Set of models fitted to assess the relationship between treatment group and trial number and (A) problem solving success, (B) latency to solve, and (C) persistence.

The error distributions used in the models are specified by model family. Model comparison was based on AICc. Np is the number of parameters, ΔAICc is the difference between the candidate model and the model having the lowest AICc, and ωi is the AICc weight of each model (i.e. measure the likelihood that a given model is the best among the candidate models). Models retained based on the lowest AICc value appears in bold, and models retained based on parsimony rules in italic font.