A man lies on a hospital bed in Tunisia in 1943. He's a soldier, mute, so shattered by combat he can't remember his own name. A doctor injects a barbiturate into his arm and then, quietly, tells him: you are back in the Kairouan Pass, and mortar shells are dropping around you. At the word "shells," the man shudders. He gets out of bed. He starts crying out a name — "Steve! Steve, are you all right?" — and lurches around the room, cowering from shells only he can hear, crouching, trembling. He describes a foxhole, two men in it, the first sergeant dead with his head blown open, the other man alive with his chest torn open, crying. God, I can still hear him crying. And then, suddenly, the soldier covers his eyes, buries his head on the doctor's shoulder — and smiles. "I remember my name," he says. "God, what a miracle that I can talk."1
That is abreaction, and it is one of the most genuinely moving things in this whole dark book — a real healing, a broken man restored to himself. It's also the exact technique that, pointed a few degrees differently, becomes the truth-drug fantasy that would waste decades and lives. The same needle that healed that soldier is the one interrogators wanted for their prisoners, and the story of why it works for the one and fails for the other is the story of a single devastating clause: even if the memory is not true.
Start with the discovery that made all this possible, because it's counterintuitive. In 1930, a University of Wisconsin doctor named William Bleckwenn gave a sedating barbiturate — amytal — to catatonic psychiatric patients. Catatonia is a state of frozen unresponsiveness: patients mute, stuporous, bodies locked in strange poses. You'd expect a sedative to make such patients worse, deeper into their stupor. Instead, paradoxically, the frozen patient "came alive."2
For a lucid interval, catatonic patients who hadn't spoken in ages became rational, aware, conversational — discussing football scores, their families, taking nourishment. Dimsdale calls the reversal of catatonia with an intravenous barbiturate "about as close to a miracle as one finds in psychiatry." It was as if the catatonia were a kind of freezing in fear, and the sedative thawed it.
That thawing effect — a drug loosening a mind locked shut — is the seed of everything downstream. If a barbiturate could unfreeze a catatonic patient into speech, maybe it could unlock other kinds of locked minds: the amnesiac, the traumatized, and — the fatal extension — the prisoner guarding a secret.
The next application was fugue: the poetic condition where a person, overwhelmed by their life, flees it entirely, losing their memory and sometimes wandering for years under a new identity, unable to say who they are.3 One early clinician compared these patients to "small children who have run away from their parents and are unable to give information about themselves."
Barbiturates, it turned out, could bring the lost self back. Within minutes of an amytal injection, memories and feelings could flood back into a fugue patient. But Dimsdale flags a crucial subtlety here, and it's the hinge of the whole chapter: the drug doesn't work alone. There's a hazy boundary between hypnosis and amytal, and in both, suggestion plays a starring role. The doctor typically tells the patient — "suggests" — that the drug will help them relax enough to restore the lost connections. So it's never the drug reincarnating the memory by pure chemistry. It's the drug plus the doctor's suggestion, working together, on a suggestible mind.
Hold that, because it's the flaw hiding inside the miracle. A technique that works through suggestion is a technique that can put things in as easily as it draws things out.
The barbiturates reached their therapeutic peak in World War II, treating combat trauma — shell shock, battle fatigue, what we now call PTSD. Under names like narcoanalysis, narcosynthesis, and narcosuggestion, doctors used the drugs to help soldiers retrieve and relive repressed traumatic memories. Roy Grinker and John Spiegel documented this on the battlegrounds of Tunisia, and their work was rushed into print by the Macy Foundation and distributed to educate doctors on treating combat fatigue.4
The technique was cathartic. Reliving the buried horror, giving voice to it, releasing the frozen emotion — that abreaction genuinely helped shattered soldiers recover. The mute infantryman who couldn't remember his name got it back and could speak again. This is real.
But here is the clause Dimsdale plants like a landmine, and it detonates the entire truth-drug enterprise: there was something about expressing these intense memories that was cathartic, even if the memories were not true.5
Read that again. The healing didn't depend on the memory being accurate. A soldier could relive, weep over, and be released from a traumatic scene that never happened that way — and be healed all the same. For the therapeutic purpose, this doesn't matter at all. If reliving a partly-fabricated horror lets a broken man function again, the fabrication is harmless; the catharsis is the point, not the accuracy.
But now point the same needle at a prisoner. The interrogator doesn't want catharsis — he wants truth, accurate facts he can act on. And the very same technique that heals regardless of accuracy will produce, under interrogation, an emotionally vivid, deeply felt, utterly convincing account that may be completely false. The suggestibility that lets the doctor guide a soldier back to a healing memory lets the interrogator (deliberately or not) guide a prisoner into a fabricated confession he'll deliver with total conviction. William Sargant, who used these drugs on WWII troops, made exactly this discovery: the drugs worked, but many of the elicited memories were false.
The therapeutic use and the coercive use are the same technique. What separates them is that the therapist doesn't care whether it's true and the interrogator desperately needs it to be — and the technique cannot tell them apart. It delivers vivid, felt, believed memory. Whether that memory is real is a question the drug has no opinion about.
