Syllogism Problems — Reversal Error That Fails Interview

Syllogism questions appear in almost every competitive aptitude test worth taking — IBPS, CAT, GATE, TCS, Infosys campus drives, and virtually every structured reasoning round at major tech companies. Recruiters love them because they test pure logical reasoning under time pressure. You cannot bluff your way through them, and experience alone does not help if you are working from the wrong mental model.

The frustrating reality is that most candidates try to answer syllogisms using common sense, and that is precisely the trap. In the real world, 'All doctors are humans' conjures images of hospitals and stethoscopes. In a syllogism, it is just a logical arrow: Doctor → Human. Your job is to follow arrows mechanically, ignoring everything your brain already knows about the subject matter. The moment you substitute real-world knowledge for stated premises, you have already failed the question — even if your conclusion happens to be factually correct.

By the time you finish this guide, you will have a repeatable three-step system that works on any syllogism question — including the possibility variants that eliminate 60% of candidates who learned only the basic rules. You will understand not just what each rule says but why it has the shape it does, so you can reconstruct the right answer under interview pressure even if you forget the formula. That depth of understanding is what separates a candidate who scores in the 95th percentile from one who memorised a table and hopes the questions are cooperative.

Why Syllogism Problems Expose Reversal Errors in Logical Reasoning

Syllogism problems test your ability to chain two given statements (premises) into a valid conclusion. The core mechanic: given "All A are B" and "All B are C", you must deduce "All A are C". The trap is reversal — assuming "All A are B" implies "All B are A" (converse error). In code, this mirrors assuming a bidirectional relationship from a unidirectional one, e.g., treating a parent-child mapping as symmetric. The key property: only certain premise pairs (e.g., universal-affirmative + universal-affirmative) yield a valid conclusion; others (e.g., particular-negative + universal-affirmative) produce no definite result. In practice, you evaluate using Venn diagrams or rules of distribution — the middle term (B) must appear exactly once in each premise and be distributed in at least one. This matters in systems that enforce access control or type hierarchies: a reversal error can grant unauthorized access or misclassify objects. For example, if "all admins can read" and "all readers can view", you cannot conclude "all viewers are admins" — yet naive rule engines often do.

The Core Framework — Statements, Conclusions, and the Four Statement Types

Every syllogism question has exactly two parts: Statements, which are the premises you must accept as absolutely and unconditionally true, and Conclusions, which are the options you must evaluate purely on the basis of those statements. Your only job is to determine whether each conclusion follows with logical necessity from the statements — not from your knowledge of the real world, not from what seems reasonable, not from what is factually accurate.

This single principle — ignore real-world knowledge entirely — eliminates more candidate errors than any memorized rule. A statement like 'All managers are lazy' is gospel for the duration of that question. Work with it.

There are four types of categorical statements, and recognizing them instantly is the foundation of every technique that follows:

Type A — Universal Affirmative: 'All S are P.' Every single member of S belongs to P without exception. In a Venn diagram, draw S completely inside P.

Type E — Universal Negative: 'No S is P.' The sets S and P are completely non-overlapping — zero shared members in any valid diagram. Draw two entirely separate circles.

Type I — Particular Affirmative: 'Some S are P.' At least one member of S is also in P. The circles overlap, but neither is fully contained by the other in the default conservative diagram. Do not assume one is inside the other unless forced by additional statements.

Type O — Particular Negative: 'Some S are not P.' At least one member of S falls outside P. This is the subtlest type — it does not say most, many, or even a significant portion. It says at least one, which is compatible with nearly anything else.

The classical mnemonic: the letters A, E, I, O come from the Latin words AffIrmo (I affirm) for the positive types A and I, and nEgO (I deny) for the negative types E and O. This mnemonic has survived two thousand years of logic pedagogy because it maps cleanly to the structure of the statements themselves.

Production-level insight that most preparation materials miss: misclassifying a statement type is not just one error — it is an error that propagates through every subsequent step. If you read 'Some S are P' as a Type A statement, your diagram is wrong, your distribution analysis is wrong, your chain validation is wrong, and your conclusion is wrong. All from one misread. This is why the classification step must happen on paper, explicitly, before you draw a single circle.

