fallacies

overview

fallacies are errors in reasoning that make arguments invalid or unsound. they appear everywhere - from political debates to everyday conversations to automated reasoning systems. understanding fallacies helps identify weak arguments and construct stronger ones.

this section covers both formal fallacies (structural errors in logic) and informal fallacies (content and context errors), with practical examples and detection methods for both human reasoning and automated systems.

types of fallacies

formal fallacies

errors in the logical structure of arguments. even with true premises, the conclusion doesn’t follow.

affirming the consequent:
if it rains, the ground is wet    (premise)
the ground is wet                 (premise)
∴ it's raining                    (invalid conclusion)

the structure is invalid because sprinklers, flooding, or spilled water could also wet the ground.

characteristics:

• structural errors: violate rules of valid inference
• form-dependent: identifiable by argument pattern
• validity issues: make arguments invalid regardless of content
• formalizable: can be detected algorithmically

learn more about formal fallacies →

informal fallacies

errors in the content, assumptions, or context of arguments. the logical structure may be valid, but the reasoning is flawed.

ad hominem:
sarah proposes increasing school funding    (premise)
sarah never went to college herself        (irrelevant attack)
∴ her proposal is wrong                     (fallacious conclusion)

the personal attack is irrelevant to the merit of the funding proposal.

characteristics:

• content errors: problems with premises, assumptions, or relevance
• context-dependent: meaning depends on subject matter
• pragmatic issues: often involve persuasion tactics
• harder to formalize: require domain knowledge to detect

learn more about informal fallacies →

common fallacies

ad hominem

attacking the person making an argument rather than the argument itself.

politician a: "we should invest in renewable energy"
politician b: "don't listen to him - he's been divorced twice"

personal characteristics are usually irrelevant to argument validity.

learn more about ad hominem →

straw man

misrepresenting someone’s argument to make it easier to attack.

actual argument: "we should have some gun regulations"
straw man version: "my opponent wants to ban all guns and leave us defenseless"

defeating the misrepresented version doesn’t address the original argument.

learn more about straw man →

false dilemma

presenting only two options when more alternatives exist.

"you're either with us or against us"

ignores neutral positions, partial agreement, or alternative approaches.

learn more about false dilemma →

fallacy categories

by logical type

by content type (informal)

by frequency in discourse

detection strategies

for formal fallacies

1. analyze argument structure

   • identify premises and conclusion
   • check if conclusion follows logically
   • test with truth tables or proof methods

2. pattern recognition

   • learn common invalid forms
   • use automated logic checkers
   • apply inference rules systematically

3. counterexample method

   • find cases where premises are true but conclusion false
   • if possible, the argument is formally invalid

for informal fallacies

1. relevance checking

   • do premises actually support the conclusion?
   • are personal attacks or irrelevant details present?

2. assumption analysis

   • what unstated assumptions are required?
   • are these assumptions reasonable?

3. context evaluation

   • is the argument appropriate for the domain?
   • are key alternatives being ignored?

automated detection

formal fallacy detection

# pseudocode for formal fallacy checking
def check_formal_validity(premises, conclusion):
    # convert to logical form
    logical_form = parse_to_logic(premises, conclusion)

    # check using proof system
    if can_derive(premises, conclusion):
        return "valid"
    else:
        return "formally invalid"

informal fallacy detection

# pseudocode for informal fallacy detection
def detect_informal_fallacies(argument):
    fallacies = []

    if contains_personal_attacks(argument):
        fallacies.append("ad_hominem")

    if misrepresents_position(argument):
        fallacies.append("straw_man")

    if limits_to_two_options(argument):
        fallacies.append("false_dilemma")

    return fallacies

practical applications

debate analysis

• political debates: identify when candidates avoid issues or attack opponents
• academic discussions: ensure arguments address actual positions
• public policy: evaluate reasoning behind proposals

media literacy

• news analysis: spot biased reporting and weak reasoning
• advertising: recognize manipulative persuasion tactics
• social media: identify fallacious arguments in discussions

automated reasoning

• argument mining: extract and evaluate arguments from text
• fact checking: identify weak or fallacious reasoning
• knowledge base validation: ensure logical consistency

education

• critical thinking: teach students to identify and avoid fallacies
• writing instruction: help construct stronger arguments
• scientific reasoning: distinguish valid from invalid inferences

common misconceptions

”fallacy = false”

misconception: if an argument contains a fallacy, its conclusion must be false

reality: fallacies make arguments weak or invalid, but conclusions might still be true for other reasons

fallacious argument: "einstein was a genius, so his political views must be correct"
(appeal to irrelevant authority)

reality: einstein's political views might be correct, but not because of his physics genius

”any criticism is ad hominem”

misconception: pointing out relevant flaws in credibility is always ad hominem

reality: credibility can be relevant when evaluating testimony or expertise

relevant: "this climate study was funded by oil companies with obvious bias"
ad hominem: "don't trust climate scientists because they're liberal"

”slippery slope is always fallacious”

misconception: any argument about consequences is a slippery slope fallacy

reality: some causal chains are well-supported; fallacy occurs when chain is unsupported

fallacious: "if we allow same-sex marriage, people will marry their pets"
reasonable: "if we cut safety inspections, more accidents will likely occur"

avoiding fallacies

in constructing arguments

1. separate person from position: attack ideas, not people
2. represent positions accurately: use strongest form of opponent’s argument
3. consider alternatives: avoid false dichotomies
4. check logical structure: ensure conclusions follow from premises
5. provide relevant evidence: make premises actually support conclusions

in evaluating arguments

1. identify the actual claim: what is really being argued?
2. check premise relevance: do premises relate to conclusion?
3. look for hidden assumptions: what unstated beliefs are required?
4. consider counterexamples: are there cases where premises are true but conclusion false?
5. evaluate evidence quality: is support adequate and appropriate?

computational approaches

natural language processing

# example: detecting straw man fallacies
def detect_straw_man(original_position, response):
    original_intent = extract_intent(original_position)
    response_target = extract_target_claim(response)

    similarity = semantic_similarity(original_intent, response_target)

    if similarity < 0.6:  # threshold for significant misrepresentation
        return "potential_straw_man"
    return "fair_representation"

machine learning approaches

• classification: train models to identify fallacy types from text
• similarity measures: detect misrepresentation in straw man fallacies
• context analysis: determine relevance for detecting ad hominem
• logical parsing: extract argument structure for formal fallacy detection

knowledge-based systems

• ontological reasoning: use formal fallacy definitions
• rule-based detection: implement fallacy patterns as logical rules
• argumentation frameworks: model defeat relations between arguments
• explanation generation: provide reasons why arguments are fallacious

further reading

classical sources

• aristotle: “sophistical refutations” (original fallacy analysis)
• francis bacon: “novum organum” (idols of the mind)
• john stuart mill: “system of logic” (informal fallacies)

modern treatments

• irving copi: “introduction to logic” (comprehensive fallacy coverage)
• douglas walton: “informal logic” (pragmatic approach to fallacies)
• charles hamblin: “fallacies” (critical historical analysis)

computational approaches

• floris bex: “arguments, stories and criminal evidence” (legal reasoning)
• chris reed: “argument & computation” (automated argument analysis)
• katarzyna budzynska: “argument mining” (extracting arguments from text)

online resources

• fallacyfiles.org - comprehensive fallacy database
• yourlogicalfallacyis.com - visual fallacy guide
• logical fallacies - purdue owl