Observing My Cannabinoid-Induced Vision from a Bayesian Brain Perspective
Note: These observations are deliberately written in bullet form. They are mostly a way for me to
document my experience, but I am also hoping that researchers far more knowledgeable than I will
contribute their ideas, and, if they are interested, directly assist with our project, especially in terms of
data analysis. If you would be interested in helping us with data analysis, and want to get a sense of our
data before contacting us, you can access it here:
I can be reached at email@example.com or at 617-694-0230. My collaborator, Jonathan Cannon, is a
computational neuroscientist and a post-doc at Brandeis University.
Brief Background (though I’d recommend skimming my paper, “Mending the Mirror” first):
•I am blind, with limited light perception, due to Leber’s Congenital Amaurosis, a congenital retinal
•Under the influence of cannabinoids, I have a subjectively visual experience, some elements of
which are compatible with the objects and features in my environment (i.e. I have experienced
some instances of object recognition).
•My collaborator has stated that he has seen evidence that my eye movements seem to be
interacting with visual stimuli while under the influence of cannabinoids, and some evidence that
they do the same when I am not under the influence. We are in the process of using eye-tracking
data to substantiate these claims.
While not under the influence of cannabinoids, I am not conscious of having any sort of visual
experience, with the exception of drastic changes in lighting (i.e. a camera flash)..
Our ultimate goal is to develop a piece of technology that will enhance visual information-
gathering skills, which would theoretically allow me and other people with severe retinal
diseases to experience some vision while not under the influence of cannabinoids.
A predictive processing hypothesis:
•My paper “Mending The Mirror” postulated that mirror neurons were responsible for my ability to
make sense of my visual experiences. This was too specific; it is more accurate to say that
predictive processing mechanisms are responsible.
•Under this hypothesis, I usually tune out the visual signal, as it has been proven to be unreliable.
However, marijuana causes alterations in the precision weighting of the visual error signal, so that
it becomes more salient. I make better use of the visual error signal, which leads to updated
predictions that match the error signal. These predictions constitute my cannabinoid-induced
visual experience (L. Swanson, personal communication, March 22, 2016).
•This is compatible with research suggesting that cannabinoids regulate the use of priors during
perception (Emrich, Lueke, & Schneider, 1997).
• Since I don’t have a fovea (though I might have a pseudofovia), and because my visual system is
underdeveloped and immature, it seems like my eyes are alternating “drifts” with saccades that
are informed by information gathered during the drifts. In other words, my eyes wander, then pick
up on something salient, and then I saccade to gain more useful information. We will be testing
this hypothesis by examining the eye-tracking data.
•It seems as if I am much more likely to recognize unattended vs. attended visual stimuli, which
means that if an experimenter asks me to look at something and gather information about it, I
disengage, because I am much more interested in exploring whatever random visual stimuli
strikes my fancy.
The role of proprioception:
•During my cannabinoid-induced visual experiences, the gain on the proprioceptive signal is also
increased, so that I feel subjectively more aware of my eye movements.
•According to Clark (2015), proprioception functions as motor commands, as motor commands
can be thought of as predicted proprioceptive states.
•Also, according to Clark, perceptual learning can occur in a largely unsupervised context because
of the use of probabilistic generative models.
•The increased proprioceptive awareness I experience under marijuana seems to have a variety of
1. It helps me overcome some of the limits of my nystagmus (uncontrolled eye movements) by
giving me information about where my eyes are in space.
2. It helps me learn how to saccade effectively. Some eye movements feel inherently rewarding,
like I’ve correctly executed a position in yoga. Cannabinoids make my eyes feel “weighted,”
and the feeling of reward seems to be related to me shifting this weight in various ways.
Whether this reward stems from the fact that I am gathering information effectively (based on
prior information gathered from “drifts”) or from learning how to saccade in and of itself, I do
3. Having increased awareness of my eye muscles gives me a strong sense of horizontal and
vertical space, which allows me to construct mental images. These images can then become
part of the generative model and can be compared with actual visual experiences. The
cannabinoid induced proprioceptive awareness can linger for a few days, and during that
time, I often find myself imagining myself seeing. I try to picture certain objects, or to imagine
certain colors, which I subjectively associate with the speed of my eye movements.
4. Proprioception also seems to offer affordance-based (action-related) information about what it
is I am seeing: If I look at an object that is taller than me, I might feel as if my eyes were
jumping over it, and might feel a corresponding “Jumping” sensation in my knees, as if I were
poised to jump. If I correctly recognize an object by sight, I do so by imagining what it would
be like to touch it (a counterfactual, as defined by Seth (2014).
