Dream Research
Dreaming without REM: the finding that complicated everything
For decades REM sleep and dreaming were treated as the same thing. Patients whose brain damage abolished one while leaving the other intact showed that they are not.
Aserinsky and Kleitman's discovery of REM sleep in the 1950s came with an immediate and compelling finding: participants woken from REM reported dreams around 80 per cent of the time, and participants woken from non-REM sleep reported them rarely.
The obvious conclusion was that REM is the physiological state of dreaming. For roughly forty years this was the working assumption, and it shaped the field — the activation-synthesis model located dream generation in the brainstem structures that produce REM.
The assumption turned out to be wrong, and the person who did most to demonstrate it was Mark Solms.
The neuropsychological evidence
Solms studied patients with focal brain lesions, asking systematically about dreaming, and published his findings in 1997.
Two categories of case were decisive.
Patients who lost dreaming but kept REM. Damage to two regions abolished dreaming entirely while REM sleep continued normally on polysomnography. One was the parieto-temporo-occipital junction — an area involved in spatial and symbolic processing. The other was the ventromesial frontal white matter, which carries dopaminergic fibres from the midbrain to the forebrain.
The second finding was striking for an additional reason. Prefrontal leucotomy — the surgical lesion performed in large numbers in the mid-twentieth century — targeted approximately this pathway, and there are historical reports of patients ceasing to dream after the operation.
Patients who lost REM but kept dreaming. Damage to the pontine brainstem structures that generate REM abolished REM while patients continued to report dreams.
If REM generated dreaming, neither of these should be possible.
The ventromesial frontal pathway is dopaminergic. Consistent with a dopaminergic basis for dreaming: L-DOPA and dopamine agonists given to patients with Parkinson's disease increase the frequency and vividness of dreams, sometimes dramatically, without increasing REM sleep. Dopamine-blocking antipsychotics reduce dream reports. REM-suppressing antidepressants, by contrast, do not abolish dreaming.
Solms's reinterpretation
His proposal was that REM sleep and dreaming are doubly dissociable — controlled by different mechanisms that normally operate together. REM is generated by cholinergic brainstem systems. Dreaming is generated by dopaminergic forebrain systems involved in motivation and appetitive behaviour.
REM would then be an unusually reliable trigger for dreaming rather than its cause, which explains the high correlation without the identity.
He drew a further conclusion that made the finding controversial: if dreaming is driven by the brain's motivational systems, this is closer to Freud's account of dreams as wish-driven than the activation-synthesis model allowed. That claim has been contested vigorously and is separable from the neuropsychological data, which is what matters here.
Non-REM dreaming
The other half of the evidence came from better questioning.
Early studies asked participants woken from non-REM sleep whether they had been dreaming. Later studies asked what had been going through their minds, and found mentation reported in a substantial proportion of non-REM awakenings — commonly 40 to 70 per cent depending on method, stage and time of night.
Non-REM mentation is typically shorter, less vivid, less narrative and more thought-like than REM dreaming. But dreams indistinguishable from REM dreams do occur in non-REM sleep, particularly late in the night and at sleep onset, where hypnagogic imagery can be extremely vivid.
The current picture
More recent work has refined rather than overturned this. Francesca Siclari and colleagues used high-density EEG to compare periods of dreaming with periods of dreamless sleep within the same sleep stage, which is the crucial control.
They identified a posterior cortical hot zone — parieto-occipital regions — whose activity distinguished dreaming from non-dreaming regardless of stage. Reduced low-frequency and increased high-frequency activity in this region predicted dream reports, and could do so with reasonable accuracy in real time.
That result is consistent with Solms's emphasis on posterior cortical regions and largely independent of the theoretical dispute.
Why it matters
Practically, it means that any statement of the form "REM sleep is when you dream" is an approximation, and that research using REM as a proxy for dreaming has been measuring an overlapping but different thing.
It also reframes the question of what dreaming is for. If dreaming can be abolished by a forebrain lesion while REM continues, then whatever functions REM serves — and the memory consolidation evidence is strong — those functions do not depend on the subjective experience of dreaming.
Which raises the uncomfortable possibility that dreaming is a by-product: a genuine experience produced by machinery that is doing something else entirely.
