What is a hallucination? The question seems simple enough. “A
hallucination is a perception in the absence of external stimulus that has qualities of real perception. Hallucinations are vivid, substantial, and are perceived to be located in external objective space.” When we think of visual hallucinations, we often think of trippy colorful images induced by psychedelic drugs (
hallucinogens).
Are
dreams hallucinations? How about visual
imagery? Optical
illusions of motion from viewing a non-moving pattern? No, no, and no (according to this narrow definition). Hallucinations are subjective and inaccessible to others, much as my recent posts discussed the
presence or absence of visual imagery in individual humans. However, people can tell us what they're seeing (unlike animals).
Visual hallucinations can occur in
psychotic disorders such as schizophrenia and schizoaffective disorder, although auditory hallucinations are more common in those conditions. Visual hallucinations are more often associated with neurodegenerative disorders. Among patients with
Parkinson's Disease, 33% to 75% experience visual hallucinations, usually related to dopaminergic or anticholinergic drug therapy.
In contrast, hallucinations in
dementia with Lewy Bodies (DLB) are diagnostic of the disease, and not related to pharmacological treatment. “Recurrent complex visual hallucinations ... are typically well-formed, often consisting of figures, such as people or animals.” The cause may be related to
pathology in subcortical visual structures such as the superior colliculus and the pulvinar, rather than the visual cortex itself. A
more specific hypothesis is that loss of α7 nicotinic receptors in the
thalamic reticular nucleus could lead to hallucinations in DLB.
Charles Bonnet Syndrome (CBS)Visual hallucinations are also caused by certain types of visual impairment, e.g. age-related
macular degeneration, which leads to the loss of central vision. Damage to the macular portion of the
retina can
cause people to “see” simple patterns of colors or shapes that aren't there, or even images of people, animals, flowers, planets, and scary figures. Individuals with CBS know that the hallucinations aren't real, but they're distressing nonetheless.
“Why are you discussing DLB and CBS here,” you might ask, “because these conditions don't involve abnormal stimulation of the visual cortex.” I brought them up because visual hallucinations in humans can occur for any number of reasons, not just from manipulation of highly specific cell types in primary visual cortex (which only occurs in optogenetic experiments with animals).
Electrical Stimulation Studies in HumansA typical starting point here would be Wilder Penfield and the history of surgical epileptology, but I'll skip ahead to the modern day. Patients with intractable epilepsy present teams of neurosurgeons, neurologists, neurophysiologists, and neuroscience researchers with a unique opportunity to probe the inner workings of the human brain. Stimulating and recording from regions thought to be the seizure focus (or origin) guide neurosurgeons to the precise tissue to remove, and data acquired from neighboring brain bits is used to make inferences about neural function and electrophysiological mechanisms.
An exciting study by Dr. Joseph Parzivi and colleagues (
2012) stimulated regions of the
fusiform face area (FFA) in the inferior temporal cortex while a patient was undergoing surgical monitoring. Two FFA subregions were identified using both fMRI and electrocorticography (ECoG).
The location of the face-selective regions converged across ECoG and fMRI studies that presented various stimuli and recorded brain responses in the FFA and nearby regions (1 = posterior fusiform; 2 = medial fusiform). Then the investigators stimulated these two focal points while the patient viewed faces, objects, and photos of famous faces and places. Electrical brain stimulation (EBS) of the FFA produced visual distortions while the patient viewed real faces. Sham stimulation, and EBS of nearby regions, did not produce these perceptual distortions. The article included a
video of the experiment, which is worth watching.
Another patient viewed pictures of faces during FFA stimulation and reported the persistence of facial images once they were gone, and the mixing of facial features, but no distortions (this is known as
palinopsia). A
third study induced the scary phenomenon of seeing yourself (self-face hallucination, or autoscopic hallucination), upon EBS of a non-FFA region (right medial occipitoparietal cortex). A video of this experiment is on
YouTube.
“But wait,” you say, “you've been describing complex visual hallucinations and distortions of the face because the EBS was in higher-order visual areas that are specialized for faces. What happens when you stimulate primary visual cortex?” The answer is less exciting (but not unexpected):
phosphenes, those non-specific images of light that appear when you close your eyes and press on your eyeballs (
Winawer & Parvizi, 2016). These can be mapped retinotopically according to their location in the visual field. {also see this
1930 article by Foerster & Penfield:
2"Stimulation of the occipital pole in area 17 produces an attack which is ushered in by an optic aura such as light, flames, stars, usually in the opposite visual field."}
But EBS of primary visual cortex is a coarse instrument. Here's where the latest refinements in
optogenetics finally enter the picture (
Marshel et al., 2019).
I won't attempt to cover the complex and novel techniques in Panel 1 and Panel 2 above. So I'll quote others who rave about what a breakthrough they are (and they are):
amazing work,
incredible breakthrough,
Key advances in current paper include multiSLM to stimulate neurons based on function, and a red-shifted opsin allowing simultaneous 2p. And one day (hypothetically speaking), I'd like to present more than direct quotes and my cartoonish version of the optogenetic ensemble and behavioral training methods. But today isn't that day.