=== STATEMENT TYPE QUICK-REFERENCE ===

Type A | Universal Affirmative   | "All cats are animals."
       | Venn diagram             | [cats] is fully INSIDE [animals]
       | What it guarantees       | Every cat is an animal. Zero exceptions.
       | What it does NOT say     | All animals are cats (WRONG — never flip A)
       | Reversible?              | NO — direction is fixed and one-way

Type E | Universal Negative      | "No cats are dogs."
       | Venn diagram             | [cats] and [dogs] are COMPLETELY SEPARATE circles
       | What it guarantees       | Zero cats are dogs. Zero dogs are cats.
       | What it does NOT say     | Anything about animals outside these two sets
       | Reversible?              | YES — 'No cats are dogs' → 'No dogs are cats' is free

Type I | Particular Affirmative  | "Some cats are black."
       | Venn diagram             | [cats] and [black] OVERLAP partially
       | What it guarantees       | At least 1 cat is black.
       | What it does NOT say     | Most or all cats are black
       | Reversible?              | YES — 'Some cats are black' → 'Some black things are cats' is free

Type O | Particular Negative     | "Some cats are not friendly."
       | Venn diagram             | At least 1 cat sits OUTSIDE [friendly]
       | What it guarantees       | At least 1 cat is unfriendly.
       | What it does NOT say     | Most cats are unfriendly / All cats are unfriendly
       | Reversible?              | NO — 'Some S are not P' does NOT give 'Some P are not S'

=== REVERSIBILITY SUMMARY ===
  Type A → NEVER reversible
  Type E → ALWAYS reversible (symmetric non-overlap)
  Type I → ALWAYS reversible (symmetric partial overlap)
  Type O → NEVER reversible (asymmetric partial exclusion)

=== SOLVED EXAMPLE 1 — Basic Type A + Type A Chain ===

Statements:
  (1) All roses are flowers.      [Type A]
  (2) All flowers are beautiful.  [Type A]

Conclusions:
  I.  All roses are beautiful.
  II. All beautiful things are roses.

Step 1 — Classify:
  Both are Type A. Middle term: flowers.
  'Flowers' is the predicate of statement 1 and the subject of statement 2.
  As subject of Type A, 'flowers' is distributed in statement 2. Chain is valid.

Step 2 — Draw the Venn diagram:
  [roses] inside [flowers] inside [beautiful]
  The nesting is clean — every rose is inside flowers, every flower is inside beautiful.

Step 3 — Test Conclusion I: "All roses are beautiful"
  Every rose → inside flowers → inside beautiful.
  In EVERY valid diagram, roses are inside beautiful. ✅ FOLLOWS

Step 4 — Test Conclusion II: "All beautiful things are roses"
  [beautiful] is the outermost circle.
  Sunsets, paintings, music — all can be beautiful without being roses.
  No statement forces every beautiful thing into [roses].
  In a valid diagram, [beautiful] extends far beyond [roses]. ❌ DOES NOT FOLLOW

Answer: Only Conclusion I follows.
What went wrong with II: Type A reversal trap — the most common error in this topic.

The Venn Diagram Method — A 3-Step System That Works on Every Syllogism

The Venn diagram method is not the most impressive-sounding approach, but it is the most reliable one — which is exactly what you need under timed exam conditions. Its power comes from forcing you to represent only what each statement literally guarantees, nothing more and nothing less. You cannot accidentally assume more than the statement says when the diagram won't let you add circles that aren't justified.

Here is the complete three-step system:

Step 1 — Classify each statement as A, E, I, or O. Write the label next to it on paper. This tells you the shape of each diagram element before you draw anything.

Step 2 — Draw the most conservative diagram the statements permit. 'All S are P' forces S inside P. 'Some S are P' forces a partial overlap, but you do NOT draw S inside P unless another statement requires it. When in doubt, draw the minimum relationship the statement forces and nothing beyond that.

Step 3 — Test each conclusion using the falsifiability principle. Ask: is there any valid Venn diagram, consistent with all given statements, where this conclusion is FALSE? If even one such diagram exists, the conclusion does not follow as definite. This is the principle most preparation materials describe as 'must be true' versus 'may be true.'

The falsifiability test is the conceptual core of the entire method. It is why 'Some S are P' does not allow you to conclude 'All S are P' — you can draw a valid diagram where only half of S overlaps P, making 'All' false in that diagram. The conclusion fails because it does not hold in all valid diagrams.

For two-statement problems, the middle term — the noun that appears in both statements — is your chain link. The distribution rule determines whether that chain is valid: a term is distributed when the statement makes a claim about every member of that category. In 'All S are P', S is distributed (we claim something about every S) but P is not (we say nothing about every P). In 'No S is P', both S and P are distributed. In 'Some S are P', neither is distributed. The middle term must be distributed in at least one statement — if it is undistributed in both, no valid chain forms and no definite conclusion follows, regardless of how plausible the conclusion sounds.

=== 3-STEP VENN DIAGRAM METHOD — WORKED EXAMPLES ===

--- EXAMPLE 2: Type I + Type A combination ---

Statements:
  (1) Some doctors are singers.   [Type I]
  (2) All singers are dancers.    [Type A]

Conclusions:
  I.  Some doctors are dancers.
  II. All dancers are doctors.
  III.Some singers are doctors.

