Abstract
Lateral prefrontal and intraparietal cortices have strong connectional and functional associations but it is unclear how their common visuomotor, perceptual and working memory functions arise. The hierarchical scheme of cortical processing assumes that prefrontal cortex issues ‘feedback’ projections to parietal cortex. However, the architectonic heterogeneity of these cortices raises the question of whether distinct areas have laminar‐specific interconnections underlying their complex functional relationship. Using quantitative procedures, we showed that laminar‐specific connections between distinct prefrontal (areas 46 and 8) and lateral intraparietal (LIPv, LIPd and 7a) areas in Macaca mulatta, studied with neural tracers, varied systematically according to rules determined by the laminar architecture of the linked areas. We found that axons from areas 46 and rostral 8 terminated heavily in layers I–III of all intraparietal areas, as did caudal area 8 to area LIPv, suggesting ‘feedback’ communication. However, contrary to previous assumptions, axons from caudal area 8 terminated mostly in layers IV–V of LIPd and 7a, suggesting ‘feedforward’ communication. These laminar patterns of connections were highly correlated with consistent differences in neuronal density between linked areas. When neuronal density in a prefrontal origin was lower than in the intraparietal destination, most terminations were found in layer I with a concomitant decrease in layer IV. The opposite occurred when the prefrontal origin had a higher neuronal density than the target. These findings indicate that the neuronal density of linked areas can reliably predict their laminar connections and may form the basis of understanding the functional complexity of prefrontal–intraparietal interactions in cognition.