Carga cognitiva em um problema de otimização: o efeito da complexidade e do formato de apresentação das informações no processamento da informação e na tomada de decisão

Abstract

Humans face complexity in many actions, from choosing a pizza to decide a university career. From the computational perspective agents face complexity because they deal with limited resources that restrict their capabilities. Thus, is necessary an adjustment between the environment and the processing structures used to solve the problem for the task load does not exceed the capabilities of the decision maker. This work investigates the relationship between the complexity and the information presentation format of an optimization problem, and the cognitive load imposed on the solver, considering the influence of these variables on information processing and decision making. An experimental task was developed based on the Knapsack optimization problem. The complexity (low, medium and high) and the information presentation format (numerical and visual) was manipulated. Data were collected from 42 individuals. Information processing was measured based on the rate of fixations and duration of fixations, and decision-making analyzed through optimization and relative performances, response time and confidence. The main findings allow us to conclude that even with the increased load, in some comparisons subjects were able to keep constant the attention employed (rate of fixations), or the evoked effort (duration of fixations) was not affected by the load. The visual task imposed an extra load processing when the instances were of high complexity. Subject’s ability to optimize and the time to find a solution are highly susceptible to changes in the load. This finding was only observed in the numerical version for the ability to find approximate solutions and for confidence, which may be the result of compensations performed in the visual version. The presentation format affected the optimization capacity, and in low complexity the numerical format led to better scores, which became equivalent when the complexity was medium or high. The response time was lower in the visual version. All these results allow the conclusion that the combination of complexity levels and information presentation formats adopted in the optimization problem favored different aspects of the task. The presentation of relevant information for solving the problem through numbers will result in less workload imposed by the task, allowing the extraction of accurate data and objective information. Specifically, when the complexity is low, numbers will lead to decision-making efficiency in terms of finding the optimal answer. On the other hand, when dealing with situations that require a quick decision process the visual presentation can bring approximate results as good as the numerical one, however, with a lower response time.