doi:10.1037/h0043688
2. Owen, A. M., McMillan, K. M., Laird, A. R., and Bullmore, E. (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
3. Kane, M. J., Conway, A. R. A., Miura, T. K., and Colflesh, G. J. H. (2007). Working memory, attention control, and the n-back task: a question of construct validity. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33(3), 615–622. doi:10.1037/0278-7393.33.3.615
4. Gray, J. R., Chabris, C. F., and Braver, T. S. (2003). Neural mechanisms of general fluid intelligence. Nature Neuroscience, 6(3), 316–322. doi:10.1038/nn1014
5. Redick, T. S., and Lindsey, D. R. (2013). Complex span and n-back measures of working memory: a meta-analysis. Psychonomic Bulletin and Review, 20(6), 1102–1113. doi:10.3758/s13423-013-0453-9
6. Schmiedek, F., Lövdén, M., and Lindenberger, U. (2014). A task is a task is a task: putting complex span, n-back, and other working memory indicators in psychometric context. Frontiers in Psychology, 5, 1475. doi:10.3389/fpsyg.2014.01475
7. Jaeggi, S. M., Buschkuehl, M., Jonides, J., and Perrig, W. J. (2008). Improving fluid intelligence with training on working memory. PNAS, 105(19), 6829–6833. doi:10.1073/pnas.0801268105
8. Redick, T. S., Shipstead, Z., Harrison, T. L., et al. (2013). No evidence of intelligence improvement after working memory training: a randomized, placebo-controlled study. Journal of Experimental Psychology: General, 142(2), 359–379. doi:10.1037/a0029082
9. Baddeley, A. (2012). Working memory: theories, models, and controversies. Annual Review of Psychology, 63, 1–29. doi:10.1146/annurev-psych-120710-100422

Put it into practice: FlashbackSign in to try it