You have a mute, shattered soldier who can't remember his name, and you're going to heal him. Watch closely, because you're also demonstrating the exact tool an interrogator will covet.
You inject the barbiturate and let it loosen the frozen mind — the same thawing that unlocks a catatonic patient. Then you suggest. You don't wait passively for memory to surface; you guide it. You tell him he's back in the Kairouan Pass, that the shells are falling. You give the locked mind a doorway and an invitation.
And it works. He relives the foxhole, the dead sergeant, the crying man, the terror — and in the reliving, the frozen emotion releases. He weeps, he collapses, and then he surfaces with his name and his voice restored. You've performed a genuine healing.
Now notice the two things you actually did. First, the drug thawed the mind. Second, your suggestion shaped what surfaced. For your healing purpose, it doesn't matter whether the foxhole scene was accurate — the catharsis heals regardless. But that's precisely the vulnerability: what tells you this technique is dangerous in other hands is that you have no way to know, and no need to know, whether what surfaced was true. You guided a suggestible mind to a vivid, felt, believed memory. An interrogator with the same syringe and a darker agenda guides his prisoner the same way — and gets a confession delivered with the same total conviction, and just as untethered from fact.
The evidence is strong clinical history: Bleckwenn's 1930 amytal-and-catatonia paper, the documented fugue recoveries, the Lindemann observations (his patient's "the little guardian just isn't there… I know what I am saying and yet I don't know"), the Grinker-Spiegel Tunisia monograph, and Sargant's finding of false elicited memories. The therapeutic efficacy and the false-memory problem are both well-documented, not inferred.
The tension is the richest in the chapter, and the book leaves it productively open: if abreaction heals "even if the memories are not true," then therapeutic memory work rests on a foundation that has nothing to do with historical accuracy — which is fine for healing but devastating for anything that needs the truth (interrogation, testimony, forensic memory). The unresolved question ripples outward: how much of recovered-memory therapy generally — not just under drugs — shares this structure, producing felt, believed, cathartic memories whose accuracy is unknowable and sometimes fabricated? Dimsdale doesn't chase that all the way, but the clause "even if not true" is a depth charge under a great deal of twentieth-century psychiatry, and the book is honest enough to plant it and let it sit.
This page is the psychiatric door into the truth-drug chapter, paired with the obstetric door in The Drug That Made Mothers Forget Giving Birth. Obstetrics gave the interrogators scopolamine (erase memory); psychiatry gave them the barbiturates (retrieve memory). Both were mercies repurposed as weapons, and this page supplies the deeper of the two lessons — that memory retrieval is inseparable from suggestion, which is what dooms it as a truth-extraction method.
It converges tightly with Robert House and the Birth of Truth Serum and with William Sargant, the Recurring Witness — Sargant being the man who worked this exact terrain and discovered the false-memory problem firsthand. Where House over-read scopolamine's power from a single anecdote, the abreaction work supplies the mechanism of why he was wrong: the drugs produce suggestible, believed, vivid memory, not verified truth.
To behavioral-mechanics — Psychic Driving: The Cameron Failure. Cameron's psychic driving and wartime abreaction are the same underlying insight — that a drugged, loosened mind is open to having content moved in or out — pointed at opposite goals. Abreaction draws (fabricated-but-cathartic) memory out to heal; psychic driving pushes (repetitive taped) messages in to remake. The insight the pairing produces: both work on the suggestible loosened state the barbiturate creates, and both hit the same wall — the state accepts input readily but doesn't reliably retain or verify it. Abreaction's "even if not true" and Cameron's failure to implant anything durable are the same limitation seen from two sides: you can vividly load a loosened mind, but you cannot make the loaded content true or permanent by chemical means. The therapeutic version got away with it because healing doesn't require truth or permanence; the coercive versions failed because control does.
To behavioral-mechanics — Memory Manipulation and False Memory Implantation. The "even if not true" clause is the entire foundation of false-memory implantation, discovered here in a healing context before it was ever weaponized. Abreaction proved that a suggestible mind will generate a vivid, emotionally complete, sincerely believed memory in response to a doctor's suggestion — and that the person cannot tell the difference between that and a real one. The connection reveals that false-memory implantation isn't a separate dark technique that had to be invented; it's just abreaction with the therapeutic intent removed. The therapist who guides a soldier to a healing (possibly fabricated) foxhole memory and the interrogator who guides a prisoner to a fabricated confession are running the identical procedure. The only thing that made one a mercy and the other a violation was what the guide wanted and whether the accuracy mattered — the drugged mind's willingness to believe what it's led to was constant.
Sharpest implication: The needle that healed a mute, shattered soldier — restoring his name, his voice, his self through cathartic reliving — is the identical needle interrogators wanted for prisoners, and the reason it heals is the reason it can't extract truth: abreaction works even if the memory is not true. The technique produces vivid, felt, sincerely believed memory shaped by the guide's suggestion, and it has no capacity to distinguish the real from the fabricated. For a therapist who needs only catharsis, that's harmless. For an interrogator who needs facts, it's fatal — and the century spent decades learning that the same tool cannot serve both masters.
Generative questions: