Modeling the performance of children on the Attentional Network Test

Recent research in attention indicates it involves three anatomical networks concerned with alerting, orienting and executive control (cf. Posner & Fan, 2007). The Attentional Network Test (ANT) provides a behavioral measure of the efficiencies of these three networks within a single task (Fan, MaCandliss, Sommer, Raz & Posner, 2002). This work adapts an ACT-R 6.0 model of adult performance on ANT (Hussain & Wood, 2009) to model the performance of children (aged 6, 7, 8, 9 and 10) on a child-friendly version of the task (Rueda, Fan, McCandliss, Halparin, Gruber, Lercari, Posner, 2004). Modifications are carried out within the framework of the ACT-R cognitive architecture (Anderson, Bothell, Byrne, Douglass, Lebiere, & Qin, 2004; Anderson & Lebiere, 1998). Models simulating the child study results indicate that improvements in latency and error rate can be attributed to incremental improvements in processing time and reduction in errors of commission respectively. In contrast the models indicate a qualitative difference between children under 9 and older age groups in both alerting efficiency attributed to specific reductions in processing surprise stimuli in the younger age groups, and executive control efficiency between 6 year olds and older age groups attributed to a slower ability in 6 year olds to focus the target in incongruent stimuli. An inhibiting effect of the alerting network on congruency, not found in the child study, was found in the model data consistent with adult studies (Callejas, Lupianez & Tudela, 2004; Fan, Xiaosi, Kevin, Xun, Fossella, Wang, Posner, 2009). Investigation of model performance under invalid spatial cueing conditions compared to adult model performance (Hussain & Wood, 2009) finds the models are differentiated by a slower ability to disengage from invalidly cued locations in the child models but are similar in benefiting from the facilitating effects of cueing on processing congruent stimuli.