TLDR: Reasoning is the largest game family on PlayMemorize · thirteen games, covering five sub-skills · pattern detection, deduction, abstraction, verbal logic, and override. They are the ones that show up most often on classic IQ tests, and they are also the ones with the most direct transfer to actual decision-making.
“Reasoning” is the umbrella every cognitive battery uses for the things that look like puzzles · find the rule, derive the answer, override the obvious-but-wrong move. It is not one skill but a stack of related ones, each with its own training format. PlayMemorize ships thirteen reasoning games, more than any other family on the site, because the sub-skills are largely independent · being a good Mastermind player tells you almost nothing about whether you will ace a verbal analogy.
What you will find here. Every reasoning game on the site, what specific sub-skill it trains, an inline round of each, and a session plan that hits all five sub-skills in twenty minutes.
What “reasoning” really means in this catalogue
For training purposes the catalogue splits reasoning into five sub-families:
Pattern reasoning. Find the rule that generates a sequence. Matrix, Sequences, Odd One Out, Spatial, and Mastermind Deduce are pattern-heavy.
Deductive reasoning. Use given constraints to eliminate options until one remains. Sudoku Force, Flag a Mine, Mastermind Deduce, and Chess Mate live here.
Abstraction. Strip away surface features to the underlying relation. Analogies and Odd One Out require this.
Verbal logic. Reasoning expressed in language · riddles, definitions, true/false claims. Riddles, Facts, and (loosely) Analogies cover this.
Cognitive override. Reasoning’s most underrated muscle · ignoring the answer your gut hands you and computing the right one. Stroop and Illusions train this directly.
Most “reasoning practice” sites focus on pattern alone, because the formats are easy to generate. The real ceiling for adults is usually somewhere else · either in deduction (where humans give up too early on the constraint search) or in override (where the gut answer wins even when the person knows better). Spreading practice across all five is the cheap way to find out which sub-skill is your weakest, and the weakest one usually has the most upside.
All thirteen reasoning games at a glance
Game-by-game
🧩 Matrix · the canonical pattern test
Matrix shows you a 3×3 grid with the bottom-right cell missing. Spot the rule that governs the other eight and pick the shape that completes the pattern. This is the format of Raven’s Progressive Matrices · the most-used non-verbal IQ subtest in the world · and the rules cascade across shape, colour, count, position, and rotation.
🎯 Odd One Out · classification and abstraction
Odd One Out gives you four to six items and asks which one doesn’t belong. To answer you have to find a category that contains all but one · which usually means peeling back at least one layer of surface description to a hidden shared trait.
🔄 Spatial · pattern reasoning under rotation
Spatial is included here because mental rotation is a reasoning task as much as a spatial one · the candidate you reject is the candidate whose rotation rule doesn’t fit. People who solve Matrix problems quickly often score well on Spatial too · the rule-search habit is the same.
🔗 Analogies · the abstraction muscle
“Doctor is to hospital as teacher is to ___?” Analogies forces you to find the relationship in the first pair (workplace) and re-instantiate it in the second. It is the cleanest abstraction drill in language · stripping away surface meaning to the relation underneath. SAT-style analogy questions correlated so highly with general reasoning that the College Board considered the format too predictive and removed them in 2005.
🌈 Stroop · cognitive override
The Stroop task is the gold-standard test of cognitive override · the word says RED but the ink is blue, and you have to name the ink. Your reading brain wants to win; your control system has to suppress it. The skill measured here predicts how often you fall for headline numbers, ad copy, and any other situation where the gut answer is wrong-but-obvious.
🔢 Sequences · numerical pattern reasoning
Sequences is pattern reasoning expressed in numbers. The rules range from arithmetic and geometric progressions to primes, Fibonacci, and powers. The cognitive habit it builds · “what could explain all the data points I have?” · is the same one mathematicians use when conjecturing.
💡 Facts · verbal logic and lie-spotting
Facts gives you several claims and asks which one is actually true. Some claims are arithmetic, some are calendar, some are geographic. The skill is parsing each claim’s logical structure and checking it · a kind of verbal logic that rewards calm rather than speed.
👁️ Illusions · perceptual override
Illusions are the visual cousin of Stroop. The drawing tells you the lines are different lengths; the math says they are equal. To win you have to override the perceptual answer with a measurement-based one. Players notice the same metacognitive habit transferring out of the game · “the chart looks dramatic, but the y-axis starts at 90 percent.”
