The advantages of Structural Equation Modelling to address the complexity of spatial reference learning

• Autor(a) principal: Moreira, Pedro Miguel Silva
• Data de Publicação: 2016
• Outros Autores: Sotiropoulos, I., Silva, Joana Margarida Gonçalves Mota, Takashima, Akihiko, Sousa, Nuno, Almeida, Hugo Leite, Costa, Patrício Soares
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/1822/45006
• topic: Auto-regressive latent trajectories; Reference learning; Longitudinal assessments; Ciências Médicas::Medicina Básica; Science & Technology

Background: Cognitive performance is a complex process influenced by multiple factors. Cognitive assessment in experimental animals is often based on longitudinal datasets analyzed using uni- and multi-variate analyses, that do not account for the temporal dimension of cognitive performance and also do not adequately quantify the relative contribution of individual factors onto the overall behavioral outcome. To circumvent these limitations, we applied an Autoregressive Latent Trajectory (ALT) to analyze the Morris water maze (MWM) test in a complex experimental design involving four factors: stress, age, sex, and genotype. Outcomes were compared with a traditional Mixed-Design Factorial ANOVA (MDF ANOVA). Results: In both the MDF ANOVA and ALT models, sex, and stress had a significant effect on learning throughout the 9 days. However, on the ALT approach, the effects of sex were restricted to the learning growth. Unlike the MDF ANOVA, the ALT model revealed the influence of single factors at each specific learning stage and quantified the cross interactions among them. In addition, ALT allows us to consider the influence of baseline performance, a critical and unsolved problem that frequently yields inaccurate interpretations in the classical ANOVA model. Discussion: Our findings suggest the beneficial use of ALT models in the analysis of complex longitudinal datasets offering a better biological interpretation of the interrelationship of the factors that may influence cognitive performance.