People With Autism Spectrum Conditions Make More Consistent Decisions

People with autism spectrum conditions (ASC) show reduced sensitivity to contextual stimuli in many perceptual and cognitive tasks. We investigated whether this also applies to decision making by examining adult participants’ choices between pairs of consumer products that were presented with a third, less desirable “decoy” option. Participants’ preferences between the items in a given pair frequently switched when the third item in the set was changed, but this tendency was reduced among individuals with ASC, which indicated that their choices were more consistent and conventionally rational than those of control participants. A comparison of people who were drawn from the general population and who varied in their levels of autistic traits revealed a weaker version of the same effect. The reduced context sensitivity was not due to differences in noisy responding, and although the ASC group took longer to make their decisions, this did not account for the enhanced consistency of their choices. The results extend the characterization of autistic cognition as relatively context insensitive to a new domain, and have practical implications for socioeconomic behavior.


Supplementary Materials
. Means for ICAR, age, and AQ in the high and low AQ groups for the two versions of the AQ study.

Figure S1. ASC study: Primary regression analysis
The panels show the regression coefficients for the mixed-effects logistic regression analyses described in the main text. Here and for all other plots of regression coefficients, the error bars show 95% Wald confidence intervals and black data points indicate significant effects (p <.05, two-tailed). The panels show the coefficients for three contrasts: decoy-selection vs any other outcome, consistent choice vs preference reversal, and attraction-effect preference reversal vs non-attraction preference reversal. (BIC values: 1531.7, 3390.0, 637.9, respectively.)

Figure S2. ASC study: First choice proportions
The plot shows the proportions of times participants in the ASC and neurotypical (NT) control group chose the target, competitor, and decoy options on the first presentation of each product pair.

Figure S3. ASC study: Regression analysis of first choices
The panels show the results of analysing participants' responses on the first occurrence of each product pair. The top panel shows the coefficients from contrasting the tendency to choose the decoy (coded 1) with the tendency to choose one of the other options (target or competitor, both coded 1); the bottom panel plots the coefficients obtained when contrasting target choices (coded 1) against competitor choices (coded 0). (BIC values 1059.2 and 3774.6, respectively).

Figure S4. ASC study: Controlling for random responding
The panels show the results of re-running the primary analysis controlling for random responding, indexed by the participant's proportion of decoy selections across the 20 test trials (pdecoy). Note that it would not make sense to include pdecoy in the Decoy vs Non-decoy contrast, so only the Consistent Choice vs Preference Reversal and Attraction-Effect Preference Reversal vs Non-attraction Preference Reversal contrasts are analysed; BIC values: 3402.9 and 644.4, respectively.

Figure S5. ASC study: Comparing decision times
As described in the main text, for our analysis of decision-times we tested whether the ASC and Control groups differed in the mean response time by running a linear regression. The plot shows the regression coefficients for this analysis (adjusted R-sq =.089).

Figure S6. ASC study: Controlling for decision times
The panels show the results of re-running the primary analysis with each person's log-transformed mean response-time (logmt) as an additional predictor. (BIC values 1516.3, 3404.7, and 641.1 for top, middle, and bottom analyses, respectively.)

Figure S7. AQ study: Primary regression analysis
The panels show the regression coefficients for the three contrasts described in the main text. As reported in the Methods section, the analyses included the Version variable (Version 1 coded 0; Version 2 coded 1) and the interaction between Version and all other variables to examine the consistency of the findings across participant samples / stimulus sets. As for other variables, Version was standardized prior to each regression, and the interaction terms were computed by multiplying the standardized predictors (e.g., the coefficient labelled aq.int is z.aq*z.version). (BIC values for the top, middle, and bottom analyses are: 1655.2, 3959.4, and 703.7, respectively).

Figure S8. AQ study Version 1: Primary regression analysis
The panels show the regression coefficients for the three contrasts described in the main text, limited to the data from the first version of the study. (BIC values for the top, middle, and bottom analyses are: 815.8, 1974.5, and 349.8, respectively).

Figure S10. AQ study: First choice proportions
The plot shows the proportions of times participants in the low-and high-AQ groups chose the target, competitor, and decoy options on the first presentation of each product pair.

Figure S11. AQ study: Regression analysis of first choices
The panels show the results of analysing participants' responses on the first occurrence of each product pair. The top panel shows the coefficients from contrasting the tendency to choose the decoy (coded 1) with the tendency to choose one of the other options (target or competitor, both coded 1); the bottom panel plots the coefficients obtained when contrasting target choices (coded 1) against competitor choices (coded 0). (BIC values 1257.2 and 4651.2, respectively).

Figure S12. AQ study: Controlling for random responding
The panels show the results of re-running the primary analysis controlling for random responding, indexed by the participant's proportion of decoy selections across the 20 test trials (pdecoy). Note that it would not make sense to include pdecoy in the Decoy vs Non-decoy contrast, so only the Consistent Choice vs Preference Reversal and Attraction-Effect Preference Reversal vs Non-attraction Preference Reversal contrasts are analysed; BIC values: 3980.2 and 703.7, respectively.

Figure S13. AQ study: Comparing decision times
As described in the main text, for our analysis of decision-times we tested whether the low-AQ and high-AQ groups differed in the mean response time by running a linear regression. The plot shows the regression coefficients for this analysis (adjusted R-sq = .088).

Figure S14. AQ study: Controlling for decision times
The panels show the results of re-running the primary analysis with each person's log-transformed mean response-time (logmt) as an additional predictor. (BIC values 1670.8, 3983.2, and 717.2 for top, middle, and bottom analyses, respectively.)