Studies on cat’s visual cortex (O’Herron et al 2016) have shown the existence of orientation and direction sensitivity in cerebral microvessels. In some cases, interestingly, orientation tuning of the vessels is quite different from the orientation tuning of its neural neighborhood. In order to account for this discrepancy, the authors of that study speculate that there could be propagation of dilatory signals from adjacent neural functional columns along vessel walls. We investigate this question using a computational model of neurovascular network and explore the right pattern of lateral interactions among the vessels that can account for the orientation tuning of the cerebral vessels and the neurons they perfuse. To this end, we use a biologically plausible self-organizing network known as Gain Controlled Adaptive Laterally connected network (GCAL) to model orientation and direction sensitivity in visual cortical neurons. This neural network is bidirectionally connected with a network of vascular units. Each unit of the vascular network integrates vasodilatory signals from the neural network through weighted connections, while each neuron integrates the flux of metabolic signals from the vascular network. The vessels are assumed to have circularly symmetric, center-surround type of lateral interactions. Three alternative forms of lateral interactions among the vessel units are considered. (1) No lateral connection (2) Excitatory center and inhibitory surround (3) Inhibitory center and excitatory surround. The afferent and lateral connections of neural and vascular layers are trained using Hebbian learning for the above three cases of lateral connectivity in the vessels. The performance of the neural and vascular layer is compared with the experimental observations (O’Herron et.al 2016) and the architecture of vascular network that best matches with the experimental observation is selected. The performance is evaluated in terms of orientation selectivity index (OSI) and directionality index (DI). It is observed that when the vessels have a lateral interaction comprising of inhibitory center and/or excitatory surround, the performance of the proposed neuro-vascular system closely matches the experimental observations. The model suggests that the vascular network has active signaling within itself in an OFF-center, ON-surround fashion. These model predictions are consistent with previously reviewed evidence for existence of dynamic signaling within microvascular beds (Secomb and Pries 2002; Pradhan and Chakravarthy 2011).