Visual Hallucinations in Psychosis: The Curious Absence of the Primary Visual Cortex

Abstract

Background and Hypothesis

Approximately one-third of patients with a psychotic disorder experience visual hallucinations (VH). While new, more targeted treatment options are warranted, the pathophysiology of VH remains largely unknown. Previous studies hypothesized that VH result from impaired functioning of the vision-related networks and impaired interaction between those networks, including a possible functional disconnection between the primary visual cortex (V1) and higher-order visual processing regions. Testing these hypotheses requires sufficient data on brain activation during actual VH, but such data are extremely scarce.

Study Design

We therefore recruited seven participants with a psychotic disorder who were scanned in a 3 T fMRI scanner while indicating the occurrence of VH by pressing a button. Following the scan session, we interviewed participants about the VH experienced during scanning. We then used the fMRI scans to identify regions with increased or decreased activity during VH periods versus baseline (no VH).

Study Results

In six participants, V1 was not activated during VH, and in one participant V1 showed decreased activation. All participants reported complex VH such as human-like beings, objects and/or animals, during which higher-order visual areas and regions belonging to the vision-related networks on attention and memory were activated.

Discussion

These results indicate that VH are associated with diffuse involvement of the vision-related networks, with the exception of V1. We therefore propose a model for the pathophysiology of psychotic VH in which a dissociation of higher-order visual processing areas from V1 biases conscious perception away from reality and towards internally generated percepts.

Introduction

Visual hallucinations (VH) are defined as visual experiences occurring in the absence of corresponding external stimulation of the eye, with sufficient sense of reality to resemble a veridical visual percept, over which subjects do not feel they have direct and voluntary control, and which occur in the awake state.¹ People with a psychotic disorder experience VH in 37% of cases during their lifetime.² In these patients, VH with frightening content are common.³ However, the mechanisms underlying psychotic VH are largely unknown, which slows the development of new therapeutic approaches. These are needed, as 8% of first-episode patients still experience mild to moderate hallucinations after 1 year of using antipsychotics.⁴ Moreover, antipsychotics can have serious side effects, which reduces medication adherence.

To elucidate the pathophysiology of VH, three previous studies imaged brain activity during actual VH.^5–7 Two case studies did not find V1 activation,^5,6 while a third study (with unknown group size) showed V1 activation on an individual level, but not at the group level.⁷

VH can be roughly divided into simple and complex subtypes, which are related to the brain regions involved. Simple VH such as featureless shapes or colors originate in V1, whereas complex VH such as animals and faces are associated with activity in the higher-order visual areas.^8–11 In line with this, both case studies reported complex hallucinations.^5,6 Of note, although complex VH are the most common type in psychosis, most patients with psychosis experience both complex and simple hallucinations; a minority also report geometric VH.¹² Recent fMRI research in our lab supports this absence of V1 involvement in psychotic VH: compared to healthy controls during rest, VH-prone patients with psychosis showed severely decreased functional communication of the right V1 with other brain regions (submitted¹³). The absence of V1 activation during psychotic VH is remarkable, as a previous study indicated that conscious perception requires V1 involvement.¹⁴ In that study, a patient who underwent surgical ablation of V1 was completely unaware of the visual stimuli in the corresponding visual hemifield, yet scored far above chance on their localization and the discrimination of their shapes. Moreover, a study from 1999 on brain activation during auditory hallucinations (AH) in schizophrenia showed activation of the primary auditory cortex (A1).¹⁵ However, a later meta-analysis did not find sufficient evidence of A1 a.tivation.¹⁶ Similarly, previous studies also suggested that psychotic VH do not require V1 involvement^6,7; V1 activity in individuals may be caused by feedback signals from higher-order visual areas that were active during VH.⁷ A special role for V1 has also been suggested in VH occurring after ocular pathology (Charles Bonnet syndrome); they proposed that VH are caused by deafferentation of V1.^17,18 This deafferentation would lead to increased spontaneous activity, and thus to VH.

