Many animals, including mice, will spend time 'investigating' a novel conspecific — for example, by sniffing or grooming them. Previous studies have shown that experimental, nonspecific activation of the prelimbic cortex (PL) reduces the amount of time that mice spend investigating another mouse; however, nonspecific inhibition of the PL does not affect social investigation. Thus, the role of the PL in this behaviour is not clear. Witten and colleagues now show in mice that PL projections to the nucleus accumbens (NAc) encode a combination of social and spatial information and may promote social investigation by enabling social–spatial learning.

The authors optogenetically activated PL projections to the NAc, the ventral tegmental area (VTA) or the amygdala in mice in a linear three-chamber arena, in which one end chamber contained an encaged 'target' mouse and the other contained a novel object. Activation of the PL→NAc projections, but not PL→VTA or PL→amygdala projections, specifically reduced their preference for inspecting the caged mouse, whereas inhibition of PL→NAc projections had no effect. PL→NAc projections may therefore mediate the reduction in social investigation resulting from nonspecific PL activation.

Credit: jvphoto/Alamy Stock Photo

almost 11% of the imaged neurons responded only when the imaged mouse was near to the social target in a specific chamber

To investigate what kind of information the PL→NAc projections encode, the authors used calcium imaging to assess the activity of these neurons while mice explored the three-chamber arena. Some PL→NAc neurons were more active while the imaged mouse was close to the target mouse. To distinguish neurons that were spatially selective from those that were selective for the spatial location of the social target, the authors analysed calcium responses in the imaged neurons when the target mouse was in the left chamber, and again when the social target was in the right chamber. This approach revealed that some neurons were selective solely for spatial position, whereas other neurons responded only when the imaged mouse was near the social target. Strikingly, almost 11% of the imaged neurons responded only when the imaged mouse was near to the social target in a specific chamber. This observation implies that some PL→NAc neurons convey a combined social and spatial code.

The authors hypothesized that the PL→NAc neuronal code might enable the learning of associations between certain locations and the target mouse. They let a test mouse explore two adjoined chambers that each contained an encaged target mouse for 1 day. Over the following 2 days, the authors inhibited or activated the PL→NAc axons of the test mice when they were near the target mouse in just one of the chambers. The following day, test mice showed lower or greater preference for the zones near the target mice that had been paired with PL→NAc inhibition or activation, respectively. Thus, the activity of PL→NAc neurons bidirectionally regulates social–spatial learning.

Together, these results demonstrate that some PL→NAc neurons encode a combination of social and spatial information, and that this encoding may enable animals to learn the locations of social encounters.