--- Step 1: Classify ---
  Statement 1 → Type I  (partial overlap: doctors ↔ singers)
  Statement 2 → Type A  (singers fully inside dancers)
  Middle term: singers
  'Singers' is the subject of statement 2 (Type A) → distributed. Chain is valid.

--- Step 2: Draw the most conservative diagram ---

  [dancers — outermost circle, biggest]
      [singers — fully inside dancers, per statement 2]
          [doctors — partially overlaps singers, per statement 1]
              [doctors extends OUTSIDE dancers for the non-singer portion]

  Why does the doctors circle extend outside dancers?
  Statement 1 says only SOME doctors are singers.
  The remaining doctors have no stated relationship to dancers.
  Adding them inside dancers would assert information the statements do not provide.
  Conservative rule: if a statement doesn't force a relationship, don't draw one.

--- Step 3: Test each conclusion ---

  Conclusion I: "Some doctors are dancers"
    The doctors who ARE singers sit inside [singers].
    [singers] is fully inside [dancers] per statement 2.
    Therefore those doctors are inside [dancers].
    This holds in EVERY valid diagram that satisfies both statements. ✅ FOLLOWS

  Conclusion II: "All dancers are doctors"
    [dancers] is the outermost circle.
    Singers who are dancers but not doctors can exist — the diagram shows
    part of [singers] (and therefore [dancers]) outside the doctors overlap.
    We can draw valid diagrams where dancers exist entirely outside [doctors]. ❌ DOES NOT FOLLOW
    Note: this is also a Type A reversal trap applied to the conclusion.

  Conclusion III: "Some singers are doctors"
    Statement 1 says some doctors are singers.
    Type I is reversible — if some doctors are singers, then some singers are doctors.
    This follows directly from reversibility without any additional chain analysis. ✅ FOLLOWS

Answer: Conclusions I and III follow.

--- EXAMPLE 3: Type E + Type I — the classic trap ---

Statements:
  (1) No pens are pencils.    [Type E]
  (2) Some pencils are books. [Type I]

Conclusions:
  I.  No pens are books.
  II. Some books are not pens.

--- Step 2: Draw ---
  [pens] and [pencils] → completely separate (Type E, zero overlap)
  [pencils] and [books] → partially overlapping (Type I)

  Key question: where do [pens] and [books] sit relative to each other?
  The statements give no information about the pens-books relationship.
  We can draw pens overlapping books (valid) OR pens outside books (valid).
  Both diagrams satisfy all given statements. The relationship is genuinely unknown.

--- Step 3: Test ---

  Conclusion I: "No pens are books"
    Invalid — we can draw a legitimate diagram where pens DO overlap books
    while satisfying both given statements. That diagram makes Conclusion I false.
    Since there exists a valid diagram where it fails → ❌ DOES NOT FOLLOW as definite.

  Conclusion II: "Some books are not pens"
    The books that overlap with pencils are definitely not pens.
    Why? Because pens and pencils are completely separate (Type E).
    So no pencil is a pen, and those pencil-overlapping books are therefore not pens.
    This holds in every valid diagram → ✅ FOLLOWS

Answer: Only Conclusion II follows.
The trap with Conclusion I: absence of a connection does not prove a universal negative.
You need a definite E-chain to conclude 'No X are Y'. Absence of information is not 'No'.

=== DISTRIBUTION TABLE — when does a term cover all its members? ===

  Statement Type  | Subject Distributed? | Predicate Distributed?
  ----------------|---------------------|----------------------
  A (All S are P) | YES                 | NO
  E (No S is P)   | YES                 | YES
  I (Some S are P)| NO                  | NO
  O (Some S not P)| NO                  | YES

  Chain rule: the middle term MUST be distributed in at least one statement.
  If undistributed in both → no definite conclusion follows, regardless of content.

=== COMBINATION TABLE — derive conclusions without drawing ===

  S1 Type + S2 Type → Strongest definite conclusion (if middle term is distributed)
  A + A → A         (All S1-subject are S2-predicate)
  A + E → E         (No S1-subject is S2-predicate)
  E + A → O*        (Some S2-predicate are not S1-subject — reversed direction)
  I + A → I         (Some S1-subject are S2-predicate)
  I + E → O         (Some S1-subject are not S2-predicate)
  I + I → No definite conclusion (middle term undistributed in both)
  O + anything → No definite conclusion (particular negative weakens chain)
  E + E → No definite conclusion (two negatives cannot create positive link)

  Note: O* means the conclusion is in the reversed subject-predicate direction.