Besides involvement of the visual network (VIS), activation of vision-related networks have been shown to play a role in the generation of complex VH. These vision-related networks include the outside-world-focused Dorsal Attention Network (DAN), the inner-world-focused and memory-related Default Mode Network (DMN), and the saliency-focused Ventral Attention Network (VAN), which functions as a switch between DAN and DMN.¹⁹ Shine et al⁹ and Menon et al²⁰ postulated that complex VH are caused by widespread dysfunction of the vision-related networks, rather than by focal pathology,¹⁹ as (1) patients with Parkinson’s Disease (PD) mostly experience complex VH, although sometimes with low clarity,²¹ and (2) patients with ocular pathology can also experience complex VH. Other models also state that complex VH in schizophrenia, PD, and Lewy Body Dementia are caused by a convoluted interplay between incoming stimuli from the outer world and influences from brain areas on attention and memory.^22–26 Moreover, in a non-clinical population, simple and geometric VH, which were induced by photic stimulation within a specific frequency and luminance range through closed eyelids, also led to extensive activation of higher visual areas and brain areas involved in attention and memory.²⁷ For psychotic disorders specifically, this widespread dysfunction would lead to poor visual processing by the VIS, suboptimal assignment of relevance to visual stimuli by the VAN, and inadequate switching between the outer world (DAN) and inner-world (DMN) representations.¹⁹ Consequently, internally generated visual information might be perceived as a representation of the external world, resulting in VH. Previous studies reported activation of DMN and DAN structures during psychotic VH^5,6 and activation of the VAN during combined audiovisual hallucinations.⁷ This appears to confirm involvement of the vision-related networks in psychotic VH.

Although several hypotheses on the origin of VH have been proposed, only a few studies have tested them. Here we tested the hypothesis that subjects with psychosis mostly experience complex VH, accompanied by a decrease in or absence of V1 activity, and a more diffuse increase in the activity of brain regions processing the specific VH content (higher-order visual areas) and in regions belonging to the visual attention- and memory-related networks (DAN, VAN, and DMN). We therefore used fMRI to acquire brain activation data in seven participants with psychosis who experienced VH during the scanning. This was combined with phenomenological data that was collected after the fMRI acquisition.

Methods and Materials

Participants

This case series was part of INZICHT2 (https://trialsearch.who.int/Trial2.aspx?TrialID=NTR6855), which included 15 participants with psychosis with VH. Participants were recruited via INZICHT1 (https://trialsearch.who.int/Trial2.aspx?TrialID=NTR5103), the Department of Psychotic Disorders, UMCG, and Lentis Center for Mental Health Groningen and Winschoten. Participants were aged 18–55. They met DSM-IV-TR criteria for schizophrenia, schizophreniform disorder, schizoaffective disorder, or psychotic disorder NOS²⁸ and had VH more than a couple of times in the last month. Exclusion criteria were interfering brain disorders and severe cognitive impairment (< 26 on the Mini-Mental State Examination²⁹). The study followed the tenets of the Declaration of Helsinki and was approved by the UMCG ethics board. Participants provided written informed consent. They received a 50 euro coupon for participating in INZICHT2.

Questionnaires

The Positive and Negative Syndrome Scale (PANSS³⁰) and the Questionnaire for Psychotic Experiences (QPE³¹) were used to assess the psychotic symptoms.

Visual Evaluation

The ophthalmic examination consisted of visual acuity (digital Snellen chart), contrast sensitivity (GECKO chart, maximum score 16) and visual fields (Humphrey Field Analyzer, 24-2 SITA; Carl Zeiss Meditec).

MRI and fMRI Acquisition

Participants underwent one 10-min fMRI scan, with dimmed lights in the scanner room. They were instructed to relax, not to perform any task or think any specific thought and to stay awake. They indicated the start and finish of periods with VH by pressing buttons with their right index and middle finger, respectively. Data was collected with a 3 T Philips Magnetic Resonance system with a standard 32-channel SENSE head coil (Intera, Philips Medical Systems). Echo-planar images were acquired, anterior-posterior, TR 2 s, TE 30 ms, voxel size 3 × 3 × 3 mm, flip angle 90° (FOV 192 × 117 × 192), 39 slices per volume. An anatomical T1 scan was also acquired (160 slices, voxel size 1 × 1 × 1 mm, FOV 256 × 160 × 224 mm). Directly after scanning participants were interviewed about the characteristics of their hallucinations.

Data Analysis

Data were analyzed per participant. Due to the small sample size and the large variation in clinical and hallucination characteristics, we did not perform group analyses or comparisons between participants. FMRI pre-processing included realignment, co-registration, normalization to MNI space, and 8 mm smoothing. ICA-AROMA was used to remove motion artifacts in functional scans.³² Subsequently, data were analyzed using SPM12 (Wellcome centre human neuroimaging), running in Matlab 2014b.³³ For each participant, VH periods were entered into the General Linear Model. Two contrasts were used: (1) increased, and (2) decreased activity during VH periods versus baseline (no VH).

The voxel-wise threshold was P < 0.001 (uncorrected); on cluster level a 0.05 family wise error rate was used.

Results

In total, eleven participants with frequent VH were included; eight of them actually experienced VH during scanning, but technical difficulties led to a missing log file for one of these participants. In total, the data from seven participants (age range 27–39 years, 3 women, 4 men) was analyzed (see Table 1 for demographics and disease characteristics). All participants had intact visual fields. Their mean visual acuity was 1.1 (SD 0.2); their mean contrast sensitivity 15.4 (SD 1.1), which are both in the normal range. Table 2 lists the characteristics of the individually experienced VH during scanning and the brain regions that showed either increased or decreased activation during VH periods. The mean number of VH periods during scanning was 15.9 (SD 24.7 range 1-–0) with a mean duration of 105.4 s (SD 87.7 s, range 6.2–225 s). Participants 1, 4, 5, and 6 kept their eyes open during scanning. All participants reported experiencing relatively complex VH such as human or human-like beings, objects, and/or animals during scanning. One of them also reported experiencing VH of low complexity. Another participant reported environmental tilt VH. Five participants reported experiencing AH. Two participants experienced AH almost continuously, while in three others AH occurred only occasionally. One participant reported experiencing a tactile hallucination. The activity patterns showed that in six out of seven participants, V1 was not active during VH. In one participant V1 showed decreased activation. In participants 1, 2, 3, 4, 6, and 7, the experience of VH was associated with an overall increase in brain activity. The most frequently activated regions of the ventral visual stream were the lateral occipital complex (LOC), the middle temporal gyrus posterior (MTGp), and the superior temporal sulcus posterior (STSp). Moreover, the Inferior Parietal Lobule (IPL) (dorsal visual stream), lateral frontal regions (DAN and VAN), and medial cortical structures (DMN) were frequently activated. In participant 5 we observed an overall decrease in activity during VH in higher visual areas in the ventral and dorsal stream, the DAN and DMN, and other areas.

Table 1.

