Activation of the mesolimbic dopamine system is positively reinforcing. After repeated activation, some individuals develop compulsive reward-seeking behavior, which is a core symptom of addiction. However, the underlying neural mechanism remains elusive.We trained mice in a seek-take chain, rewarded by optogenetic dopamine neuron self-stimulation. After compulsivity was evaluated, AMPA/NMDA ratio was measured at three distinct corticostriatal pathways confirmed by retrograde labeling and anterograde synaptic connectivity. Fiber photometry method and chemogenetics were used to parse the contribution of orbitofrontal cortex afferents onto the dorsal striatum (DS) during the behavioral task. We established a causal link between DS activity and compulsivity using optogenetic inhibition.Mice that persevered when seeking was punished exhibited an increased AMPA/NMDA ratio selectively at orbitofrontal cortex to DS synapses. In addition, an activity peak of spiny projection neurons in the DS at the moment of signaled reward availability was detected. Chemogenetic inhibition of orbitofrontal cortex neurons curbed the activity peak and reduced punished reward seeking, as did optogenetic hyperpolarization of spiny projection neurons time-locked to the cue predicting reward availability.Our results suggest that compulsive individuals display stronger neuronal activity in the DS during the cue predicting reward availability even when at the risk of punishment, nurturing further compulsive reward seeking.