Modern Neuroanatomy Favors Freud and Jung on Dreams. A 2025 Meta-Review Explains Why.

Do dreams mean anything, or is your brain just firing randomly while you sleep?

For three decades, the official neuroscience answer was: random. The dominant theory, proposed by Harvard psychiatrist J. Allan Hobson in 1977, held that dreams originate in the brainstem during REM sleep. According to this model, the pons fires off neural signals with no psychological content, and the cortex does its best to stitch that noise into something that resembles a narrative. The storylines, the emotions, the recurring characters and symbols of your dream life were considered accidents of neural architecture. Meaningless.

The psychoanalytic tradition said the opposite. Freud argued that dreams are generated by the unconscious mind to express repressed desires and unresolved emotional conflicts. Jung expanded on this, proposing in his framework for Jungian dream analysis that dreams serve a compensatory function: they surface what the waking ego has neglected, repressed, or refused to see. Under this model, dreams are not noise. They are messages from a part of you that knows what you are ignoring.

For most of modern neuroscience, Freud and Jung were treated as historical curiosities. Hobson's model was the textbook answer.

Then brain imaging technology matured. And when researchers finally looked at which brain areas actually activate during dreaming, what they found did not match the textbook at all.

In 2025, a team of researchers at Medical School Hamburg published the first systematic meta-review of neuroimaging studies on dreams, using the same methodology (PRISMA guidelines) that governs clinical medical research. They evaluated 22 review articles covering fMRI and PET data on dreaming. They scored each article against both models: the psychoanalytic model and Hobson's Activation-Synthesis Model. The results were not subtle.

Thirteen out of 22 reviews aligned with the psychoanalytic model. Zero aligned with the Activation-Synthesis Model. Five showed overlap with both. Four aligned with neither. Not a single review article, out of 22, clearly supported the theory that dominated neuroscience for 30 years.

The score was 13 to zero. And the brain areas explain why.

What Each Model Predicts

Before looking at the imaging data, it helps to understand what each model claims about the brain during dreaming. They make different predictions about which regions should be most active.

The Activation-Synthesis Model predicts that dreams originate in the brainstem, specifically the pontine tegmentum. During REM sleep, this region generates bursts of neural activity that propagate upward into the cortex. The cortex then "synthesizes" this random input into a dream experience. Under this model, the key brain areas should be the brainstem (as the generator) and the sensory and motor cortices (as the synthesizers). The emotional content of dreams is incidental, a byproduct of the cortex doing pattern-matching on random signals.

The psychoanalytic model predicts something different. If dreams express unconscious emotional content, then the brain areas most active during dreaming should be the ones involved in emotion, memory, and self-referential processing. The limbic system (amygdala and hippocampus) should be hyperactivated. The prefrontal cortex, which handles logical reasoning and executive control, should be suppressed, because conscious rational oversight is exactly what the psychoanalytic model says goes offline during dreaming. The ego, in Jung's language, steps aside so the unconscious can speak.

These predictions are testable. The imaging data tests them.

The Limbic System: Emotion and Memory at Full Volume

The single most consistent finding across the neuroimaging literature on dreams is limbic system hyperactivation. Of the 22 articles reviewed, 11 explicitly reported that the amygdala, hippocampus, or limbic system broadly shows elevated activity during dreaming. Five articles specifically addressed amygdala activation. Six more addressed the limbic system as a whole.

The amygdala is the brain's emotional processing center. It handles threat detection, fear conditioning, anxiety, and the assignment of emotional significance to experience. The hippocampus sits next to it and serves as the brain's memory intermediary, binding new experiences to existing memory networks and determining what gets consolidated into long-term storage.

When both of these regions are hyperactivated simultaneously during dreaming, the result is exactly what dreamers report: intensely emotional experiences built from fragments of memory, with emotional significance that often exceeds what the dreamer would feel during waking life. This is why your dreams can be terrifying, euphoric, sexually charged, or profoundly sad in ways that seem disproportionate to their content. The emotional volume is turned up because the amygdala is running hot, and the memories being pulled into the dream are selected by emotional relevance, not logical relevance, because the hippocampus is working in concert with the amygdala rather than the prefrontal cortex.

This pattern maps directly onto what Freud described. Dreams are driven by emotional forces, specifically repressed desires and unresolved conflicts, that operate outside conscious awareness. The amygdala-hippocampus complex during dreaming is the neuroanatomical correlate of what Freud called the unconscious wish. The brain is not generating random noise. It is doing emotional work.