Possibility Cases — The Question Type That Eliminates 60% of Candidates

Once you are comfortable with definite conclusions, aptitude tests throw possibility conclusions at you. These look like: 'Some roses can be trees' or 'It is possible that all cats are dogs' or 'All managers being singers is a possibility.' They are designed to catch candidates who learned the basic rules and stopped there — because the possibility framework runs on exactly opposite logic from the definite conclusion framework.

Here is the key insight: a possibility conclusion is TRUE if there exists even one valid Venn diagram — without contradicting any given statement — where it holds. You do not need it to be true in all diagrams. You need it to be true in at least one.

Conversely, a possibility conclusion is FALSE only if every valid Venn diagram makes it impossible — meaning the given statements definitively rule it out across all cases.

This flips the direction of your test. For definite conclusions, you ask: 'Is there any valid diagram where this fails?' If yes, it fails. For possibility conclusions, you ask: 'Is there any valid diagram where this holds?' If yes, it succeeds.

The golden rule for possibilities: if two groups are not universally separated by a Type E chain — meaning no sequence of statements establishes 'No A is C' definitively — then it is always possible for them to overlap. Even if no definite conclusion links them positively, the possibility of overlap generally survives. The absence of a definite positive relationship is not the same as the presence of a definite negative one.

Conversely, if the given statements establish a definite negative relationship — like 'All A are B' combined with 'No B is C' yields definitively 'No A is C' — then 'Some A can be C' is impossible, not just uncertain. A definitively derived 'No X is Y' kills the corresponding possibility completely.

The mental model you need: definite conclusions use the universal quantifier (must hold in every valid diagram). Possibility conclusions use the existential quantifier (must hold in at least one valid diagram). Mixing these two quantifiers is the core error that sends candidates to wrong answers on possibility questions.

One more pattern that appears frequently in exams: the complementary pair. When neither 'Some A are C' nor 'Some A are not C' follows as a definite conclusion, both can be possibilities simultaneously. The exam answer in this case is often 'Either I or II follows' — meaning one of the pair must be true without being able to specify which. Recognizing this pattern immediately when I + I or O + anything appears as the premise combination saves significant time.

=== POSSIBILITY CONCLUSIONS — COMPLETE FRAMEWORK ===

Core Rule Comparison:
  Definite conclusion  → must hold in ALL valid diagrams
  Possibility conclusion → must hold in AT LEAST ONE valid diagram

Possibility is BLOCKED when:
  The given statements definitively establish the OPPOSITE relationship.
  Example: if statements yield 'No A is C' as a definite conclusion,
  then 'Some A can be C' is IMPOSSIBLE — not just unprovable.

Possibility SURVIVES when:
  No E-type chain definitively separates the two terms being tested.
  Even if no positive definite conclusion links them, overlap remains possible.

--- EXAMPLE 4: Possibility from I + A combination ---

Statements:
  (1) All roses are flowers.   [Type A]
  (2) Some flowers are red.    [Type I]

Definite Conclusions (applying combination table: A + I in reverse = I + A → I):
  'Some roses are red' — does this hold in ALL valid diagrams?
  Draw [roses] inside [flowers]. Draw [red] overlapping [flowers].
  The [red] overlap with [flowers] might sit entirely in the non-rose part of [flowers].
  In that diagram, zero roses are red. Conclusion fails in that diagram. ❌ NOT DEFINITE.

Possibility Conclusion: 'All roses being red is a possibility'
  Can we draw a valid diagram where ALL roses are red, without violating any statement?
  Draw [roses] inside [flowers], draw [red] fully containing [roses] while overlapping [flowers].
  Statement 1 satisfied: roses inside flowers. ✅
  Statement 2 satisfied: some flowers (the rose ones) are inside red. ✅
  The diagram is valid and all roses are red in it. POSSIBILITY FOLLOWS. ✅

--- EXAMPLE 5: Possibility blocked by definite E chain ---

Statements:
  (1) All A are B.    [Type A]
  (2) No B is C.      [Type E]

Definite conclusion chain:
  A+E combination → definite conclusion: No A is C (E type, using A+E → E rule)
  Verify: every A is inside B, and no B overlaps C → every A is outside C → No A is C. ✅

Possibility test: 'Some A can be C'
  The definite conclusion 'No A is C' holds in EVERY valid diagram.
  There is no valid diagram where any A is inside C.
  The possibility is completely blocked by the definite negative chain. ❌ IMPOSSIBLE.

--- EXAMPLE 6: Complementary pair — Either/Or answer ---

Statements:
  (1) Some cats are dogs.   [Type I]
  (2) Some dogs are birds.  [Type I]

Middle term: dogs — appears as predicate in both → undistributed in both → I + I.
No definite conclusion about cats and birds follows.

