To re-examine the paradoxical effect hypothesis of abused drugs, the present study concerned whether different doses of morphine disparately affect neuronal activity and associations among the subareas of the medial prefrontal cortex (mPFC: cingulate cortex 1-Cg1, prelimbic cortex-PrL, infralimbic cortex-IL), the subregions of the nucleus accumbens (NAc; both core and shell), and the basolateral amygdala (BLA) following conditioned taste aversion (CTA) and conditioned place preference (CPP). All rats were given a 0.1% saccharin solution for 15-min, and they were intraperitoneally injected with saline or 20, 30, or 40 mg/kg morphine to form the aversive CTA learning. Later, half of the rats were tested for CPP (including the CTA and then CPP tests) for 30-min. Finally, the immunohistochemical staining with c-Fos was conducted after the behavioral test. After the CTA test, c-Fos (%) in the Cg1 and PrL (but not the IL) was more in 20-40 mg/kg of the morphine groups; c-Fos (%) in the NAc core, NAc shell, and BLA was more in the 30-40 mg/kg morphine group. After the CPP test, the Cg1, PrL, IL, and BLA showed more c-Fos (%) in 20 mg/kg morphine; the NAc core showed fewer in c-Fos (%) in the 30-40 mg/kg morphine groups. The mPFC subregions (e.g., Cg1, PrL, and IL), NAc subareas (e.g., NAc core and NAc shell), and BLA were involved in the different doses of morphine injections. The correlation analysis showed that a positive correlation was observed between PrL and IL with NAc core with low doses of morphine and with NAc shell with increasing doses of morphine after the CTA test. After the CPP, an association between PrL and NAc core and NAc shell at low doses and between IL and BLA and NAc shell with increasing doses of morphine. Therefore, different neural substrates and the neural connectivity are observed following different doses of morphine and after the CTA and CPP tests. The present data extend the paradoxical effect hypothesis of abused drugs.