Grid cells are believed to be key neurons in spatial navigation. Their mechanisms and operation have been described via the theoretical framework of continuous attractor networks (CANs). While increasing evidence points to a CAN mechanism for grid cells, one fundamental aspect of grid cell CAN models remains untested, namely the prediction that population activity is determined by specific patterns of recurrent connectivity. In particular, many CAN models require a Mexican hat connectivity profile with excitatory connections between functionally similar neurons and inhibition between more dissimilar cells.To experimentally test this prediction, we have developed an all-optical recording and stimulation approach allowing us to investigate functional connectivity between grid cells with known phase differences. Our four-step procedure consists of: i) using a miniature two-photon (2P) microscope (MINI2P) to identify grid cells in freely-moving mice, ii) transferring the mice to a benchtop 2P photostimulation system where the field-of-view is aligned to the field-of-view recorded with MINI2P, iii) photoactivating channelrhodopsin expressing grid cells identified in step i, and iv) measuring stimulation triggered responses in untargeted grid cells and relating these responses to the neurons’ functional similarity to the targeted cells. Our data show that we can record well over 100 grid cells from one module using MINI2P and that on average more than 75% of these neurons can be refound and photoactivated in our benchtop 2P system. We are currently applying this approach to directly test predictions of CAN models regarding connectivity between grid cells with known firing correlates.