Conclusions presented:
  I.  Some cats are birds.
  II. Some cats are not birds.

Can we draw a valid diagram where some cats ARE birds?
  Put cats and dogs overlapping, dogs and birds overlapping, and the cats-dogs overlap
  also inside birds. Valid — satisfies both I statements. Conclusion I can hold. ✅ POSSIBLE

Can we draw a valid diagram where some cats are NOT birds?
  Put cats and dogs overlapping outside the bird circle entirely. Valid.
  Conclusion II can hold. ✅ POSSIBLE

Both are possible. Neither is definite. Together they form a complementary pair
(one of them MUST be true in any given scenario). → 'Either I or II follows' is correct.

=== POSSIBILITY QUICK-REFERENCE ===

  Situation                          | Possibility verdict
  -----------------------------------|--------------------
  No statement connects terms A & C  | A and C CAN overlap → possibility SURVIVES
  Definite 'Some A are C' derived    | Stronger than possibility — definite follows too
  Definite 'No A is C' derived       | Possibility is BLOCKED entirely
  I + I premises → no definite       | Possibility usually survives; check for Either/Or
  O + anything → no definite         | Possibility usually survives for non-excluded term
  A + E → definite No link           | Possibility blocked for that exact relationship

The 5-Second Shortcut — Using the Combination Table Under Time Pressure

The Venn diagram method is complete and reliable, but in competitive exams you often need to process 5-6 syllogism questions in under three minutes. For straightforward two-statement problems, the combination table lets you derive the strongest possible definite conclusion in under 10 seconds — no diagram required.

The table works by mapping statement type pairings to the type of conclusion they produce. Once you classify both statements and identify the middle term, you look up the combination and immediately know the answer type. Then you scan the options for the one that matches that type, verify the subject-predicate direction, and move on.

The complete set of productive combinations: A + A → A (Universal Affirmative — 'All S1-subject are S2-predicate') A + E → E (Universal Negative — 'No S1-subject is S2-predicate') E + A → O (Particular Negative reversed — 'Some S2-predicate are not S1-subject') I + A → I (Particular Affirmative — 'Some S1-subject are S2-predicate') I + E → O (Particular Negative — 'Some S1-subject are not S2-predicate') A + I → I (Particular Affirmative reversed — 'Some S2-predicate are S1-subject')

The starred entries () indicate reversed conclusions — the subject and predicate swap direction. This is the source of the O error in the E + A combination, where candidates often get the direction backwards.

All other combinations — I + I, O + anything, E + E, E + O, O + O — yield no definite conclusion. When you see these pairings, stop computing and go directly to possibility and Either/Or analysis.

The method's limitation: it works for the strongest single conclusion from a linear two-statement chain. For problems with three statements, longer chains, or constraint conditions, you still need the full Venn diagram. The combination table is a speed tool for the most common question format, not a replacement for the underlying logical framework.

=== 5-SECOND COMBINATION TABLE ===

  S1 + S2        | Conclusion Type | Direction          | Example
  ---------------|-----------------|--------------------|-----------------------------------------
  A + A          | A               | S1-subj → S2-pred  | All roses are flowers + All flowers are red → All roses are red
  A + E          | E               | S1-subj ≠ S2-pred  | All dogs are pets + No pets are wild → No dogs are wild
  E + A          | O*  (reversed)  | S2-pred ≠ S1-subj  | No fish are birds + All birds are animals → Some animals are not fish
  I + A          | I               | S1-subj → S2-pred  | Some cats are pets + All pets are loved → Some cats are loved
  A + I          | I*  (reversed)  | S2-pred → S1-subj  | All roses are flowers + Some flowers are red → Some red things are roses
  I + E          | O               | S1-subj ≠ S2-pred  | Some books are pens + No pens are pencils → Some books are not pencils
  E + I          | O*  (reversed)  | S2-pred ≠ S1-subj  | No cats are dogs + Some dogs are pets → Some pets are not cats

  NON-PRODUCTIVE COMBINATIONS (no definite conclusion):
  I + I          | None            | —                  | Middle term undistributed in both premises
  O + A          | None            | —                  | Particular negative weakens the chain irreparably
  O + E          | None            | —                  | Two negatives + particular = no positive link
  E + E          | None            | —                  | Two universally negative premises cancel each other
  O + O          | None            | —                  | Two particular negatives yield nothing

=== APPLYING THE TABLE — Step by step ===

  EXAM QUESTION:
  Statements:
    (1) No dogs are cats.    [Type E]
    (2) All cats are animals. [Type A]

  Step 1: Identify types → E + A
  Step 2: Look up E + A → O* (reversed direction)
  Step 3: O* direction means: Some S2-predicate are not S1-subject
           S2-predicate = animals (predicate of statement 2)
           S1-subject   = dogs (subject of statement 1)
           Conclusion   = Some animals are not dogs
  Step 4: Scan options for 'Some animals are not dogs' → match found
  Total time: approximately 8 seconds

  WHY the conclusion is O* not E:
  'No dogs are cats' + 'All cats are animals' does not mean 'No dogs are animals'.
  That would require all animals to be cats, which statement 2 does not say.
  The E + A combination can only support the weaker O* conclusion, not a full E.
  Candidates who conclude 'No dogs are animals' have over-strengthened from E to full E
  when only O* is supported — a classic exam trap.

