Human Attentional Networks: A Connectionist Model Hongbin Wang 1 and Jin Fan 2 Abstract & Recent evidence in cognitive neuroscience has suggested that attention is a complex organ system subserved by at least three attentional networks in the brain, for alerting, orienting, and executive control functions. However, how these different networks work together to give rise to the seemingly unitary mental faculty of attention remains unclear. We describe a con- nectionist model of human attentional networks to explore the possible interplays among the networks from a computational perspective. This model is developed in the framework of leabra (local, error-driven, and associative, biologically realistic algorithm) and simultaneously involves these attentional net- works connected in a biologically inspired way. We evaluate the model by simulating the empirical data collected on normal hu- man subjects using the Attentional Network Test (ANT). The simulation results fit the experimental data well. In addition, we show that the same model, with a single parameter change that affects executive control, is able to simulate the empirical data col- lected from patients with schizophrenia. This model represents a plausible connectionist explanation for the functional structure and interaction of human attentional networks. & INTRODUCTION Although the concept of attention is well understood by many, the mechanism by which this function operates remains unclear. Recent advances in cognitive neurosci- ence have prompted a renewed interest in treating at- tention as a neural organ system and have bolstered interest in seeking its neural underpinnings (see Posner & Fan, in press; Posner, 2004). One such organ theoretic account advocates that there exist multiple attentional networks in the brain, with each responsible for a dif- ferent aspect of attention (e.g., Raz & Buhle, 2006; Posner & Dehaene, 2000; Posner & Raichle, 1994; Posner & Petersen, 1990). At least three different attentional net- works, for alerting, orienting, and executive control, have been distinguished (see Fan, Raz, & Posner, 2003, for a review). Specifically, alerting refers to the function of achieving and maintaining a heightened internal state of arousal in preparation for coming task-related events. The alerting network has been associated with thalamic, frontal, and parietal areas. An example of the alerting network pathology is hemispatial neglect syndrome, a neuropsychological impairment following unilateral su- perior parietal damage in which patients neglect objects and events in the contralateral hemispace and appear to be blind, deaf, and numb on this side of space. Orienting refers to selectively focusing on one or a few items out of many candidate inputs. The orienting network includes parts of the superior and inferior parietal lobe, frontal eye fields, the subcortical collicular pathway (the superior colliculus of the midbrain and the pulvinar), and reticular nucleus of the thalamus. A particular important structure in this network is the posterior parietal cortex, a part of the dorsal ‘‘where’’ pathway (Ungerleider & Mishkin, 1982), which is thought to host multiple supramodal spatial representations in egocentric frames and to guide movements ( Wang, Johnson, Sun, & Zhang, 2005; Colby & Goldberg, 1999; Egeth & Yantis, 1997). Finally, execu- tive control refers to monitoring and resolving conflicts in planning, decision making, error detection, and over- coming habitual actions. The executive control network includes the midline frontal areas (anterior cingulate cortex [ACC]), lateral prefrontal cortex, and the basal ganglia. These regions are the target areas of the ventral tegmental dopamine system (Montague, Hyman, & Cohen, 2004; Holroyd & Coles, 2002; Braver & Cohen, 1999). The role of executive control is evidential in a range of Stroop- related tasks (see MacLeod & MacDonald, 2000; MacLeod, 1991). The attentional networks account of human attention has been systematically examined in recent years using a variety of approaches (e.g., see Raz & Buhle, 2006). Ge- netically, Fan, Wu, Fossella, and Posner (2001) explored the heritability of the networks by comparing attentional performance between monozygotic and dizygotic twins and found that different networks possessed different degrees of heritability. The underlying genetic variation that might contribute to the brain activation related to executive control has also been investigated (Fan, Fossella, Sommer, Wu, & Posner, 2003). Developmentally, evidence has suggested that different attentional networks possess different developmental profiles. Using children aged from 6 to 10 years old, one study has shown that whereas the 1 University of Texas Health Science Center at Houston, 2 Mount Sinai School of Medicine D 2007 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 19:10, pp. 1678–1689