SciFi Futures
← Concepts
Neurotechnology

Dream recording and directed dreaming

Instrumentation that reads what a sleeper is dreaming and steers the content of the dream on command.

In shortDream recording and directed dreaming, from The Lathe of Heaven (Ursula K. Le Guin, 1971), is at technology readiness level 2 of 9 (technology concept formulated) at its weakest link. Held back by Mechanistic understanding of what sleep is for and Mechanistic theory of neural computation. Science status: consistent with known physics; key effects unproven at scale. Estimated new funding to unlock: $109B–550B (a floor; some parts are priced only to early research).

Dr Haber's Augmentor records and amplifies George Orr's dream state and, under hypnotic suggestion, directs what he dreams about.
The Lathe of Heaven by Ursula K. Le Guin, 1971Set in 2002Overdue24 years past the story's date; weakest link still at TRL 2

Closest todayDormio targeted dream incubation, two-way dialogue with lucid dreamers, and fMRI decoding of dream and imagery content.

Dependency stack

Research TRL 1–3Development TRL 4–6Deployment TRL 7–9
Sort by
  • Reading or specifying dream imagery means decoding internally generated experience
    ✕3 blockers›
    • Models are subject-specific; decoders do not transfer across brains without recalibration
    • Non-invasive signals are slow and spatially coarse; invasive coverage is a few square centimetres
    • Encoding (write) models are far behind decoding; we cannot specify the activity for an arbitrary percept
    TRL 3
    • Fast, fine-grained signals needed for anything beyond coarse category decoding
      ✕2 blockers›
      • Tissue response degrades signal over years
      • Channel count and wireless bandwidth
      TRL 6
      • Channel count is limited by wireless data and power budget
        ✕3 blockers›
        • Tissue heating caps implant power at a few hundred milliwatts, which bounds radio and compute
        • Raw broadband neural data exceeds the link, forcing on-implant spike detection and compression
        • Battery or wireless-power lifetime versus the decades a permanent interface implies
        TRL 5
        • Implant power density sets what telemetry and on-board compute are affordable
          ✕2 blockers›
          • Cycle life of lithium-metal anodes
          • Thermal runaway safety
          TRL 6
    • Deep generative and language models supply the latent spaces that brain activity is mapped onto
      ✕2 blockers›
      • Reliability and calibration over long horizons
      • Continual learning after deployment
      TRL 7
      • Training and inference compute
        ✕2 blockers›
        • Grid power and interconnection queues
        • Advanced-node chip and HBM memory supply
        TRL 8
  • Steering a dream requires detecting sleep stage and cueing in real time
    ✕3 blockers›
    • Cues seed a theme but do not determine narrative content
    • Decoding dream content requires per-subject models and MRI-grade signals, not wearables
    • Individual variability in sleep staging and arousal thresholds
    TRL 6
  • Intervening nightly for long periods requires knowing what sleep is doing that must not be disrupted
    • No consensus on sleep's core function, so no way to verify a shortened-sleep design is not accruing damage
    • No biomarker of sleep debt, so safety can only be assessed by lifetime outcome studies
    • Known short-sleeper families are tiny cohorts with no controlled long-term follow-up
    TRL 2
  • Directed dreaming works by reactivating chosen memory traces during sleep
    ✕4 blockers›
    • Human engrams cannot be labeled; the mouse toolkit depends on transgenic c-fos tagging
    • Memory is distributed across an engram complex in many regions, not one addressable site
    • Writing content requires knowing the ensemble code for arbitrary experience, which is unknown
    • Specificity; stimulation that strengthens one memory also perturbs neighbours
    plausibleTRL 3
    • Writing a memory requires knowing what code represents its content
      • Unknown sufficient level of biological detail
      plausibleTRL 2
      • Wiring diagrams constrain and test circuit models
        ✕3 blockers›
        • Human proofreading of AI segmentation dominates cost; 2025 methods imply tens of billions of dollars for one mouse brain
        • Imaging throughput and storage; a human brain at EM resolution is of order 100+ exabytes of raw data
        • Sectioning or expanding whole large brains without tissue loss or distortion
        TRL 3
        • Ultrastructure must survive preservation before it can be imaged
          ✕3 blockers›
          • Ischemic damage starts within minutes of circulatory arrest, before perfusion can begin
          • Aldehyde fixation may destroy molecular state (phosphorylation, receptor conformation) that could carry memory
          • No assay proves preserved tissue still contains the information an emulation would need
          TRL 4
        • Automated segmentation and tracing of exabyte-scale image volumes
          TRL 8
    • Activity-dependent labeling in humans would need brain-wide gene delivery
      ✕3 blockers›
      • Receptor targeting is species-specific; capsids binding human TfR1 often fail in macaque
      • Pre-existing anti-AAV immunity excludes a large fraction of patients and blocks redosing
      • Dose needed for brain-wide coverage drives liver toxicity and manufacturing cost
      TRL 4
      • Designed capsid and shuttle proteins are the main route to receptor-specific transcytosis
        ✕1 blocker›
        • Designing catalysis and conformational dynamics
        TRL 6
    • Engram ensembles must be addressed at single-cell scale in deep structures
      TRL 6
  • Influencing thalamic and limbic sleep circuits without implants
    ✕4 blockers›
    • Skull aberration and heating; each subject needs a CT-based acoustic model or custom metalens
    • Mechanism unclear (mechanosensitive channels vs thermal vs cavitation), so dose-response is empirical
    • Excitation versus inhibition depends on parameters in ways not yet predictable
    • Effects on a structure, not a circuit; cannot address specific cell types
    TRL 5

