Abstract
Preclinical-clinical translation of cognitive functions has been difficult in Alzheimer's disease (AD) research but is crucial to the (predictive) validity of AD animal models. Reversal learning, a representation of flexibility and adaptability to a changing environment, might represent such a translatable feature of human cognition. We, therefore, examined visual discrimination (VD) and reversal learning in the APPPS1-21 mouse model of amyloid-based AD pathology. We used touchscreen operant cages in novel and translationally valid, as well as objective testing methodology that minimizes within- or between-trial handling. Mice were trained to associate a visual cue with a food reward (VD learning), and subsequently learned to adjust their response when this rule changed (reversal learning). We assessed performance at two different ages, namely at 6 months of age, considered an early disease stage, and at 9 months, a stage of established pathology. Both at 6 and 9 months, transgenic animals needed more sessions to reach criterion performance, compared to wild-type controls. Overall, transgenic animals do not show a general cognitive, motivational or motor deficit, but experience specific difficulties to adapt to reward contingency changes, already at an early pathology stage.