vaccine experimental inflammation model together with computational modelling of an instrumental learning task performed during fMRI (1). Twenty-two healthy participants were each tested 3-h after blindly administered typhoid vaccine (0.025 mg) and placebo (0.5 ml 0.9% NaCl) injection. Typhoid but not placebo was associated with a threefold increase in plasma IL-6 (t(21) = 5.20, p < 0.001) confirming induction of mild inflammation. Following inflammation participants demonstrated greater avoidance of a punishing cue (loss £1) and significantly reduced selection of a rewarded cue F(21,1) = 6.41, p = 0.019. Computational modelling of individuals' choices using a Q-learning model (free parameters alpha (learningrate), beta (choice-randomness) and R (reward value)) demonstrated significant reduction in reward cue value and increase in value of punishment cues following inflammation F(21,1) = 3.63, p < 0.05. fMRI confirmed previous correlations of appetitive and aversive prediction error (PE) within ventral striatal (VS) and anterior insula (AI) regions of interest respectively (FWE p < 0.05) across conditions. However, inflammation significantly impaired VS responses to appetitive PE and enhanced AI responses to aversive PE (FWE p < 0.05) suggesting that inflammation modulates the value of aversive and appetitive cues via actions on VS and AI prediction error signals respectively. (1) Pessiglione et al., 2006. Nature 442, 1042-45. http://dx.Despite decades of progress, the biological mechanisms underlying addiction are unknown. Glia, including astrocytes and microglia, profoundly influence neuronal function and may play a critical role in the development of drug addiction. Previously, we have shown that a neonatal handling procedure (which promotes enriched maternal care) reduces glial activation to an acute morphine challenge and prevents reinstatement of morphine conditioned place preference; likewise, similar results were obtained by pharmacological manipulation of glia within the nucleus accumbens (NAc). Here, we investigated the effects of opioid self-administration on central immune signaling within the NAc and its modulation by early-life experience. Male rat pups were handled (15 min/day) from postnatal day (P) 2 to P20 or left undisturbed. In adulthood, all rats were tested for self-administration of intravenous remifentanil (a short-acting opioid) in daily 1 h sessions for 2 weeks. Brains were then collected to examine persistent alterations in glial activation. Neonatal handling reduced acquisition of remifentanil; this effect was specific for opioids, as acquisition of food reward was unaffected. Remifentanil self-administration substantially increased glial antigen expression (Iba1 and GFAP) and proinflammatory gene expression in NAc, including the innate immune receptor TLR4. However, only handled rats showed increased expression of the anti-inflammatory cytokine IL-10. These data establish neural-glial interactions in the NAc as an essential component of self-motivated drug acquisition that can be alte...