Scientists in Japan say they have found a puzzling mismatch between how much fuel the brain sends itself during REM sleep — the stage most associated with dreaming — and how much usable energy its neurons actually have on hand. In a study of mice published in Communications Biology on July 27 and highlighted this week by ScienceDaily, researchers at Tohoku University led by Yusuke Takahashi and professor Ko Matsui report that blood volume in the brain begins climbing roughly 50 seconds before REM sleep starts, beginning in the posterior cortex — but the neurons' immediately available energy, measured as ATP, drops even as that fuel supply increases.
To watch the process unfold, the team used a UV-curable resin to keep the skulls of living mice transparent, then tracked brain activity during natural sleep with wide-field fluorescence imaging. Alongside blood volume, they measured neuronal ATP, the molecule that directly powers cell activity, and astrocytic pyruvate, a compound that links glucose arriving from the bloodstream to the brain's energy-processing machinery. Pyruvate levels rose during REM sleep as expected, confirming more fuel was reaching the tissue — which made the simultaneous ATP decline more surprising.
A challenge to a simple assumption
The finding complicates a common assumption in neuroscience: that pumping more blood, and therefore more glucose, into brain tissue should straightforwardly raise its available energy. Instead, the Tohoku team's data suggest REM sleep places such heavy demands on neural circuits — the intense, rapid signaling thought to underlie dreaming — that neurons burn through energy faster than the new fuel supply can replenish it, at least in the moments captured by this technique.
Sleep may appear peaceful, but the brain is highly active — especially when dreaming.
Tohoku University research team, Communications Biology
The researchers also found that fluctuations in theta-band brain activity during non-REM sleep could predict the subsequent rise in blood volume, hinting at a signaling pathway that prepares the brain's fuel delivery system in advance of a dream state.
Because the work was conducted in mice using an invasive skull-transparency technique, it cannot yet confirm that the same energy paradox occurs in the human brain, and the published research does not report how many animals were used in the underlying experiments — a detail that would help gauge how robust the effect is. The authors frame the results as a step toward understanding the brain's metabolic efficiency during sleep broadly, with implications the team says could extend to conditions involving disrupted sleep and energy metabolism. Further work, including replication in other species and eventually non-invasive human studies, would be needed before the findings reshape clinical understanding of sleep disorders.