=== DIRECTION MEMORY AID FOR REVERSED CONCLUSIONS ===

  For O* (from E + A or E + I):
    The conclusion subject comes from S2's predicate
    The conclusion predicate comes from S1's subject
    Read backwards: S2-pred are not S1-subj

  Memory shortcut: when the combination is 'E + A', think 'the answer
  goes home to the A statement's predicate first, then points back at E's subject.'

The Only Syllogism That Matters for FAANG — The "Either-Or" Trap

You've memorised the four statement types. You can slam a Venn diagram in ten seconds. Great. Then Amazon throws you an "Either A or B" conclusion with a "Some A are not B" premise and watches you choke.

Here's the war story: The "either-or" conclusion is the single most gamed question in every major tech interview. Why? Because it's not about logic. It's about definition enforcement. In standard logic, "Either A or B" means exactly one of them is true. But in syllogism problems—especially at FAANG—they use the inclusive OR: at least one is true, possibly both. Candidates who don't ask "which OR are we using?" burn twenty seconds and three points.

The fix: Always split the conclusion into its two exclusive cases. If both cases fail under the same Venn, it's invalid. If one case holds, it's "either-or" valid only if the problem explicitly states exclusive OR. Most problems don't. They want you to catch that "either-or" with overlapping sets is automatically false unless the sets are mutually exclusive. Production lesson: never assume the semantics of your input. Validate the contract first.

// io.thecodeforge — interview tutorial

// Detect whether an "either-or" conclusion holds
// under the standard syllogism inclusive OR.

def check_either_or(premises, conclusion_a, conclusion_b):
    # Model sets as Venn regions
    # Return True if at least one conclusion holds
    # (inclusive OR default)
    a_valid = all(premises[conclusion_a] for p in premises)
    b_valid = all(premises[conclusion_b] for p in premises)
    return a_valid or b_valid

# Example: All A are B. Some C are A.
# Conclusion: Either Some C are B OR Some C are not B
# Both can't be true simultaneously if sets are overlapping.
premises = {
    "All A are B": True,
    "Some C are A": True
}

result = check_either_or(premises, "Some C are B", "Some C are not B")
print(f"Either-or holds under inclusive OR: {result}")

Production-Ready Syllogism: Modeling Ambiguity With Null Regions

You think Venn diagrams are the gold standard? They're a crutch. Every production system that processes syllogisms—yes, they exist in NLP parsers, rule engines, and compliance checkers—uses a region-based model. Three circles, seven non-overlapping regions. The trick isn't drawing the circles. It's tracking which regions are null, which are non-null, and which are flat-out impossible.

Here's the pattern that kills seniors: "All A are B" plus "No B are C." Classic. But the Venn shows A sits inside B, and B has zero overlap with C. So A and C are disjoint. That's a valid conclusion. Easy. But when you have "Some A are B" plus "Some B are not C," your Venn has two possible layouts: A could be entirely inside B's non-C region, or partially inside B's C-overlap region. Ambiguity. That's not a bug—it's a feature. In production, you model ambiguity as a set of possible worlds, not a single diagram.

The implementation: Represent each region as a bitmask (7 bits for 3-set Venn). Mark regions as 'occupied', 'empty', or 'ambiguous'. Then, for a conclusion, you check if it's true in all possible worlds (definite) or just some (possible). FAANG interviewers watch you do this in Python. They don't want the answer. They want to see you handle uncertainty like a system that won't crash on bad input. That's the difference between a coder and an engineer.

