Revisão de treinos automatizados baseados na RFT e alterações metodológicas no treino SMART

Abstract

SMART is an automated training protocol based on relational frame theory (RFT) that allows participants to develop derived relational responding (DRR). Besides SMART there are other RFT-based training protocols that aims the development of DRR. Two studies were carried out to explore theoretical and empirical aspects of methodological alternatives to these training protocols. The first study consisted of a systematic review on RFT-based training protocols, including SMART, aiming to compare methodological variables and participants' performance on SMART and those other automated RFT-based training protocols that develop DRR. The articles were selected based on two techniques, the search for keywords and the use of snowballing. After the selection step, the 27 experimental studies were compared according to the sample; performance and mastery criteria in combinatorial entailment (CE) tests; use of psychometric tests; training protocol, structure and duration; stimuli used and relational responses; and use of multiple- exemplar training. The results of the review suggested a heterogeneity among the articles found, which made the comparison between them difficult. A series of independent variables that can be manipulated in future experiments were highlighted. The second study consisted of the application of a computer game based on SMART, with four stages. Forty children between 9 and 12 years old participated, and were splitted into four experimental groups according to the insertion of multiple sets of stimuli in stage two, stage three, in both or none of them. Participants' scores in the four stages and the probability of success in the different types of tasks in the first stage were analyzed according to the nodal distance, type of relationship (sameness or opposition) and direction of the questions present in the tasks. The results showed that there were no significant differences between groups regarding the use of multiple sets of stimuli. Stage three proved to be easier than the others in all groups. The types of tasks showed significant differences based on the probability of success according to the nodal distance and the relation of the question, but the linearity of the question did not impact the chances of getting it right. The results support the hypothesis that the SMART training order is not efficient in terms of gradual increase in difficulty. This aspect can impact the performance of some participants in training and can be critical if we consider the use of SMART for other populations, such as children with atypical development. The findings of the two studies, the review article and the empirical article, offer suggestions for researchers interested in studies based on automated training based on RFT, highlighting the SMART protocol.