Memory Match

About memory match (Concentration)

Memory Match is the digital descendant of Concentration, a paper-card parlour game first published by Milton Bradley in 1958 and later made famous by the long-running NBC television show of the same name (1958 to 1991). The board on this page uses 16 face-down cards arranged 4 by 4, containing 8 distinct animal emoji pairs (dog, cat, fox, bear, lion, panda, koala, frog). Each turn you flip two cards; matches stay face up, mismatches flip back. The goal is to clear all 8 pairs in the fewest moves and shortest time. Move count is the harder metric to game; clock time rewards speed but does not punish guesses.

The game loads visuospatial working memory because each newly revealed card adds a (symbol, location) tuple to your active rehearsal set. Once you have seen N distinct mismatched pairs, you are tracking 2N positions in working memory, which on this 16-card layout maxes at 16 active positions, well beyond the 4-item span Cowan reported in 2001. Beating the game in under 14 moves therefore requires either chunking by quadrant or relying on positional cues (corners and centres are easier to remember than middle-edge cells).

How scoring works

Board     = 16 cards (8 pairs) shuffled
Per turn  = flip card A, flip card B, then evaluate
Match     = both cards stay face up, +1 pair, no penalty
Mismatch  = both flip back after 800 ms delay
Score     = (moves, time_seconds)  lower is better on both axes
Win       = all 8 pairs found

• Moves count each two-card turn, not each single flip. A perfect game is theoretically 8 moves; realistic optimum is 12 to 14.
• Time starts on the first flip, not the New button. The 800 ms mismatch delay is included.
• Pairs counter shows X/8 in real time.
• No penalty for going slow; the only cost of speed is forgetting positions you saw earlier.

Worked example: a 14-move expert run

An optimal strategy is the "scan first, match second" approach. Here is a 14-move solve sketch on an 8-pair board:

1. Moves 1 to 4 (intentional scan): deliberately flip 8 unique cards across the board, accepting the mismatches, to map symbol locations. You now have full information on roughly half the board.
2. Moves 5 to 6 (close known pairs): two of the symbols you exposed will have appeared twice in your scan. Close those 2 pairs immediately.
3. Moves 7 to 10 (controlled probing): flip one new card, then check if it matches any symbol you remember; if so, close. Add 4 more pairs this way.
4. Moves 11 to 14 (cleanup): with 12 cards already revealed and remembered, the final 2 pairs close from memory.

Difficulty by board size

This page uses the classic 4x4 layout. Larger boards add cards in pairs; expected moves grow super-linearly because each new pair raises both the search space and the memory load.

Common pitfalls

• Flipping the same first card twice in a row. If the second flip mismatched, you already know the first card. Pick a new unknown for your second-turn opener instead.
• Ignoring positional clusters. Cards in corners and centres are easier to remember than middle-edge cells. Burn early scans on edges so corners can carry the memory load.
• Rushing through the 800 ms mismatch animation. That window is your encoding time. Looking away wastes it and forces a re-scan.
• Verbal labelling. Saying "dog top-left, cat middle" pushes encoding into the phonological loop, which has lower capacity than the visuospatial sketchpad for spatial layouts.
• Greedy matching. Closing a pair the moment you have full information costs 0 moves saved over closing it 2 turns later, but reveals less new board. Sometimes it is worth holding a known pair to extract another scan.
• Restarting after one bad turn. The score reflects the whole run, but the encoded board state from the early scan does not transfer. Finishing a slow run still trains the underlying memory.

Related tools on 3Tej

Frequently asked questions