Demographic and illness characteristics

Partic.	Age	Gender	Diagnosis	Psychiatric/neuro-logical comorbidity	Disease dur (year)	Antipsychotics	Other ­medication	PANSS*				QPE, severity#				Vis. acuity Contrast
								P3	P	n	Tot	AH	VH	Htot	Del	ODS~	ODS’
1	34	f	Schizoaffective disorder	Personality dis NOS,traits of a borderline and avoidant pers dis	9	Aripiprazol 7.5 mg Zuclopentixol 10 mg Propranolol 10 mg Promethazine (s.o.s) Clonazepam (s.o.s)	Lithium 800 mg Mirtazapine 30 mg	5	17	13	66	19	12	31	0	100	15
2	28	f	Schizophrenia	Dysthymic disorder Migraine-like headache	13	–	Rizatriptan (s.o.s) Oral ­contraceptive	5	11	11	45	19	3	22	0	80	16
3	32	m	Schizoaffective disorder	–	12	Clozapine 250 mg Aripiprazol 10 mg	–	5	17	24	81	32	14	16	10	80	16
4	36	m	Psychotic dis NOS	Bipolar disorder	24	Aripiprazol 10 mg	Lithium 800 mg Levothyroxine 75 mcg	5	21	19	82	22	11	36	13	100	16
5	31	f	Schizoaffective disorder	Traits of borderline personality disorder	13	Haloperidol 5 mg Aripiprazol 2.5 mg	Lithium 800 mg Biperiden 2 mg Fluoxetine 20 mg Diazepam 9 mg Temazepam 10 mg Loratidine 10 mg Proponalol 20 mg Omeprazol 20 mg	4	13	17	69	12	15	27	11	120	16
6	39	m	Schizophrenia	Gilles de la Tourette	2	–	Omeprazol 40 mg	4	28	18	100	8	20	33	18	150	16
7	27	m	Psychotic ­disorder NOS	–	2	–	–	5	25	25	100	23	16	39	9	80	13

The Positive and Negative Syndrome Scale ³⁰. P3, hallucinatory behavior; P, total score positive symptoms; N, total score negative symptoms; Tot, total score.

Questionnaire for Psychotic Experiences ³¹. Severity AH (Auditory Hallucinations): the sum of items A1, A3–A6, A11–A13, A15. Severity VH (Visual Hallucinations): the sum of items V1, V3–V6, V11–V13, V15. Severity Htot (total hallucinations): the sum of items A1, A3–A6, A11–A13, A15, V1, V3–V6, V11–V13, V15, O1, O3. Severity Del (delusions): the sum of items W10–W13.

Visual acuity of Oculi Dextra and Sinistra simultaneously (digital Snellen chart).

Contrast sensitivity of Oculi Dextra and Sinistra simultaneously (GECKO chart).

Table 2.