The Role of Somatatopic maps
•Ruey-song Huang and colleagues (2010) found multisensory representations of the face and
body in the posterior parietal cortex. In collaboration with visual and motor areas, The PPC
synthesizes multisensory information about obstacles coming into contact with the face and body,
thus “guarding the body from head to toe.”
•These findings are consistent with my own tendency to locate objects in external space by using
a parapersonal reference frame. For example, when looking at the sides of an object, I feel a
sensation at the edges of my cheeks.
•My mouth becomes particularly sensitive during cannabinoid-induced visual experiences. I
become very aware of a sense of “forwardness” (towards my lips) and a sense of “backwardness”
(towards my throat.
•We will be looking at the eye-tracking data to see if there is a consistent mapping between my
subjective experience of objects being projected onto my face and body and their locations in
external space. If these mappings prove consistent, then they may be acting as a form of
bootstrapping that allow me to learn about the relationship between physical space and abstract
(brain-based or internal) space.
•In my experience, the more I feel as if I am being tested (i.e. the more controlled the experiment
is), the more my visual system disengages.
•In contrast, the more I am allowed to explore my environment in an unrestricted way, the more
vivid my visual experiences are. However, in these circumstances my eyes jump from stimulus to
stimulus, and it becomes nearly impossible to gather reliable eye-tracking data, or to match my
subjective descriptions with my eye movements.
•I have a strong desire to verbally describe what I am seeing and have those descriptions
corroborated by the experimenter. This is consistent with the work of Harris (2012) on how
children learn through testimony and Clark’s (2015) description of how language acts as artificial
•If I feel as if I am being judged, or if my visual perceptions are “wrong” (i.e. I lose confidence), I
•Though my conversations with the experimenter feel helpful and informative to me, they diminish
the integrity of the data, because it becomes influenced by his subjective observations (the so-
called Clever Hans effect).
•The closest we came to designing a controlled experiment was by having the experimenter ask
me where I felt an object on my face. This task was not difficult for me, and I did not feel judged.
However, my visual system did disengage after a few trials because the task lost its novelty, even
though I was still emotionally and intellectually engaged.
•We plan to use more precise eye-tracking devices to gather conclusive data about the interaction
between stimuli and my eye movements.
•If you would be interested in helping with the data analysis, please contact me. I have some
funding available. This would entail de-noising the data and reviewing and coding the video
according to our hypotheses. Since this study is exploratory, we are also open to other
•After reviewing our eye-tracking data, we will be making changes to our eye-training game so that
it is more engaging. This will make it a more effective training tool, and may improve our eye
tracking data by keeping me engaged during experiments while allowing me to explore my
environment in a somewhat controlled manner.
•Currently, the game uses a Tobii Eye-x to provide auditory information about what direction my
eyes are pointing. However, because of my nystagmus, my eyes often go off the screen, and I
lose feedback altogether.
•In addition, the game only rewards gross eye movements, not the small movements that feel so
rewarding during my cannabinoid experiences.
•The game could possibly be improved by converting the target into a moving avatar that would
only move when the eye-tracker is able to track my eye movements. There would also be alert
tones when approaching the midlines of the horizontal and vertical axes, as well as the edges of
the screen. However, it is uncertain whether our current eye-tracking setup is sensitive enough to
measure smaller eye movements and their velocities. To that end, if you have access to an
eyetracking setup that would better suit our purposes, please contact me.
Clark, A. (2015). Surfing Uncertainty: Prediction, Action, and the Embodied Mind (1 edition). Oxford;
New York: Oxford University Press.
Emrich, H. M., Leweke, F. M., & Schneider, U. (1997). Towards a cannabinoid hypothesis of
schizophrenia: cognitive impairments due to dysregulation of the endogenous cannabinoid
system. Pharmacology, Biochemistry, and Behavior, 56(4), 803–807.
Harris, P. L. (2012). Trusting What You’re Told: How Children Learn from Others. Cambridge, Mass:
Huang, R.-S., Chen, C., Tran, A. T., Holstein, K. L., & Sereno, M. I. (2012). Mapping multisensory
parietal face and body areas in humans. Proceedings of the National Academy of Sciences of
the United States of America, 109(44), 18114–18119.
Seth, A. K. (2014). A predictive processing theory of sensorimotor contingencies: Explaining the
puzzle of perceptual presence and its absence in synesthesia. Cognitive Neuroscience, 5(2),