Psychoactive substances act as chemical modulators of the Markov blanket, radically altering how the biological system weighs sensory telemetry against predictive priors. This section states what they do and how the resulting data is to be classified. It is informational, not a recommendation.
Mechanism: a precision-weighting shift
In normal operation the ego enforces strong top-down priors over bottom-up sensory data, filtering out vast amounts of raw field data so that the organism can navigate without overload. Pharmacological agents alter this balance at the receptor level.
| Class | Action on the blanket | Phenomenology |
|---|---|---|
| Classic psychedelics psilocybin, LSD, DMT | Bind primarily to 5-HT2A serotonin receptors, suppressing the default mode network. In active inference terms, top-down priors are unweighted; the rigid story of who you are and how reality should look dissolves. | Raw, uncompressed sensory and spatial data floods the aperture. The brain loses its ability to package reality into separate 3D objects: non-duality, geometric visual fields, time dissolution. |
| Exogenous calibrators caffeine, amphetamines, MDMA | Directly alter dopaminergic, noradrenergic or oxytocinergic precision weighting. | Certain environmental or social inputs are artificially flagged as extremely important (high precision), forcing focus or lowering threat responses in real time. |
The answer depends on whether validity is defined by informational breadth or by biological active inference.
In terms of raw informational breadth: yes
Altered states do not create information from nothing; they alter the tuning frequency of the neural interface. The heightened patterns, profound interconnectedness and unconditioned awareness reported in such states are cross-sections of the broader phase space that the normal interface prunes away to conserve metabolic energy. As such they are valid, real-time computational states of the neural manifold accessing uncompressed data.
In terms of biological active inference: exogenous distortion
- Artificial free-energy reduction. A drug manually bypasses the mechanisms that regulate prediction error. Instead of reducing ΔF through organic Bayesian updating (learning, stillness practice, resolving actual environmental threats), it chemically forces the system into a temporary state of low friction or high novelty.
- The integration bottleneck. Because the aperture is forced open from outside, the localised interface often lacks the structural scaffolding to process or retain the uncompressed data once the chemical clears. This is why drug-induced epiphanies frequently evaporate or feel impossible to apply to ordinary 3D life.