VH during scanning: phenomenology and fMRI activity patterns

Pt	VH PHENOMENOLOGY	INCREASED ACTIVATION DURING VH					DECREASED ACTIVATION DURING VH
		Anatomical region	MNI X, y, z	k	T max	P-value	Anatomical region	MNI X, y, z	k	T max	P-value
1	5 VH, total duration: 225 s	Amygdala r	25, 4, −17	434	5.44	0.000
	An american author standing peacefully and quiet next to her, wearing a white robe, staring ahead	Thalamus r	13, −26, −2		5.13
	A spider, which was terrifying. Luckily she was able to send it to Mars with her thoughts	Hippocampus r	13, −8, −23		4.95	0.000
	The whole room was filled with cloudy, green-blue sea water. This did not scare her, as she knew it was not real	Superior temporal sulcus post r	55, −41, 4	472	5.36
	A 15 cm bee flying across the room. In her thoughts, she opened the air grate to let it out	Middle temporal gyrus r, t-o	61, −62, −2		4.85
	A starfish with sweet little eyes, attached to the fMRI scanner. It made squeaking sounds, continuously saying “I like it here”	Supramarginal gyrus r	61, −41, 25		4.42	0.000
	A 3–4 m spider with scary feet. It changed into Dalek (Doctor Who)	Superior colliculus l	−5, −35, −5	365	5.26
	A swimming dolphin, making dolphin sounds	Hippocampus l/WM	−20, −17, −11		4.59
	Sméagol (Lord of the Rings) putting his hand on her left knee, which she did not feel	Putamen l/WM	29, −14, −8		4.46	0.000
	The whole room was filled with cloudy, green-blue sea water. This did not scare her, as she knew it was not real	Medial frontal gyrus r	7, 22, 40	321	5.23
		Corpus callosum r	4, 10, 25		4.89
		Anterior cingulate cortex l	−2, −2, 31		4.32	0.001
		Middle temporal gyrus l, t-o	−53, −56, 4	168	4.86
		LOC inferior l	−59, −68, 7		4.06
		LOC inferior l	−50, −77, 13		3.69	0.001
		VLPFC r	25, 40, 28	164	4.64
		VLPFC r	22, 49, 16		3.98
		OPFC r	13, 61, 22		3.26	0.004
		VLPFC l	−44, 13, 16	131	4.63
		Frontal operculum l	−38, 22, 7		4.62
		Middle temporal gyrus ant l	−59, −5, −14		4.11	0.037
		Temporal pole l	−47, 10, −20	78	3.76
2	5 VH, total duration: 27,9 s	Precentral sulcus r	46, 7, 28	151	4.69	0.001	Postcentral gyrus r	7, -41, 70	132	4.20	0.001
	People	VLPFC r	43, 13, 19		4.47	0.006	Precentral gyrus r	10, -29, 70
	Someone sitting behind a desk	Middle temporal gyrus l, t-o	−65, −50, 1	99	4.43		Precentral gyrus r	4, −32, 61
	Scenes	Superior temporal sulcus post r	−56, −53, 7		4.17	0.000
	A tree	LOC inferior l	−62, −62, 10		3.30
		Supramarginal gyrus r	49, −35, 52	228	4.41
		Supramarginal gyrus r	55, −38, 22		4.25	0.028
		Sup temporal gyrus r, t-o	61, −38, 10		3.91
		Supramarginal gyrus r	−59, −35, 37	71	3.99
		Supramarginal gyrus r	−62, −44, 43		3.87
		Supramarginal gyrus l	−56, −47, 55		3.38
3	70 VH, total duration: 172,7 s	Premotor cortex r	28, −8, 49	4202	6.74	0.000	–
	White, short, beamer-like flashes of light, originating out of his forehead. They indicated the answer to the questions he heard in his head	Precentral sulcus l	−26, −8, 46		6.64
	White, non-transparent energy, of variable sizes and shapes	Premotor cortex r	52, 4, 25		6.25
		Angular gyrus r	61, −47, 31	804	6.27	0.000
		Middle temporal gyrus r, t-o	49, −53, 4		5.57
		Middle temporal gyrus r, t-o	55, −56, −5		5.18
		LOC inferior l	−47, −62, −2	767	5.12	0.000
		Cerebellum VI r	28, −50, −26		4.98
		Cerebellum VI l	−32, −59, −29		4.87
		Precuneus r	7, −56, 55	100	4.96	0.031
		Superior parietal lobule r	22, −68, 58		3.64
		Precuneus r	10, −71, 55		3.59
		Postcentral sulcus r	49, −26, 46	146	4.12	0.007
		Intraparietal sulcus r	34, −50, 52		3.90
		Postcentral sulcus r	40, −41, 61		3.59
4	8 VH, total	Angular gyrus l	−62, −59, 22	170	5.26	0.000	Cuneal cortex r	4, −83, 43	1286	5.02	0.000
