Simon Is Quietly on Your Side: The Memory Trick Built Into Every Round

Simon is one of the few toys that can claim a launch party at Studio 54. Milton Bradley released it in New York in 1978: a plastic disc with four coloured pads that lit up in a sequence, which you then had to repeat. It was designed by Ralph Baer, the engineer behind the first home video game console, together with Howard Morrison, and it borrowed its idea from an Atari arcade machine called Touch Me. The name came from the playground game, in which you do only what Simon says.

The rules have not changed since. The machine plays one step. You copy it. It plays the same step plus one more. You copy that. The round you fail on is your score, and the only thing that grows is the length of the sequence.

Put that way, Simon sounds like a straight test of working memory, the small mental scratchpad that holds things for a few seconds while you use them. Psychologists have a test almost exactly like it. In 1972 Philip Corsi, a graduate student working with the neuropsychologist Brenda Milner in Montreal, built a board of nine wooden blocks. The examiner taps a sequence of blocks, and you tap it back. It is Simon without the colours, and it measures spatial span the way the digit test we wrote about at cerdikcerdik.com/blog/number-memory-digit-span measures verbal span.

The typical result is humbling. Most adults manage around five blocks in the Corsi test, a couple fewer than the seven or so digits they can repeat back. Simon has fewer places to go — four pads rather than nine blocks — but that does not make each step cheap to hold. By that measure, a player who has never practised should start dropping steps at around round five or six.

Plenty of people go well past that on their first evening, and the reason is the one thing Simon does that the Corsi test does not. In a lab, every sequence is new. In Simon, every round begins by replaying the round you have just completed. The first seven steps of round eight are the whole of round seven, which were the whole of round six. By the time you are working on step eight, you have watched and reproduced step one eight times.

That turns out to matter far more than it sounds. In 1961 the psychologist Donald Hebb ran a digit span experiment with a quiet twist. Participants heard a list of nine digits and repeated it back, over and over, with a fresh list each time — except that every third list was secretly the same one. Recall of the repeated list climbed steadily across the session, while recall of the fresh lists stayed flat. Many participants never noticed the repetition at all.

This is now called the Hebb repetition effect, and it is one of the cleanest demonstrations of a short-term memory quietly turning into a long-term one. You do not have to try to learn the sequence. Being asked to hold it, again and again, is enough.

Simon runs that experiment on you every round, and far more aggressively than Hebb did. His repeated list came back every third trial. Simon’s comes back every single time, and only its tail is new. So the working memory load in round ten is not really ten items. It is a well-rehearsed opening that you have started to know, plus a short, fragile end that you are still holding by effort. Your score measures how long you can keep turning the first into the second.

That also explains where runs fall apart. The newest step feels like the risk, because it is the one you have seen only once, but the mistake is usually a few steps before it — in the middle of the sequence, in the part that has had several rehearsals but not yet enough to be automatic. The opening is safe and the last step is fresh in mind. The middle is where the sequence is changing from one kind of memory into the other, and it is the weakest point.

It is worth knowing what the sequence can and cannot contain. Each new step in the version on this site is a fresh random pick from all four pads, so a colour can follow itself. Across a ten-step sequence the chance of at least one doubled colour is better than 92 percent, and doubles are a classic way to lose, because the pad goes dark for only 150 milliseconds between the two flashes and it is easy to register one long light instead of two short ones. When a colour seems to stay on for longer than usual, it has almost certainly lit twice.

The size of what you are remembering is also smaller than it looks, and larger than you might hope. Four pads means each step is one of four equally likely choices, which is two bits of information. Eight steps is sixteen bits, or 65,536 possible sequences. A decimal digit carries about 3.3 bits, so an eight-step Simon sequence holds about as much information as a five-digit number. Ten steps is over a million possibilities — roughly six digits. The point is not that the numbers match. It is that Simon feels easier than a phone number of the same information content, and the repetition is most of the reason.

The version here makes a few choices that differ from the 1978 original, and they change how you should play. It is silent. The original gave each pad its own note — an A major chord, chosen so that any sequence sounded like a bugle call — and many players remember the tune rather than the lights. Without sound, the pattern has to live as positions and timing, so it helps to watch the shape the sequence makes across the board: a diagonal, a back-and-forth on one side, a lap around all four corners.

It also does not hurry you. The original ended your turn if you hesitated too long. Here the game waits as long as you like between presses, so there is no prize for speed while you are repeating. Pause, replay the sequence in your head from the beginning, and then press. The only clock that matters is the playback.

The playback is what the difficulty levels change. Each step is lit for 700 milliseconds on Easy, 500 on Medium and 320 on Hard, with a 150 millisecond gap between steps. A ten-step sequence takes about eight and a half seconds to play on Easy, and under five on Hard. At the slow speed there is time to name each colour to yourself. At the fast speed there is not, and players who rely on naming find their scores drop sharply, because the technique rather than the memory has run out. Your best score is kept separately for each difficulty, and so is the leaderboard, because they are not measuring the same thing.

The practical version is short. Treat the opening as something you are learning, not something you are re-reading, and let it become automatic. Put your effort on the middle, where runs actually break. Expect doubles, and count flashes rather than trusting duration. Use the time you are given before your first press. And play in short sessions spread across days rather than one long one — the same spacing that helps any skill stick, which we made the case for at cerdikcerdik.com/blog/spacing-effect-daily-practice.

See where your ceiling is on each speed at cerdikcerdik.com/games/simon-says. If it is higher than you expected, now you know who to thank: not your working memory, but a game that keeps telling you the answer.