Discussion of nunuphar's comment: In this extensive anatomical study Susan Haber et al trace in detail the reciprocal connections between the different striatal regions and the midbrain dopamine neurons that provide input to them. They first make an admirable effort to functionally divide the striatal regions (in primate) into limbic (VMS), associative (CS), and sensorimotor (DLS) by mapping the regions by the inputs from limbic corticies/amygdala/hippocampus, dlPFC, and motor/premotor regions respectively. They then did many anterograde, retrograde, and bidirectional tracer injections in various striatum and the midbrain sites. Among their many findings, the one that warranted great attention and a theoretical framework about network function was that for each successive striatal region moving in the ventromedial-dorsolateral direction their inputs to the midbrain had both overlap with where their own DA inputs originated and extended past the origin region in the ventral direction (moving inversely to the striatal axis) to exert influence on the midbrain neurons that would be sending input to the “next step up” in the striatum. They conjecture that the part of the inputs going back to the origin DA cells have direct synapses with the DA cells to provide feedback inhibition while the part of the inputs going to the adjacent midbrain area synapse with inhibitory interneurons thus disinhibiting the next set of DA cells. They dub this the “ascending spiral” that allows striatal regions to have influence on the regions just dorsolateral to them by modulating their dopamine inputs.