Objective: Case report notions of unexpected memory retrieval in patients with severe dementia near to death are starting to alter the central "irreversible" paradigm of dementia and locate dementia as a problem of memory retrieval, not consolidation. We suggest that the most likely central tenet of this paradoxical memory retrieval is the fluctuation of neuromodulators projecting from the brain stem to the medial prefrontal cortex and the hippocampus. The neuromodulation-centric explanation of this phenomenon aims to open the "irreversible" paradigm of dementia up for discussion and suggest a plausible treatment strategy by questioning how the devastating process of death fluctuates memory performance in severe dementia. Background: Supporting demented patients, who are mostly unresponsive, without making demands or asking a question and regarding them as valuable human beings unexpectedly improve their memory performance around the time of death. New lucidity hypothesis: Around the time of death, neurological signs (hyper-arousal and-attention) of demented people point out that neurotransmitter discharges are dramatically changed. Relatively resistant neuromodulator circuits to neurodegeneration can maintain optimal levels of arousal and attention for memory processing. In this way, unexpected episodes of lucidity can be triggered. Also, corticotropin-releasing peptides might increase mental clarity by increasing the excitability of the neuromodulator circuits. The science of memory retrieval is more complicated and nuanced than retrieval observations in case reports, but the rapid development of new techniques holds promise for future understanding of lucidity in severe dementia. Major challenge for the model: There is no an animal or human model to test this hypothesis; however, the similarities between neurological signs (instantaneous cognitive fluctuations) of delirium and paradoxical lucidity could provide a unique window to understand neural events of terminal lucidity on a modified animal model of delirium. Likewise, similarities between unexpected consciousness signs of terminal lucidity and lucid dreaming suggest that lucid dreaming episodes might be considered a human model for terminal lucidity research.