Analytical Puzzles & Critical Reasoning — UPSC CSAT

Weightage: ~25–30 questions of CSAT's 80 — the largest section, split roughly evenly between puzzle-solving and critical reasoning.

1. Seating arrangements — the diagram-first method

Worked example 1.1. Five friends — P, Q, R, S, T — sit in a row facing north. Q sits second from the left end. R sits immediately to the right of Q. P sits at one of the ends. S does not sit adjacent to T. Who sits at the two ends?

Solution. Draw 5 blank positions: _ _ _ _ _ (positions 1-5, left to right). Q is second from left → position 2. R is immediately right of Q → position 3. P sits at one end → position 1 or 5. Remaining people (S, T) fill the two leftover positions, with the constraint S is NOT adjacent to T. If P is at position 1: remaining positions 4, 5 go to S and T — but positions 4 and 5 ARE adjacent, violating the constraint. So P must be at position 5. Remaining positions 1, 4 go to S and T — NOT adjacent (positions 1 and 4 have positions 2,3 between them), satisfying the constraint. Final arrangement: position 1 = S or T, position 2 = Q, position 3 = R, position 4 = T or S, position 5 = P. The two ends are occupied by P (position 5) and either S or T (position 1) — the puzzle doesn't fully determine which of S/T is at position 1 versus 4 without further clues, but P at one end is confirmed.

2. Blood relations — the family-tree method

Worked example 2.1. Pointing to a photograph, Ravi said, "She is the daughter of my grandfather's only son." How is the woman in the photograph related to Ravi?

Solution. Draw a family tree. Ravi's grandfather → Ravi's grandfather's "only son" — since it's the ONLY son, and Ravi is presumably part of this family line, this only son IS Ravi's father (assuming Ravi is male and descended through this line — a standard assumption unless stated otherwise). The woman is "the daughter of [Ravi's father]" — meaning she is Ravi's father's daughter, i.e., Ravi's sister.

3. Coding-decoding — the pattern-first method

Worked example 3.1. If in a certain code, "FLOWER" is written as "GMPXFS", how would "GARDEN" be written in the same code?

Solution. First, identify the transformation pattern by comparing each letter's position, one pair at a time: F(6)→G(7): +1. L(12)→M(13): +1. O(15)→P(16): +1. W(23)→X(24): +1. E(5)→F(6): +1. R(18)→S(19): +1. Every letter shifts forward by exactly 1 in the alphabet — a consistent rule confirmed across ALL six letters, not just the first one or two. Applying the same +1 shift to GARDEN: G(7)→H(8), A(1)→B(2), R(18)→S(19), D(4)→E(5), E(5)→F(6), N(14)→O(15). GARDEN codes to HBSEFO.

4. Statement-assumption — must-be-true, not merely-reasonable

Worked example 4.1. Statement: "The government has announced free public transport for students to reduce urban traffic congestion." Assumption: "Students currently form a significant enough share of urban traffic for their transport-mode shift to meaningfully affect congestion." Is this assumption implicit in the statement?

Solution. Yes. The statement's LOGIC only holds if this assumption is true — if students were a negligible share of traffic, free transport for them wouldn't meaningfully "reduce congestion" as the statement claims is the goal. This assumption is a NECESSARY, unstated premise the argument depends on, distinct from a merely-related fact.

5. Statement-conclusion — what necessarily follows

Worked example 5.1. Statements: "All city buses run on electric power. Some electric vehicles receive a tax subsidy." Conclusion: "Some city buses receive a tax subsidy." Does this conclusion necessarily follow?

Solution. No. This is a classic syllogism trap. "All city buses run on electric power" places buses WITHIN the electric-vehicle category. "SOME electric vehicles receive a subsidy" does NOT specify WHICH electric vehicles — the subsidised subset might not include ANY buses at all. The conclusion is possible but NOT necessarily true from the given statements — a critical distinction between "possible" and "necessarily follows."

Common traps UPSC sets here

• In seating/arrangement puzzles, always check EVERY constraint against your candidate arrangement before finalising it — a single missed constraint check (like the adjacency rule in Worked example 1.1) can produce a wrong final answer even with correct initial placements.
• In blood-relation puzzles, "only son/daughter" is a strong, specific clue that eliminates ambiguity — don't treat it as equivalent to "a son/daughter" (implying possibly one of several).
• In coding-decoding, test your hypothesised rule against ALL given letter pairs, not just the first one or two, before applying it to the new word — a rule that fits the first two letters may fail on the fifth.
• In syllogism/statement-conclusion questions, "SOME X are Y" does NOT mean "ALL X are Y," and does NOT tell you WHICH specific X's are Y — Worked example 5.1's trap is a direct application of this: "some electric vehicles" being subsidised doesn't specify buses are among them.
• Assumption questions test what the argument's LOGIC REQUIRES, not what's merely true or related — an assumption must be NECESSARY for the stated conclusion/goal to make sense, not just a plausible supporting fact.

Memory aids

• "Draw before you derive" — diagram every puzzle before attempting a verbal solution.
• "Only means only" — treat "only son/daughter" and similar exclusive language as eliminating ambiguity, not as a soft hint.
• "Test the whole word, not just the first letter" — for coding-decoding, verify a hypothesised rule against every letter pair before applying it.
• "Some ≠ all, and some doesn't say which" — the core syllogism trap, worth repeating as a fixed mental check.
• "Necessary for the argument, not just true in general" — the assumption-question filter.

Exam protocol

• For arrangement puzzles, after deriving a candidate solution, re-check EVERY given constraint against it one final time before marking your answer.
• For blood-relation puzzles, draw the family tree with clear generational levels (grandparent/parent/self) and gender markers as you read each clue.
• For coding-decoding, write out the given word-code pair letter-by-letter with positions before hypothesising a rule, and verify the rule against all letters before applying it to a new word.
• For syllogism/conclusion questions, be alert to "some" statements — they almost never license a "some...are...Y" conclusion about a specific named subgroup unless explicitly stated.
• For assumption questions, ask "does the argument's stated goal/logic collapse if this assumption is FALSE" — if yes, it's a valid assumption; if the argument still basically holds without it, it's not.