Skip to main content
PlayMemorize
← Back to blog

How to Master Inference

TLDR: Master Inference by first eliminating impossible bags (those missing a colour you drew), then picking the survivor whose colour ratios most closely match your handful. Ratio matching is not a rule of thumb here · it is exactly the rule the game scores by. And the round always has one right answer, built to win by a clear margin.

What Is Inference, and Why It Matters

Inference is the reasoning game that teaches you to think like a detective, doctor, or scientist. Unlike deduction games that run from a rule to a guaranteed answer, Inference works backward: you see a handful of coloured tokens drawn from one hidden bag and must pick which of several candidate bags most likely produced it.

The game trains abductive reasoning - the everyday logic of incomplete evidence. In real life, you rarely have perfect information. A patient shows symptoms; a witness describes what they saw; a sample arrives at the lab. You must weigh which explanation best fits the facts, even though multiple explanations remain possible. Inference builds that skill in a compact, playable form.

Each round shows three or four bags, and every bag holds exactly six tokens in a mix of colours. A handful of four is drawn from one of them · which one is what you have to work out. You win when you pick the bag that best explains the draw, and the reveal then names that bag and says why.

Tip: Inference is not a memory game. You can see all the candidate bags throughout the round. Focus your brain on comparing, not memorizing. The screen holds all the information you need.

The Core Skill: Elimination and Matching

Success in Inference rests on two moves, always in this order.

First, eliminate the impossible. If you drew a red token, no bag that contains zero red tokens can be the source. Cross it out mentally. A single missing colour rules out the entire bag forever. This step is decisive and binary: either the bag has that colour or it does not.

Second, match and compare. Among the survivors, ask which mix most closely resembles what you drew. If you drew two blue tokens, one green and one yellow, you want the bag that is about half blue, a quarter green and a quarter yellow · of six tokens, that is three blue, and roughly one or two of each other colour. The bag whose ratios mirror your handful is not just a good heuristic here, it is the exact rule the game scores by.

This two-step process is simple in principle but demands care in execution. Be aware, though, that the first step fires less often than you would expect: in the opening rounds only about 46 percent of deals contain a bag you can eliminate outright. Roughly half the time you go straight to step two, so do not treat elimination as the main engine. A single missing colour is still a complete disqualification when it appears.

InferenceOpen game
Loading…

The Elimination Rule. Before considering any probabilities, rule out every bag missing a colour you drew. This shrinks your search space and makes the remaining choice far clearer. Do this first, always.

Reading the Bags and the Draw

Each bag is shown as a row of coloured tokens, six of them, with its letter beside it. Your draw sits above in the same colours, so you can compare directly. Nothing is hidden and nothing has to be remembered · every token you need is on the screen at once.

Pay close attention to the proportions. A bag of 3 blue, 2 green and 1 red is not the same as one of 1 blue, 2 green and 3 red, even though both hold six tokens. If your draw leans toward blue, the first is the better match. Because every bag holds the same six tokens, comparing counts and comparing ratios come to the same thing · you never have to correct for one bag being bigger than another.

Tip: Count the tokens in your handful before looking at the bags. Know exactly what you drew: how many of each colour, and in what ratio. Then compare that pattern to each candidate. Precision here cuts noise and speeds your decision.

As the rounds go by, a fourth bag joins, then a fourth colour, and the draw grows from four tokens to five. What does not happen is the bags growing more alike. The generator refuses to deal a round unless the best bag explains the draw at least 1.6 times better than the runner-up, and that guard applies at every level · measured across 400 deals at each, the closest margin is 1.69 times at the opening difficulty and 1.69 times at the hardest. A later round has more to read, not a finer call to make.

Partial Matches Require Care: Most surviving bags will match on some colours and differ on others. Do not pick the first bag that “looks right.” Compare every survivor and pick the one that minimises overall difference, not the one that matches a single strong colour. The winner is always there by a clear margin · if two bags feel equally good to you, you have misread one of them.

Common Mistakes and How to Avoid Them

Mistake 1: Skipping elimination. Many players jump straight to comparing all bags without first ruling out those missing a colour. This wastes mental effort and leaves you comparing impossible sources. Always eliminate first.

Mistake 2: Over-focusing on one colour. Your handful contains multiple colours. Do not fixate on matching just the most common one. Compare the full pattern. A bag might have lots of blue but too little green; another bag with less blue but perfect green might be the better overall match.

