Thinking in Black-and-White: Adults with High Autistic Traits Sharpen Vague Predicates

Background: Thinking in autism spectrum disorders shows atypical cognitive flexibility and processing of local information independently of context. Recent evidence suggests that decision-making is also atypical, being more consistent and conventionally rational in adults with ASD and non-ASD controls with high autistic traits (Farmer et al., 2017). There does not appear to be any investigation of black-and-white, logical thinking that is reported clinically in ASD. Vagueness, a property of natural language, may provide a window on logical thinking. Vague predicates (e.g., “tall”) admit borderline cases; those where it is not clear whether the predicate applies. Thus, the truth of a statement like “X is tall” is indeterminant. This contrasts with sharp predicates where classical bivalent logic applies. Truth-gap theories are a suggested solution to this problem, which minimise the divergence from classical logic. Alxatib and Pelletier (2011) (AP) found evidence for gap theories in a truth-judgment task. They argued that if the predicate “tall” was sharp, rather than vague, participants would be equally likely to accept ‘“X is tall” is True’ as to reject ‘“X is not tall” is False’. However, they found participants were more likely to accept the statement ‘“X is tall” is False’ than reject ‘“X is not tall” is True’, which is consistent with the presence of a truth-gap.

Objectives: If black-and-white thinking is characteristic of ASD, the indeterminacy of vague predicates may be problematic. We investigated if individuals with high autistic traits may, therefore, treat vague concepts as sharp, implying a rejection of any divergence from the classical bivalent logical paradigm.

Methods: Participants recruited via social media participated in an online experiment, including the adult Autism Spectrum Quotient and AP’s truth-judgment task. They were presented with a drawing of a police line-up showing five men, ranging in height from 5’4” to 6’6”, with 5’11” as a borderline case. They responded "True", "False" or "Can't tell" to a series of statements (e.g., “#1 is tall”; “#1 is not tall”) about each man, presented in random order.

Results: Data from 121 participants (median age = 27 (range = 18-80) years; Male = 26, Female = 95) were analysed. A median split of AQ score (median = 12 (range = 2-33)) was used to create low (N = 63) and high (N = 58) autistic traits groups. Cochran's Q tests were run to compare the frequency of truth judgments to “#x is tall” and “#x is not tall” statements in the low AQ and high AQ groups. For the critical borderline case, the low AQ group were more likely to reject the statement ‘“#2 is tall” (i.e., judge it False) than accept the statement “#2 is not tall” (i.e., judge it True) (Q(1) = 8.33, p = .004). This was not the case for the high AQ group (Q(1) = 1.00 , p = .317 ).

Conclusions: Participants with high autistic traits, in contrast to those with low autistic traits, treated vague concepts as sharp, appearing to reject any divergence from the classical bivalent logic.