Empirical pattern: Borderline cases elicit hedging and uncertainty.
For individuals near the inferred threshold:
- Speakers hedge or refuse to answer
- Judgments show high variance across informants
- Neither "A" nor "not A" feels fully appropriate
Source: @cite{lassiter-goodman-2017} Section 1, @cite{kennedy-2007}
- adjective : String
The adjective
- context : String
Description of the context/comparison class
- clearPositive : String
Clear positive case (definitely A)
- clearPositiveValue : String
Clear positive value
- clearNegative : String
Clear negative case (definitely not A)
- clearNegativeValue : String
Clear negative value
- borderline : String
Borderline case
- borderlineValue : String
Borderline value
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
House price borderline example.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Height borderline example.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Empirical pattern: The sorites paradox.
Given:
- A clearly Adj individual (premise 1)
- A tolerance principle: "If x is Adj and y is ε smaller, then y is Adj" (premise 2)
- Iterated application leads to a clearly non-Adj individual (conclusion)
Empirical observations:
- People accept premise 1 (the clear case)
- People accept individual instances of premise 2 (each step seems valid)
- People reject the conclusion (the absurd case)
- People show gradient acceptance as cases approach the threshold
Source: @cite{edgington-1997}, @cite{lassiter-goodman-2017} Section 5
- adjective : String
- scale : String
- toleranceGap : String
- clearPositive : String
- positiveValue : String
- clearNegative : String
- negativeValue : String
- numSteps : Nat
- acceptPremise1 : Bool
- acceptToleranceSteps : Bool
- acceptConclusion : Bool
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
The problem of higher-order vagueness.
If "tall" has borderline cases, what about "borderline tall"? Is there a sharp boundary between "borderline tall" and "clearly tall"?
- First-order vagueness: borderline cases of "tall"
- Second-order vagueness: borderline cases of "borderline tall"
- Third-order vagueness: borderline cases of "borderline borderline tall" -... and so on
This threatens any theory that posits sharp boundaries anywhere.
Source: @cite{fine-1975}, @cite{williamson-1994},
- basePredicate : String
- clearlyPositive : String
- borderline : String
- clearlyNegative : String
- sharpClearlyBorderline : Bool
- sharpBorderlineNot : Bool
- iteratesUpward : Bool
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
The "definitely" operator and higher-order vagueness.
If "Definitely tall" means "clearly tall" (not borderline), then:
- "Definitely tall" should have sharper boundaries than "tall"
- But "definitely" is itself vague.
- So we get: "borderline definitely tall"
Iterating: "definitely definitely tall", etc.
Source: @cite{fine-1975}, @cite{williamson-1994}
- predicate : String
- eliminatesBorderline : Bool
- definitelyIsVague : Bool
- borderlineDefinitely : Bool
- iterationHelps : Bool
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
Instances For
Penumbral connections: logical relationships that hold even in borderline cases.
Even if we don't know whether John is tall:
- "John is tall ∨ John is not tall" is true (excluded middle)
- "John is tall ∧ John is not tall" is false (non-contradiction)
- If John = 5'9" and Mary = 5'9", then "John is tall ↔ Mary is tall" (same-height)
These are "penumbral truths" - true in the borderline region.
Supervaluationism: true iff true on ALL precisifications. Degree theories: must explain why these have degree 1.
Source: @cite{fine-1975}, @cite{keefe-2000}
- connectionName : String
- formalStatement : String
- alwaysTrue : Bool
- borderlineExample : String
- supervaluationismCaptures : Bool
- degreeTheoryCaptures : Bool
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
The tolerance principle: the central ingredient in sorites paradoxes.
Tolerance: If x is F and y differs from x by only a tiny amount, then y is also F.
This seems true for vague predicates:
- 1mm can't make the difference between tall and not-tall
- $1 can't make the difference between expensive and not-expensive
- 1 grain can't make the difference between heap and not-heap
But iterated application leads to absurdity (the sorites).
Source: @cite{wright-1976}, @cite{fara-2000}
- predicate : String
- scale : String
- toleranceMargin : String
- individualStepAcceptable : Bool
- iterationAbsurd : Bool
- proposedResolution : String
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Probabilistic sorites analysis.
Each tolerance step is highly probable, not certain.
