In a double-blind, placebo-controlled study at Washington State University (Cuttler & McLaughlin, Journal of Psychopharmacology, March 2026), 120 regular cannabis users were randomly assigned to vaporize placebo, 20 mg THC, or 40 mg THC, then completed 21 memory tests spanning verbal recall, visuospatial memory, prospective memory, source memory, false memory, episodic content, and temporal order.
Cannabis produced significant impairment in 15 of 21 measures. The expected deficits — worse verbal recall, impaired prospective memory (forgetting to do something later) — appeared as predicted. The unexpected finding was false memory: participants on THC were substantially more likely to “recall” words that were never presented. THC didn't just delete memories. It created them.
The dose-response curve was flat. Moderate doses (20 mg) produced impairment comparable to high doses (40 mg). The memory system didn't degrade proportionally with intoxication — it degraded categorically. Once THC crossed a threshold, most memory domains were equivalently affected regardless of dose.
The false memory result is the structurally interesting one. Memory distortion and memory loss are different failure modes. Losing a memory means the signal was never encoded or was erased. Fabricating a memory means the system actively generated content that passed its own reality check. The generation mechanism is intact; the verification mechanism is compromised. THC doesn't silence the memory system. It makes it generative without making it accurate.
This maps onto a broader principle about memory architecture: memory is not a recording. It's a reconstruction. Every act of recall is a synthesis — assembling fragments into a coherent narrative. What THC appears to do is loosen the constraints on that synthesis, allowing fragments to combine into narratives that were never experienced. The fabrication is the normal memory process operating with reduced fidelity, not a separate pathological mechanism.