Fig 1.
An example of a simple and a complex reference game trial.
Table 1.
Annotation tags used to label participants’ reported reasoning strategies with examples.
Fig 2.
Average proportions of responses in the reference game per utterance type.
Error bars represent standard deviations.
Fig 3.
Proportion of target choices on simple and complex trials by participant.
Dot size indicates number of participants.
Table 2.
Regression model output for the model without individual differences (N = 254) and for the model with individual differenceseak (N = 167).
Effects for which the 95% CrI does not include 0 are bolded.
Table 3.
Regression model output from Franke & Degen (2016) (F&D; N = 51) and from Experiment 4 of Mayn & Demberg (2023) (M&D; N = 60).
Effects for which p<0.05 for F&D and effects for which the 95% CrI does not include 0 for M&D are bolded.
Fig 4.
Mean proportion of target selections per annotation tag and the frequency of that tag in the two critical conditions.
Error bars represent standard deviations.
Fig 5.
LPA classes vs. classes based on annotations.
Table 4.
Confusion matrix comparing classes based on LPA (rows) and based on annotation (columns).
Fig 6.
Correlations of the individual difference measures. Insignificant correlations are greyed out.
Table 5.
Summary statistics for the individual difference measures (N = 167).
Table 6.
PCA factor loadings.
Fig 7.
A boxplot showing the distribution of the three individual differences for each reasoning class identified by LPA.
The y-axis represents the composite score for the corresponding individual difference obtained using PCA. Ordinal regression revealed the effect of logical reasoning and ToM but not of memory on reasoning class.
Fig 8.
Proportion of correct responses of individual participants in the two implicature conditions as a function of logical reasoning and ToM.