RESUMEN
INTRODUCTION: Semantic verbal fluency is a useful neuropsychological tool since it involves language and executive abilities that can be impaired in patients with neurodegenerative diseases in comparison to healthy controls. The present study explores retrieve and executive control processes using traditional quantitative and qualitative raw scores and examines the utility of multidimensional scaling combined with linear regression to provide new insights about the underlying semantic network in mild cognitive impairment and in healthy older adults. METHOD: A total of 165 Spanish older adults, 81 patients and 84 controls, were assessed in different cognitive domains and evoked animal names in one minute. Group differences on fluency raw scores were first explored. Regressions using tests to predict groups' fluency scores were also performed. The 12 animals that had been produced more frequently were selected to perform a multidimensional scaling analysis for each group. Four features related to animal names were extracted from normative studies and then were used as predictors in linear regression to provide an interpretation of the resulting dimensions' coordinates. RESULTS: Patients performed worse on memory and naming and produced a shorter list of animals than controls. In controls, naming and visual memory explained a small part of variance related to the total of animals produced and to the number of switches. Both groups exhibited similar semantic maps. CONCLUSIONS: Results suggest that patients' map is influenced by words with a dense associative neighborhood that were acquired at an early age, whereas in controls none of the predictors explained dimensions.