Mistake 3: Confusing “most likely” with “perfect.” Inference is not about finding the bag that matches perfectly - that almost never happens. It is about finding the bag that best explains the evidence given. Pick the closest fit, not a perfect fit.

Proportional Matching. Do not just count tokens; think in ratios. If your draw is roughly half blue and half green, a bag with a 50-50 blue-green split is better than one that is 70-30, even if the total counts differ. The proportion matters more than absolute numbers.

Tip: When two bags seem equally close, count again rather than guessing. Every round has exactly one right answer, and the game will not deal a round where two bags tie · a felt tie is always a miscount. Go colour by colour through your draw and check each candidate against it.

How Difficulty Shapes Strategy

There is no difficulty menu, and nothing for you to set. The game opens every run at the same level and steps up by one rung after every eighteen rounds you survive, to a ceiling it reaches around round ninety. Because the rung is tied to your streak, a lost run puts you back at the opening level · the harder deals are something you reach, not something you select.

So the shape of a run is fixed and worth knowing. Rounds 1 to 18 deal three bags of six tokens in three colours, with a draw of four. From round 19 a fourth bag joins. A fourth colour arrives later, and the draw grows to five tokens later still. In every one of them: count each colour in your draw, count each colour in every survivor, and ask whether this bag sits closer to the draw than the last one did.

More to read, not a finer call: Each rung adds a bag, a colour or a token to the draw, so a late round takes longer to read. The decision itself stays the same width, because the winning margin is guarded at every rung. The skill does not change; the amount of counting does.

InferenceOpen game
Loading…

Practice Routine for Steady Improvement

To master Inference, use a focused practice routine.

Rounds 1-5: Learn elimination. The opening rounds deal three bags, so name what you rule out before you choose. Say it aloud: “I drew red, and Bag A has no red, so it is out. That leaves B and C.” About half the time there will be nothing to rule out, and that is the round telling you to go straight to matching.

Rounds 6-15: Practice proportional matching. Now focus on comparing the remaining candidates. For each round, pick the two or three most likely bags and compare their mixes in detail. Ask yourself: which one’s ratios match mine best? Explain your choice before clicking.

Rounds 16-25: The fourth bag arrives. You cannot switch difficulty up · it steps up on its own at round 19, if you get that far. Repeat the routine with the extra candidate. Elimination and matching still work unchanged; there is simply one more bag to read each round.

Rounds 26+: Speed and intuition. Once the four-bag deals feel routine, play several rounds in succession without deliberate narration. Let your brain process the comparison faster. You are training intuition grounded in solid reasoning, and you need the speed · the later rungs are only reached by surviving the earlier ones in one unbroken run.

The Narration Habit. In early rounds, narrate your reasoning aloud or in writing. “I drew 2 blue, 1 green, 1 red. Bag A has no green, so out. Bag B is 3 blue, 2 green, 1 red · that is close to my half-blue draw. Bag C leans too red. I pick Bag B.” This habit builds discipline and reveals where your thinking breaks down.

Consistency Over Speed: Do not rush. A careful round where you reason well is better practice than five fast rounds where you guess. Inference trains the quality of your thinking, not your reaction time. Build the habit first, then let speed follow naturally.

Plateau Is Normal: You may find your streak stopping in the same place for several sessions. This is normal, and it is not you being held at a level · every run restarts at the opening rung, so a plateau is about how far you get, not where you sit. Plateaus mean your brain is consolidating the skill. Keep practising - improvement often comes in sudden jumps after a period of apparent flatness.

Why Inference Matters Beyond the Game

Inference teaches the logic of diagnosis, investigation, and discovery. Doctors use it when a patient’s symptoms could fit several diseases; detectives use it when evidence points toward multiple suspects; scientists use it when data supports competing hypotheses. None of these professionals have certainty. They weigh the evidence, eliminate the implausible, and choose the best explanation.

Inference trains that exact thinking in a safe, feedback-rich environment. Every round is immediate and clear. You pick a bag, you learn whether it was right, and you see why. Over time, your ability to reason from incomplete evidence strengthens - a skill that transfers directly to real judgment under uncertainty.

There is no comfortable setting to start from, so start with the two moves instead and practise them until they are automatic. The harder deals will arrive on their own once your streak carries you there. Inference rewards careful, patient thinking. Master it, and you master the everyday logic of experts.

Advertisement
Which jar did the marbles come from? · Brain Games Ep. 10

New game videos on YouTube We make a short video for each game. Subscribe to see the new ones.

OK