Let P(step) = 0.99 (each step is 99% acceptable) After N steps: P(all steps) = 0.99^N
For N = 762 (mm from 7'4" to 4'10"): 0.99^762 ≈ 0.0005 (extremely unlikely)
The paradox dissolves: the argument is valid but unsound. Each premise is probably true, but the conjunction is probably false.
Source: @cite{edgington-1997}, @cite{lassiter-goodman-2017} Section 5
- predicate : String
- singleStepProbability : Float
- numSteps : Nat
- cumulativeProbability : Float
- premise1Accepted : Bool
- eachStepAccepted : Bool
- fullArgumentAccepted : Bool
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Formal constraints that any adequate theory of vagueness should satisfy.
These are theory-neutral desiderata. A theory's success is measured by how well it accounts for these patterns.
Source: Various
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Interest-relativity of vague extension.
The extension of a vague gradable adjective can change when an agent's interests change, even if the precise measured property stays constant. This is evidence that the degrees of vague quantities incorporate information about interests, not just objective measurements.
Source: @cite{fara-2000}, @cite{dinis-jacinto-2026} §5.3
- adjective : String
- individual : String
- preciseProperty : String
- preciseValueUnchanged : Bool
- interestsChanged : String
- extensionBefore : Bool
- extensionAfter : Bool
- extensionFlipped : Bool
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Prince William / Charles III: as William ages, his interests shift toward considering people with m hairs bald whom he previously didn't. Charles III has the same number of hairs, but was bald∅ before and isn't now — his baldness degree changed because degrees partially reflect interests.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Context-dependent expensiveness: same price, different interests.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Tolerance step non-uniformity.
Not all tolerance steps feel equally acceptable. In a Soritical sequence where adjacent elements differ by one precise unit (one hair, one mm, one dollar), some steps feel like negligible differences and others feel like significant jumps — even though the precise difference is identical.
This is evidence that the vague difference between adjacent elements is not a simple function of the precise difference.
Source: @cite{fara-2000} on salient similarity, @cite{dinis-jacinto-2026} §6.1
- adjective : String
- preciseDifference : String
- clearRegionAcceptance : String
Steps near clear cases feel negligible
- thresholdRegionAcceptance : String
Steps near the threshold feel significant
- nonUniform : Bool
Same precise difference, different perceived significance
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Baldness tolerance: removing one hair from someone with 100,000 hairs feels negligible; removing one hair from someone near the "boundary" can feel significant.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Height tolerance: 1mm difference between two clearly tall people feels negligible; 1mm difference near the threshold feels more significant.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Borderline contradiction acceptance.
Experimental data on whether subjects accept sentences of the form "X is P and not P" (contradictions) and "X is neither P nor not P" (gaps) for borderline cases of vague predicates.
Key finding: acceptance rates for both contradictions and gaps are significantly higher for borderline cases than for clear cases. This is evidence against both classical logic (which rejects all contradictions) and standard supervaluationism (which makes contradictions super-false even for borderline cases).
The data is compatible with the TCS framework (@cite{cobreros-etal-2012}), which predicts that borderline cases tolerantly satisfy P ∧ ¬P.
Source: @cite{alxatib-pelletier-2011}, @cite{ripley-2011}, @cite{serchuk-hargreaves-zach-2011}
- study : String
Study identifier
- predicate : String
The vague predicate tested
- stimulusType : String
Stimulus type (visual, scenario-based, etc.)
- borderlineCase : String
Description of the borderline case
Acceptance rate for "X is P and not P" (contradiction)
Acceptance rate for "X is neither P nor not P" (gap)
- borderlinePeaks : Bool
Whether rates are significantly higher for borderline than clear cases
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
@cite{alxatib-pelletier-2011}: 5 men of different heights (visual). Man #2 (5'11", median size) shows peak contradiction/gap acceptance.
Equations
- One or more equations did not get rendered due to their size.
Instances For
@cite{ripley-2011}: 7 pairs (A-G) of decreasing nearness (visual). Pair C (median distance) shows peak contradiction acceptance.
Equations
- One or more equations did not get rendered due to their size.
Instances For
@cite{serchuk-hargreaves-zach-2011}: scenario-based (Susan's wealth). Forced-choice with 6 options including "Both", "Neither", "Partially True".
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
All three studies find borderline-peaking: contradiction/gap acceptance is significantly higher for borderline cases than for clear cases. This is the key empirical finding that TCS accounts for and standard supervaluationism does not.