Neural circuits and emotional processing in rapid eye movement sleep

For decades, the scientific consensus held that the brainstem was the sole engine of REM sleep. Classical studies showed that while brainstem lesions abolished REM, cortical lesions actually preserved sleep cycles. This suggested the cortex was merely a passenger. However, a growing body of evidence now points to a much more complex, distributed network. This shift changes how we understand the link between sleep and mental health.

The shifting role of the cortex

The traditional view positioned the brainstem as both necessary and sufficient for REM generation. This meant the cortex was seen as secondary to the sleep-wake switch. But recent research suggests the cortex plays a vital, complementary role. New studies show that cortical regions help regulate the transition into REM and manage sleep homeostasis (the biological drive to recover lost sleep).

Current research faces a hurdle in moving from correlation to causation. We often observe specific neurons firing during REM, but proving they drive the state is difficult. Furthermore, clinical data remains inconsistent. Patients with Major Depressive Disorder (MDD) show varied REM profiles. Some experience shortened latency (the time taken to enter REM), while others show prolonged duration. This variation makes it difficult to create a single, universal model for REM-driven emotion.

A distributed network of command and control

The authors of this review describe a distributed architecture for REM regulation. Instead of one command center, REM emerges from three interacting systems:

1. The Cortical Layer: The cortex acts as a bidirectional regulator. In mice, the occipital cortex helps manage REM homeostasis. The retrosplenial cortex (RSC) serves as a starting point for cortical calcium activity waves. These waves correlate with shifts between quiescent REM (qREMS, with theta oscillations of 6.5–7.5 Hz) and active REM (aREMS, with faster oscillations of 8.2–10 Hz).
2. The Brainstem Engine: This remains a key site for REM generation. Nuclei like the laterodorsal tegmental (LDT) and pedunculopontine tegmental (PPT) act as "REM-ON" centers. They use cholinergic (acetylcholine-releasing) neurons to trigger the state. Meanwhile, the sublaterodorsal nucleus (SLD) helps produce muscle atonia (temporary muscle paralysis).
3. The Hypothalamic Modulator: The lateral hypothalamus (LH) provides stability. It uses melanin-concentrating hormone (MCH) neurons to inhibit "REM-off" signals and orexin neurons to maintain REM homeostasis.

As shown in, this creates a feedback loop. Cortical signals can descend to the brainstem, and brainstem signals can ascend to the cortex.

Mapping the architecture of affect

The importance of these circuits lies in their link to emotion. The review categorizes how these pathways intersect with three primary domains:

• Anxiety: Disruptions in the SLD and the lateral habenula (LHb) can increase anxiety-like behaviors in rodents. In mice, REM sleep deprivation (REMSD) has been linked to neuroinflammatory responses (inflammation in the nervous system) in the prefrontal cortex (PFC).
• Depression: The connection involves neuroplasticity (the brain's ability to change). The paper reports that REMSD can decrease levels of brain-derived neurotrophic factor (BDNF)—a protein vital for neuron growth. Clinical findings vary, however. Some MDD patients show shortened REM latency, while others show extended latency.
• Fear and PTSD: REM is critical for fear memory extinction (learning that a stimulus is no longer a threat). The authors report that activity in the infralimbic cortex during REM facilitates this process. Failure in this mechanism may contribute to the fear consolidation seen in PTSD.

As shown in, these domains share overlapping brain regions like the hippocampus and mPFC. This suggests a common biological foundation for many neuropsychiatric conditions.

Limits of the current paradigm

Several limitations remain. First, there is a "methodological confound." Most studies rely on REM sleep deprivation (REMSD). However, depriving an animal of sleep is a major physiological stressor. It is hard to tell if anxiety results from lost REM function or the stress of sleep loss.

Second, the "translation gap" is significant. Much causal evidence comes from rodent models using optogenetics (using light to control specific neurons). While brain structures are similar, human REM involves complex dreaming and emotional nuances that mice do not mirror. Finally, the specific functional role of theta oscillations in regulating emotion remains an open question. We see the rhythmic pattern, but the exact mechanism is not yet decoded.

The verdict: A blueprint for neuromodulation

The field is moving toward treating psychiatric disorders through sleep manipulation. We have shifted from viewing REM as a simple reflex to seeing it as a cortical-integrated process. Identifying specific nodes, such as the RSC or the SLD, provides targets for future therapy.

Instead of systemic drugs, the next step may involve circuit-specific neuromodulation (targeting specific neural pathways). This could include deep brain stimulation or methods to stabilize theta rhythms. However, researchers must first decouple the stress of sleep deprivation from the biological effects of REM loss. Only then can these models reliably guide clinical treatment.

Figures from the paper