🤔 Riddles · verbal logic under wordplay
Riddles mix classical puzzles, lateral-thinking riddles, math riddles, and wordplay. Every type rewards the same trick: read each word literally rather than skim it. Wordplay riddles are the closest format on the site to legal-document parsing · the answer is hiding in a word that has two meanings.
♔ Chess Mate · deduction under board geometry
Chess Mate gives you a position with a forced mate-in-one. Even chess novices can train the deductive layer here · for each candidate move, ask “if I make this move, can the king escape?”. The answer space is small enough to enumerate, so the game becomes pure constraint elimination.
🔟 Sudoku Force · pure single-step deduction
Sudoku Force gives you one cell where exactly one digit fits, and asks you to find that digit. The reasoning is row + column + box constraint intersection · the cleanest single-step Sudoku drill on the site.
🚩 Flag a Mine · constraint-satisfaction deduction
Flag a Mine shows a partially revealed Minesweeper board where one cell is provably a mine. To find it you read each numbered cell as a constraint and combine them · the same kind of reasoning that makes a logician happy.
🎲 Mastermind Deduce · deduction across many constraints
Mastermind Deduce gives you a history of past guesses and the peg feedback for each. Crack the code. Unlike Sudoku Force, the reasoning here is multi-step · you have to chain constraints · which makes it the deepest pure-deduction puzzle on the site.
🖼️ Mosaic · inference under partial information
Mosaic reveals a hidden subject row by row from coloured emoji squares. The reasoning task is inferential · “what shape is consistent with the rows I have seen so far, and at what point have I seen enough to commit?” Each new row eliminates candidate subjects from the multiple-choice list, and an early-but-correct guess scores far more than a late one. The skill is calibrated commitment · knowing when your evidence is enough, exactly the failure mode behind every mis-diagnosed bug and prematurely-closed argument.
How to actually train reasoning
Reasoning is half technique, half slowing down. Most failed rounds on the games above are not impossible · they are rounds where the player’s first instinct was wrong and they didn’t pause to check. The single biggest improvement most adults can make to their reasoning score is a habit of two-second pause before clicking.
Three habits that consistently lift scores: first, when you fail a round, replay it in your head before moving on · ask which step of the reasoning broke. Second, alternate pure pattern (Matrix, Sequences) with deduction (Sudoku Force, Mastermind) · the brain learns rule-search better when the rule type keeps changing. Third, do Stroop or Illusions at the end of every session · they are the override drills, and they are the easiest to skip because they feel like they “shouldn’t” need practice.
Don’t grind reasoning when tired. Reasoning games are unique on this site in that fatigue actively hurts your score · not just slows it. A tired brain reaches for the gut answer instead of overriding it, which means you can hard-code the wrong instinct by playing late at night. Save Stroop and Mastermind for fresh-brain time.
A 20-minute reasoning workout
- 2 minutes Stroop · cognitive override warm-up
- 3 minutes Matrix · pattern reasoning
- 3 minutes Mastermind Deduce · deductive reasoning
- 2 minutes Analogies · abstraction
- 2 minutes Sequences · numerical pattern
- 2 minutes Riddles · verbal logic
- 2 minutes Sudoku Force · single-step deduction
- 2 minutes Flag a Mine · spatial deduction
- 2 minutes Illusions · perceptual override
One round of each game beats two rounds of half. The skill that does best with mixed practice is exactly reasoning, because the brain’s rule-search engine is what’s being trained, not any one rule.
Where this matters in real life
Reasoning skill predicts almost everything we colloquially call “smart” · job performance in cognitively complex roles, ability to learn new domains, ability to spot a bad argument before it commits you. The transfer from games to real life is not “you become smarter” but “you reach more quickly for the second-order check” · the habit of asking “is the answer my gut handed me actually right?”.
The everyday transfer test: the next time you read a startling statistic in the news, do one Stroop-style override before quoting it · ask whether you are reacting to the number or to the framing. Whether you can or can’t is a more honest signal than any score on this site.
Polymath
Cross-game streak roulette drawn from the whole PlayMemorize catalogue. Pure full-spectrum test · every round can be any game
Play now - it's freeNo account needed. Works on any device.
