In this talk we introduce a novel theoretical framework that merges verbal reports with computational cognitive models. This approach takes advantage of the context-sensitive features of advanced Natural Language Processing (NLP) models to analyze participants’ verbal descriptions about their problem-solving strategies. We will use a JavaScript-based plugin that complements a JsPsych library that we recently developed. This tool enables
real-time speech recognition, where spoken reports are automatically converted into text and analysed using large-language models, greatly reducing the typical burden of analysing self-report data. Our goal is to use these verbal descriptions to validate the formal components in our models, while also exploring their more implicit assumptions.