Why People on LSD or Mushrooms Experience the Same Hallucinations

The perception of hallucinations by those under the influence of substances like LSD or psilocybin (mushrooms) has long intrigued researchers and scientists. One popular hypothesis centers around how the human brain processes visual information. According to this theory, the brain's visual processing is structured into about 7 layers of neurons, each responsible for a different level of processing and abstraction.

The Layers of Visual Processing

The lowest level of these layers receives input from the optic nerves after the left and right nerves cross and intermingle in the middle of the brain. This layer detects basic visual elements such as edges at specific angles and movement. Each subsequent layer increases the complexity of processing, until the highest layer is responsible for recognizing full objects and their relationships.

The Role of Lower-Level Information Disclosure

When considering hallucinogens, it appears that lower-level, unprocessed visual data "leaks" to the conscious level of the brain. Typically, the brain lacks mechanisms to interpret this raw data. Consequently, it may interpret this information in a way that results in hallucinations. Alternatively, the leakage might relate to noise from lower levels of visual processing that usually get suppressed. Noise here refers to the background random firings of neurons that don't have sensory meaning.

Neural Networks and Hallucinations

To further explore this hypothesis, researchers have utilized artificial neural networks (ANNs). These networks, designed to recognize objects, have been fine-tuned to detect objects even if they don't exist. By doing so, these networks sometimes generate images that closely resemble human hallucinations. This experiment provides strong evidence suggesting that drug-induced hallucinations may be linked to how the human brain processes and interprets visual data.

Confirmation from Artificial Neural Networks

Google researchers recently conducted an experiment with image recognition networks. They increased the sensitivity of these networks to detect objects that didn't exist, then adjusted the networks to generate images based on these erroneous detections. The results were striking, often producing images that closely mimicked human hallucinations. This study, titled 'This is What Happens When Deep Learning Neural Networks Hallucinate', can be found on The New Stack.

The Implications

The confirmation from artificial neural networks provides compelling evidence that drug-induced hallucinations are deeply connected with the intricate workings of human image processing. By understanding how these lower-level visual processing stages work, researchers may gain insights into the mechanisms behind visual hallucinations. This could have significant implications for fields such as neuroscience, psychiatry, and even artificial intelligence.