Using ChRmine [a fancy new opsin] together with custom holographic devices to create arbitrarily specified light patterns [horizontally or vertically drifting gratings], we were able to measure naturally occurring large-scale 3D ensemble activity patterns during visual experience and then replay these natural patterns at the level of many individually specified cells. We found that driving specific ensembles of cells on the basis of natural stimulus-selectivity resulted in recruitment of a broad network with dynamical patterns corresponding to those elicited by real visual stimuli and also gave rise to the correctly selective behaviors even in the absence of visual input.
Briefly, the investigators captured patterns of activity in
V1 layer 2/3 neurons and layer 5 neurons that responded to horizontal or vertical gratings, and then played back the same patterns to those neurons in the absence of a visual stimulus. There goes the coarseness of EBS-induced phosphenes in humans... But obviously, the one great advantage of human studies is that your subjects can tell you what they see. Nonetheless, everyone wants to say that laser-activated nerve cells cause the mice to hallucinate vertical bars.
What really happened is that mice were trained to discriminate between horizontal and vertical gratings. The task required them to respond to the vertical, but not the horizontal. After training, visual stimulation with gratings was compared to optogenetic stimulation of classifier-identified neural ensembles in the absence of gratings. How well did the mice perform with optogenetic-only stimulation?
Modified from Fig. 5 (Marshel et al., 2019).(A) Discrimination performance during visual-only stimulation (black) and tuned-ensemble stimulation (red) over several weeks. (B) Discrimination performance for tuned-ensemble stimulation versus visual trials (P > 0.1 paired t test, two-tailed, n = 112 sessions across five mice). Eventually the mice did just about as well on the discrimination task with optogenetic stimulation of the horizontally or vertically-tuned neurons, compared to when the horizontal or vertical stimuli were actually presented. Were these mice “hallucinating” vertical gratings? Or did they merely learn to respond when a specific neural ensemble was activated? Isn't this somewhat like neurofeedback? During training, the mice were rewarded or punished based on their correct or incorrect response to the “vertical” ensemble stimulation. They can't tell us what, if anything, they saw under those conditions.
And the authors themselves noted the following limitation, that “mice initially required some training involving paired optogenetic and visual stimuli before optogenetic activation alone sufficed to drive behavioral discrimination.”
Marshel et al. correctly invoked the “it takes a village” explanation that many other cortical and subcortical regions are required to generate a full natural visual percept.
My frustration with the press coverage stems from inaccurate language and overblown interpretations.
3 [So what else is new?] From the
New York Times:
Why Are These Mice Hallucinating? Scientists Are in Their HeadsIn a laboratory at the Stanford University School of Medicine, the mice are seeing things. And it’s not because they’ve been given drugs.
With new laser technology, scientists have triggered specific hallucinations in mice by switching on a few neurons with beams of light. The researchers reported the results on Thursday in the journal Science.
The technique promises to provide clues to how the billions of neurons in the brain make sense of the environment. Eventually the research also may lead to new treatments for psychological disorders, including uncontrollable hallucinations.
The
Stanford press release doesn't use “hallucination” in the title, but a few are sprinkled throughout the text for dramatic effect: “
Hallucinations are spooky” and “
Hallucinating mice.”
Should we classify the following as a spooky hallucination: Optical stimulation of 20 vertical bar neurons in behaviorally trained mice who then perform the task as if the drifting vertical gratings were present in their visual field. I would say no. To be fair, in the
Science paper the authors used the word “hallucinations” only once, and it wasn't to describe mouse percepts.
Studying specific sensory experiences with ensemble stimulation under different conditions may help advance development of therapeutic strategies . . . for neuropsychiatric symptoms such as hallucinations or delusions. More broadly, the ability to track and control large cellular-resolution ensembles over time during learning, and to selectively link cells and ensembles together into behaviorally relevant circuitry, may have important implications for studying and leveraging plasticity underlying learning and memory in health and disease.
I'm focusing on only one small aspect of the study, albeit the one that grabs media attention. The results were highly informative in many other ways, and I do not want to detract from the monumental technical achievements of the research team.
Footnotes1 This is a terrific resource, with loads of information, additional artistic renderings, an
eBook, and a must-see
video.
2 There's no escaping Penfield...
3 See Appendix for expert opinion, since I am
not an expert...
ReferencesFoerster O, Penfield W. (1930).
The structural basis of traumatic epilepsy and results of radical operation.
Brain 53:99-119.
Marshel JH, Kim YS, Machado TA, Quirin S, Benson B, Kadmon J, Raja C, Chibukhchyan A, Ramakrishnan C, Inoue M, Shane JC, McKnight DJ, Yoshizawa S, Kato HE, Ganguli S, Deisseroth K. (2019).
Cortical layer-specific critical dynamics triggering perception.
Science Jul 18.
Parvizi J, Jacques C, Foster BL, Witthoft N, Rangarajan V, Weiner KS, Grill-Spector K. (2012).
Electrical stimulation of human fusiform face-selective regions distorts face perception.
J Neurosci. 32(43):14915-20.
Winawer J, Parvizi J. (2016).
Linking Electrical Stimulation of Human Primary Visual Cortex, Size of Affected Cortical Area, Neuronal Responses, and Subjective Experience.
Neuron 92(6): 1213-1219.
AppendixBefore lodging this critique, I consulted select experts on Twitter...