// io.thecodeforge — interview tutorial

// Model 3-set Venn regions as 7-bit masks.
// Track occupancy states: occupied, empty, ambiguous.

from enum import Enum

class RegionState(Enum):
    OCCUPIED = 1
    EMPTY = 2
    AMBIGUOUS = 3

# Regions: [A_only, AB, B_only, BC, C_only, AC, ABC]
regions = [RegionState.EMPTY] * 7

# Apply "All A are B": A_only is empty, everything A must be inside AB or ABC
regions[0] = RegionState.EMPTY  # A_only empty
regions[1] = RegionState.AMBIGUOUS  # AB can be anything
regions[6] = RegionState.AMBIGUOUS  # ABC can be anything

# Apply "No B are C": BC empty, ABC empty
regions[3] = RegionState.EMPTY
regions[6] = RegionState.EMPTY

# Check conclusion: No A are C -> A and C can't share region AC or ABC
# AC is region index 5, ABC is 6
if regions[5] == RegionState.EMPTY and regions[6] == RegionState.EMPTY:
    print("Conclusion 'No A are C' is DEFINITE")
else:
    print("Conclusion 'No A are C' is POSSIBLE")

The Two Pillars of Immediate Inference: Conversion and Obversion

Most syllogism training jumps straight to Venn diagrams or combination tables, skipping the two most fundamental inference techniques: conversion and obversion. Conversion swaps subject and predicate. For 'All A are B,' conversion gives 'Some B are A.' For 'No A is B,' it gives 'No B is A.' For 'Some A are B,' it gives 'Some B are A.' The critical trap: 'Some A are not B' cannot be converted. Obversion flips quality and negates the predicate. 'All A are B' obverts to 'No A is non-B.' 'No A is B' becomes 'All A is non-B.' These two methods are the bedrock for solving 'possibility' and 'either-or' conclusions under time pressure. When you see a conclusion like 'Some B are A,' ask: can I derive this from the given statement by conversion? If not, it's false. Interviewers use this to test your understanding of logical versus natural language.

// io.thecodeforge — interview tutorial

# Conversion: All A are B → Some B are A
def convert(statement):
    if 'All' in statement:
        return statement.replace('All', 'Some').split(' are ')[::-1]
    elif 'No' in statement:
        return 'No ' + statement.split(' is ')[1] + ' is ' + statement.split(' is ')[0]
    elif 'Some' in statement and 'not' not in statement:
        parts = statement.split(' are ')
        return 'Some ' + parts[1] + ' are ' + parts[0]
    return 'Cannot convert'

print(convert('All cats are mammals'))  # Some mammals are cats
print(convert('No dog is cat'))  # No cat is dog

The 'Only a few' Construction — Nuanced Partial Overlap

The phrase 'Only a few X are Y' appears frequently in high-level syllogism questions. It means: some X are Y, but most X are not Y. Crucially, this implies that Y can be completely inside X, partially overlapping, or even a subset of X — but some X must fall outside Y. This is not the same as 'Some X are Y,' which leaves open the possibility that all X are Y. When combined with other statements, 'Only a few' introduces a necessary exclusion region. For example: 'Only a few land is red. All red is pink. Some pink is tree.' From this, you can conclude that some land is not red, and some land could be pink — but you cannot conclude that no red is tree, because the 'only a few' constraint only affects land, not pink or tree. The standard error is treating 'Only a few' as synonymous with 'Some.' It is not. The 'only a few' forces at least one element of X to be outside Y.

// io.thecodeforge — interview tutorial

# Model 'Only a few X are Y' as: some X in Y, some X not in Y
def only_a_few(x, y):
    return f"At least one {x} is {y}, but not all {x} are {y}"

def check_conclusion(stmt, conc):
    # Example: Only a few land is red. Some land can be pink?
    if 'land' in stmt and 'red' in stmt:
        return 'possible' if 'pink' in conc else 'invalid'
    return 'unknown'

print(only_a_few('land', 'red'))
print(check_conclusion('Only a few land is red', 'Some land can be pink'))

The 'Only X are Y' Identity — Exhaustive Containment

The statement 'Only X are Y' means that if something is Y, it must be X. In formal logic, this is the converse of 'All Y are X.' All members of Y are contained within X, but X may have members that are not Y. For example: 'Only perfumes are quiet' means every quiet thing is a perfume — but not all perfumes are necessarily quiet. This is a common reversal trap. When you then combine 'Only perfumes are quiet' with 'Some perfumes are shift,' you cannot conclude anything about quiet and shift directly. The correct inference: since all quiet are perfumes, and some perfumes are shift, it is possible that some quiet are shift — but not necessary. The phrase 'Only' forces a one-way relationship: Y implies X, but X does not imply Y. Draw it as a circle of Y entirely inside X. Interviewers test this by giving 'Only X are Y' and then asking if 'All Y are X' is true (it is) or if 'All X are Y' is true (it is not).