Who could unlock it

Mechanistic understanding of what sleep is forTRL 2bottleneck
  • Ying-Hui FuResearcher, University of California, San Francisco
    Identified the familial natural short sleep genes, five so far, and showed short-sleep alleles reduce dementia-linked aggregates in mouse models
  • Louis PtacekResearcher, University of California, San Francisco
    Co-leads the short-sleeper family cohort work with Fu, including the 2019 ADRB1 A187V discovery
  • Maiken NedergaardResearcher, University of Rochester and University of Copenhagen
    Showed norepinephrine-driven slow vasomotion pumps glymphatic clearance during sleep and that zolpidem suppresses it (Cell, January 2025)
  • Giulio TononiResearcher, University of Wisconsin-Madison
    With Chiara Cirelli, authored the synaptic homeostasis hypothesis, the main rival account of sleep's function, still being tested ultrastructurally in 2025
Mechanistic theory of neural computationTRL 2bottleneck
  • Philip ShiuResearcher, Eon Systems
    Built the leaky integrate-and-fire whole fly brain model that predicts sensorimotor circuit responses (Nature 2024)
  • Srinivas TuragaResearcher, HHMI Janelia Research Campus
    Co-led connectome-constrained networks that predict neural activity across the fly visual system (Nature 2024)
  • Larry AbbottResearcher, Columbia University
    Theoretical neuroscientist; co-director of the Zuckerman Institute's Center for Theoretical Neuroscience
  • Allen InstituteAcademic lab
    MICrONS functional connectomics and a 230,000-cell biophysical model of mouse visual cortex
Bidirectional encoding models of perception and imageryTRL 3
  • Tomoyasu HorikawaResearcher, NTT Communication Science Laboratories
    Mind captioning: generated text descriptions of viewed and recalled videos from fMRI (Science Advances 2025)
  • Alexander HuthResearcher, University of Texas at Austin
    Non-invasive semantic language decoding (2023) and its cross-participant transfer (Current Biology 2025)
  • Francis WillettResearcher, Stanford University
    Co-led real-time decoding of inner speech from motor cortex in BrainGate participants (Cell 2025)
  • Edward ChangResearcher, UCSF
    Speech neuroprosthesis work decoding attempted speech to text and voice from cortical surface recordings
Engram-level memory read and writeTRL 3
  • Susumu TonegawaResearcher, MIT Picower Institute
    His lab first optogenetically reactivated and implanted false memories in tagged engram cells in mice (2012-2013)
  • Sheena JosselynResearcher, SickKids and University of Toronto
    Showed engram allocation, erasure and artificial memory formation in mice; 2025 Cell paper on stress-inflated engrams
  • Steve RamirezResearcher, Boston University
    Engram manipulation to alleviate maladaptive memory states in mice
  • Robert HampsonResearcher, Wake Forest University School of Medicine
    Hippocampal memory-decoding-model stimulation facilitated content-specific recall in epilepsy patients (Front. Comput. Neurosci. 2024)
Sleep-stage-targeted sensing and interventionTRL 6
  • Ken PallerResearcher, Northwestern University
    Targeted memory reactivation in REM sleep steered dreams toward unsolved puzzles and doubled next-day solving (Neuroscience of Consciousness 2026)
  • Karen KonkolyResearcher, Northwestern University
    Led two-way communication with lucid dreamers (Current Biology 2021) and smartphone-based lucid dream induction (2024)
  • Adam Haar HorowitzResearcher, MIT Media Lab
    Created Dormio for targeted dream incubation at sleep onset (Scientific Reports 2023)
  • ElemindStartup
    MIT spin-out selling an EEG headband that times acoustic stimulation to brainwaves to speed sleep onset (2024)

Beyond engineering

  • Mental privacy of recorded dream content
  • Clinical oversight of therapy that alters memory while the patient is unconscious