Jung would add that this emotional work is purposeful. It compensates for whatever the waking ego has failed to integrate. The imaging data does not prove that claim on its own, but it is entirely consistent with it: the emotional and memory systems are not passively replaying the day. They are actively selecting and combining material in ways that the conscious mind did not choose and cannot control.

The Prefrontal Cortex: The Ego Goes Dark

The second most consistent finding in the neuroimaging literature involves the prefrontal cortex, and it tells the other half of the story.

The dorsolateral prefrontal cortex (DLPFC) consistently deactivates during dreaming. This region is the seat of executive function. It handles working memory, logical reasoning, planning, self-monitoring, reality testing, and volitional control. In the language of depth psychology, the DLPFC is the neural substrate of the ego. It is the part of your brain that decides what to pay attention to, evaluates whether your experience makes sense, and maintains the boundary between what is real and what is imagined.

When the DLPFC goes offline during dreaming, you lose all of those capacities. You cannot think logically. You cannot detect contradictions. You cannot recognize that you are dreaming. You accept impossible events as real. This is why you can fly, have conversations with dead relatives, or find yourself in a building that is simultaneously your childhood home and your workplace, and none of it strikes you as strange.

But here is the part that matters for the psychoanalytic model: while the DLPFC deactivates, the medial prefrontal cortex (MPFC) and ventromedial prefrontal cortex (VMPFC) remain active or even increase their activity. These regions handle self-referential processing, emotional evaluation, and the integration of emotional information with autobiographical memory. The VMPFC is also part of the dopaminergic circuit that Solms (2000) identified as critical to dream generation: patients with ventromedial frontal damage lose the ability to dream entirely, and this region receives projections from the mesolimbic dopamine pathway that Panksepp (1998) characterized as the SEEKING system.

So the brain during dreaming is not simply "shut down" at the top. It is selectively shut down. The logical, reality-testing, ego functions go offline. The emotional, self-referential, meaning-making functions stay online or intensify. The brain is actively creating conditions for emotional material to be processed without the interference of rational oversight.

This is the neuroanatomical version of what Jung described when he said the unconscious speaks when the ego steps aside. The ego (DLPFC) literally steps aside. The unconscious (limbic system + MPFC/VMPFC) literally takes over.

The Parietal-Temporal-Occipital Junction: Where Meaning Gets Made

Mark Solms' lesion studies from the 1990s and 2000s identified a brain region that may be the most important of all for understanding why dreams are meaningful rather than random: the parietal-temporal-occipital junction.

Solms found that patients with damage to this area completely lost the ability to dream. Not reduced dreaming. Not altered dreaming. Complete cessation. Their REM sleep continued normally. Their brainstems were intact. But they could not dream at all.

This region sits at the intersection of three major processing areas: the parietal lobe (spatial awareness and body schema), the temporal lobe (language and auditory processing), and the occipital lobe (visual processing). Its primary function is transforming concrete sensory percepts into abstract representations. It takes raw experience and converts it into meaning.

This is exactly the kind of process that would be required for symbolic dream content. When you dream of teeth falling out, your brain is not randomly activating dental imagery. It is converting an abstract psychological state (loss of power, vulnerability, shame) into a concrete sensory image. That transformation, from abstract meaning to symbolic image, is what the parietal-temporal-occipital junction does. And when it is destroyed, dreaming stops entirely, even though the brainstem (which the Activation-Synthesis Model says generates dreams) is still functioning normally.

Only one of the 22 reviewed articles addressed this region in detail. This represents a significant gap in the literature, but the lesion data from Solms is among the most powerful evidence against the brainstem-origin theory and in favor of the view that dreams are generated by higher-order meaning-making systems.

The Brainstem and Sensory Cortices: Where the Activation-Synthesis Model Needed Support and Did Not Get It

If the Activation-Synthesis Model were correct, the brainstem and sensory/motor cortices should be the most frequently reported and most consistently activated regions in the dreaming brain. They are the foundation of the model: the pons generates the signal, the cortex synthesizes it.

Of the 22 articles reviewed, only four addressed brainstem activity during dreaming. Only two addressed sensory and motor cortex activation.

This is not a case of mixed results favoring one side over the other. This is a case of the predicted activation pattern for one model barely appearing in the literature at all, while the predicted activation pattern for the other model dominates it.