// io.thecodeforge — interview tutorial

# Only X are Y means: All Y are X
def only_are(x, y):
    return f"All {y} are {x}"

def check_identity(statement, conc):
    if 'Only' in statement:
        # parse: Only X are Y
        parts = statement.replace('Only ', '').split(' are ')
        x, y = parts[0], parts[1]
        if conc == f"All {y} are {x}":
            return True
        elif conc == f"All {x} are {y}":
            return False
    return None

print(only_are('perfumes', 'quiet'))
print(check_identity('Only perfumes are quiet', 'All quiet are perfumes'))
print(check_identity('Only perfumes are quiet', 'All perfumes are quiet'))

Venn Diagram Method for Syllogism: All, Some, No

The Venn diagram method is a visual approach to solving syllogisms by representing sets as overlapping circles. This technique is particularly effective for handling 'All', 'Some', and 'No' statements, which form the foundation of categorical logic. To apply this method, follow a three-step system: first, draw two overlapping circles representing the two categories (e.g., A and B). Second, shade regions that are empty based on universal statements ('All A are B' shades the part of A outside B; 'No A are B' shades the overlap). Third, place an 'X' in regions that are occupied for particular statements ('Some A are B' places an X in the overlap; 'Some A are not B' places an X in the part of A outside B). For example, consider the statements: 'All cats are mammals' and 'Some pets are cats'. Draw circles for 'cats' and 'mammals' with cats inside mammals (shade cats outside mammals). Then draw circles for 'pets' and 'cats' with an X in the overlap of pets and cats. The conclusion 'Some pets are mammals' is valid because the X in the pets-cats overlap lies within the mammals region. This method eliminates ambiguity and is production-ready for interviews. Practice with combinations like 'All A are B, No B are C' to see that 'No A are C' follows. The Venn diagram method is a reliable system that works on every syllogism, making it indispensable for FAANG interviews.

def venn_syllogism(statements):
    # Simplified implementation for demonstration
    # Returns valid conclusions based on Venn logic
    # Example: statements = [('All', 'A', 'B'), ('Some', 'B', 'C')]
    # Returns ['Some A are C']
    pass

Either-Or Cases and Possibility Cases

Either-or cases and possibility cases are advanced syllogism types that frequently eliminate candidates in FAANG interviews. An either-or case presents two conclusions where exactly one must be true (e.g., 'Either Some A are B or No A are B'). A possibility case asks if a conclusion is possible given the statements (e.g., 'It is possible that all A are B'). To solve these, use the combination table: list all possible relationships between sets (e.g., disjoint, partial overlap, subset). For either-or, check if the two conclusions are complementary (cover all possibilities) and mutually exclusive. For example, given 'Some A are B', the conclusions 'Some A are not B' and 'All A are B' form an either-or pair because exactly one must hold. For possibility cases, verify if there exists a scenario consistent with the statements. For instance, from 'No A are B', it is possible that 'All A are C' and 'Some C are B'? No, because A and B are disjoint. The key is to test with a single counterexample. Practice with: Statements: 'All pens are pencils. No pencils are erasers.' Conclusions: 'Either some pens are erasers or no pens are erasers.' The correct answer is 'No pens are erasers' follows, so the either-or is invalid because one conclusion is definitely true. Master these to avoid reversal errors.

def check_either_or(statements, conclusion1, conclusion2):
    # Returns True if exactly one of the conclusions must be true
    # Example: statements = [('All', 'A', 'B')]
    # conclusion1 = ('Some', 'A', 'B'), conclusion2 = ('No', 'A', 'B')
    # Returns False because 'All A are B' implies 'Some A are B' is true
    pass

Reverse Syllogism: Finding Conclusions from Given Statements

Reverse syllogism is a common interview twist where you are given statements and must identify which conclusions logically follow. Unlike traditional syllogisms, here you often have multiple conclusions to evaluate. The key is to apply the core framework: convert statements into immediate inferences (conversion and obversion) and then combine using the combination table. For example, given 'All A are B' and 'No B are C', possible conclusions: 'No A are C' (valid), 'Some A are not C' (valid), 'Some C are not A' (valid). But 'Some A are C' is invalid. To solve systematically: list all statements, derive their conversions (e.g., 'All A are B' converts to 'Some B are A'), then check each conclusion against the combined set. Use the Venn diagram method to verify. Practice with: Statements: 'Only a few A are B. Some B are C.' Conclusions: 'All A being C is a possibility' (true), 'Some A are not C' (false). Reverse syllogism tests your ability to think backwards and avoid reversal errors. Production tip: In data analysis, reverse syllogism helps infer missing relationships from partial data.

def reverse_syllogism(statements, conclusions):
    # Returns list of conclusions that logically follow
    # Example: statements = [('All', 'A', 'B'), ('No', 'B', 'C')]
    # conclusions = [('No', 'A', 'C'), ('Some', 'A', 'C')]
    # Returns [('No', 'A', 'C')]
    pass