Fig 1.
Three classic transmission biases and a new one.
(a) Direct or content-based bias favours the adoption of variants depending on their perceived attractiveness, utility, ease etc. (b) Model-based bias favours variants depending on who produced (or modeled) those variants. (c) Frequency-based bias disproportionately favours variants that have high (or low) frequency. (d) Context-congruence or associative bias favours variants that are associated with the current context, i.e. that were learned, observed or produced in the same context).
Fig 2.
Fig 3.
Experimental design showing what a participant P learns and transmits onwards.
Fig 4.
Experimental results showing the number of times the expert’s and peer’s variant strategy was produced in each onward transmission context.
Table 1.
Description of each step of the simulation.
Table 2.
Calculation of the probability that the congruent and incongruent variants are produced in each experimental condition as a function of ExpertBias and CongruentBias.
Fig 5.
The CongruentBias x ExpertBias parameter space showing the number of times (out of 5000) that the simulation results matched the experimental results for each parameter value combination.
Lighter colour represents more matches, indicating the parameter values that best fit the experimental results. Matches cluster in an area of around positive values of both CongruentBias and ExpertBias, indicating biases in favour of the congruent variant and the Expert’s variant.
Fig 6.
CongruentBias and ExpertBias parameter spaces for different values of PrimacyBias, showing the number of times (out of 5000) that the simulation results (approximately) matched the experimental results, for each parameter value combination.
Lighter colour represents more matches, indicating the parameter values that best fit the experimental results. Only positive values of CongruenceBias and ExpertBias are shown, as there were no matches in negative values (i.e., values favouring the incongruent and peer’s variants, respectively).