The brainstem is active during REM sleep. That is not disputed. But being active during REM is not the same as generating dream content. The distinction Solms made in 2000, that REM sleep and dreaming are controlled by different brain mechanisms, has been reinforced by the imaging data. The brainstem manages the physiological state of REM. The limbic system, prefrontal cortex, and parietal-temporal-occipital junction manage the psychological experience of dreaming. The Activation-Synthesis Model conflated the two.

Motor cortex inhibition during REM is real and well-documented. This is why your body does not act out your dreams (and why REM sleep behavior disorder, where this inhibition fails, results in patients physically acting out dream content). Freud described dreams themselves as "the guardian of sleep", arguing that dreams discharge unconscious impulses in imagery rather than action, preventing them from waking the dreamer. The discovery that the motor system is actively inhibited during REM is consistent with this framework: the brain blocks motor output so that dream content can be experienced without physical enactment.

What About Nightmares and Lucid Dreams?

The meta-review found limited but interesting data on two special dream states.

Nightmares show similar activation patterns to normal dreams, with the limbic system and brainstem both engaged, but at higher intensity. The amygdala in particular appears to be more strongly activated during nightmares, which is consistent with the emotional-processing model: nightmares are dreams where the emotional content has exceeded the system's capacity for integration, producing a fear response strong enough to wake the dreamer.

Lucid dreams, where the dreamer becomes aware they are dreaming, show increased prefrontal activation, particularly in regions associated with self-awareness and metacognition. This makes neuroanatomical sense: lucid dreaming is the partial reactivation of the executive functions that normally go offline during sleep. The ego partially "wakes up" within the dream. Few studies have examined this phenomenon with imaging, and the differentiation from normal dream activation patterns remains unclear.

The Methodology That Makes This Different

Most articles making claims about neuroscience and dreaming are opinion pieces, literature reviews with no systematic method, or pop-science summaries of individual studies. The Borawski et al. (2025) meta-review is different in several ways that matter.

First, it followed PRISMA guidelines. This is the same reporting framework required for systematic reviews in clinical medicine. It means the search strategy, inclusion criteria, and analysis methods are transparent and reproducible. Another researcher could follow the same steps and arrive at the same pool of articles.

Second, the study was preregistered on the Open Science Framework before data collection began. This means the researchers committed to their methodology before seeing the results. They could not adjust their classification system after the fact to produce a preferred outcome.

Third, they developed a novel classification system: a dimensional scale from psychoanalytic to Activation-Synthesis alignment. Each of the 22 articles was scored on this scale based on which brain areas it reported and how those areas mapped onto the predictions of each model.

Fourth, they conducted a Risk of Bias assessment on all 22 reviewed articles. The results were sobering: only 1 of 22 was preregistered. Only 3 of 22 had independent reviewers. Only 2 reflected on potential source bias. Some articles were authored by the founders of the models being evaluated, creating obvious conflicts of interest. The meta-review itself met a higher methodological standard than the literature it was reviewing.

This matters because the finding is not just "the psychoanalytic model won." It is "the psychoanalytic model won even in a literature base that was methodologically weak and not optimized to produce that result." If anything, the true score may be even more lopsided than 13 to zero, because higher-quality studies with better controls might produce cleaner separations.

What This Means for Freud, Jung, and Your Dreams

The brain during dreaming is not a random signal generator. It is an emotional processing system with the rational governor turned off.

The amygdala and hippocampus (emotion and memory) are the most consistently hyperactivated regions. The dorsolateral prefrontal cortex (logic, reality testing, executive control) consistently deactivates. The medial and ventromedial prefrontal cortex (self-reflection, emotional evaluation) stay online. The parietal-temporal-occipital junction (meaning-making, symbolic transformation) appears to be critical for dreaming to occur at all. The brainstem, which the dominant neuroscience theory said was the origin point of dreams, barely registers in the systematic review of imaging data.

Freud said dreams are the royal road to the unconscious. Jung said dreams compensate for what the waking ego refuses to acknowledge. The neuroanatomy says the emotional brain takes over when the rational brain steps down, and it uses that window to process material that waking consciousness could not or would not handle.

These are not identical claims. But they are compatible claims, and the imaging data is far more consistent with all of them than it is with the idea that your dreams are meaningless noise.

The question was settled before most people realized it was still being asked. Dreams mean something. The brain is telling you that. The 30-year detour through "dreams are garbage" is over, and the data brought us back to where Freud and Jung started.

The only question left is whether you are paying attention.