Modelos mentales erróneos y persistentes en programación

Abstract

Learning involves creating mental representations analogous to what is learned, known as mental models. These models allow us to make predictions and decisions. The closer they are to reality, the better our predictions and decisions will be, indicating that effective learning consists of refining these models. Programming, on the other hand, requires carrying out the reverse process: from the desired effects, create a program that produces them. This process, more complex than merely refining a mental model, demands multiple accurate models and associative skills to navigate between them in both directions. Teachers instruct on conceptual models, which are clearly defined abstractions. Acquiring these models involves generalizing and extracting patterns from numerous specific examples. Consequently, initial mental models often link to circumstantial details of the examples, both to the noise and the signal. Although these models may be erroneous, they can work reasonably well in some specific situations, be sufficient to pass, and persist over time. This could explain future difficulties in programming. This study seeks to identify incorrect mental models in programming. We propose an iterative method that starts from hypotheses based on teaching experience and reaches empirical evidence. This approach also allows discovering new models and refining our knowledge. We present its application with fourth-year students and first-year concepts. Our results highlight potential erroneous models and suggest other unforeseen ones that require further investigation.