	duration: 155,5 s	Mid temporal gyrus l, t-o	−62, −59, 4		3.96		Intracalcarine cortex r	13, −56, 10		4.80
	A black, unfamiliar cat, walking back and forth on his chest	Supramarginal gyrus l	−50, −47, 31		3.95		Cuneus WM r	13, −83, 25		4.58
	A man bending over him. He wore a black coat and hat. His face was expressionless. The participant did not know this man, although he had seen him before in his hallucinations
5	2 VH, total	–					DLPFC l	−23, 43, 43	524	5.36	0.000
	duration: 15,6 s						DLPFC l	−23, 19, 64		4.35
	Via the mirror of the fMRI scanner, she saw heads and necks of puppy’s popping up						DLPFC l	−29, 16, 58		4.23	0.000
	During her second VH, she saw herself tilted in the mirror, half upside down, while she also felt being turned herself						Superior parietal lobe l	−44, −47, 61	1527	5.32
							Precuneus r	1, −62, 55		4.96
							Angular gyrus l	−53, −53, 34		4.86
							Temporal fusiform cortex l	−47, −62, −23	102	4.89	0.011
							DLPFC r	28, 19, 49	121	4.80	0.005
							DLPFC r	22, 34, 49		3.87
							Inferior parietal gyrus r	40, −71, 43	277	4.63	0.000
							Angular gyrus r	40, −50, 34		4.46
							Angular gyrus r	46, −50, 55		4.04
							Angular gyrus r	64, −47, 13	69	4.55	0.048
							Mid temporal gyrus post r	64, −23, −8	116	4.49	0.006
							Mid temporal gyrus post r	64, −35, −14		3.70
							Inf temporal gyrus post r	55, −38, −14		3.68
							Cerebellum crus I r	25, −74, −20	286	4.15	0.000
							Cerebellum crus I r	43, −68, −26		4.11
							Occipital fusiform gyrus r	34, −80, −20		3.97
6	20 VH, total duration: 135,1 s	LOC inferior l	−44, −62, 10	1858	7.01	0.000	–
	Seeing a person moving just outside the scanner, like someone was looking at him to see if he could see him. Sometimes he saw an arm, sometimes a leg. The arm moved, like the person was waving at him	LOC inferior l	−56, −68, 7		6.75
		Cerebellum crus I l	−38, −59, −29		5.61	0.000
		LOC inferior r	55, −62, 7	717	6.72
		Middle temporal gyrus r, t-o	40, −56, 10		6.22
		LOC inferior r	43, −59, 1		5.91	0.001
		VLPFC l	−35, 31, 28	119	6.24
		DLPFC l	−32, 46, 28		3.91	0.000
		Precentral gyrus r	43, −2, 58	211	6.12
		Precentral gyrus r	52, 4, 37		4.91
		Precentral gyrus r	37, 1, 46		4.74	0.000
		Sup temporal sulcus post r	58, −29, 1	628	5.87
		Supramarginal gyrus r	64, −38, 25		5.66
		Supramarginal gyrus r	61, −26, 22		5.54	0.000
		Frontal opercular cortex r	37, 25, 4	365	5.49
		Temporal pole r	52, 19, −11		5.28
		OPFC r	52, 16, 1		5.20	0.000
		VLPFC r	37, 46, 22	290	5.49
		DLPFC r	31, 40, 46		5.19
		DLPFC r	16, 49, 31		4.98	0.000
		Premotor cortex r	10, −2, 73	312	5.31
		SMA r	1, −5, 70		5.13
		Premotor cortex l	−17, −2, 67		4.40	0.003
		Angular gyrus r	55, −47, 52		4.79
		Superior parietal lobe r	40, −44, 52	102	4.68
		Intraparietal sulcus r	43, −32, 40		3.43	0.001
		Temporal pole l	−56, 10, 4	124	4.38
		Precentral gyrus l	−59, 10, 22		4.16
		Temporal pole l	−62, 7, −5		4.13
7	1 VH, total duration: 6,2 s	Cerebellum crus I l	−44, −50, −38	495	5.22	0.000	Cerebellum vermis X/ r IX	4, −47, −38	57	4.02	0.044
	He saw the head of an unfamiliar female creature taking a sip of coffee. The head was normal-sized and was located in the middle-right visual field	Cerebellair peduncle l	−14, −35, −41		4.68		Cerebellum vermis X/ l IX	−5, −47, −35		3.93
		Cerebellum l WM	−26, −53, −41		4.67	0.000
		Lingual gyrus (V2) r	10, −86, −8	402	4.82
		Occipital pole WM r	19, −89, 13		4.52
		Intracalcarine cortex l	−2, −80, 13		4.46	0.002
		Precentral gyrus l	−38, −23, 61	107	4.64
		Precentral gyrus l	−35, −23, 70		4.50	0.000
		Medial frontal gyrus r	7, 37, 64	140	4.49
		SMA l	−2, −8, 76		4.07	0.001
		DLPFC r	16, 25, 64		3.86
		OPFC r	22, 37, −14	122	4.45	0.013
		Frontal pole med r	10, 46, −14		4.16	0.007
		OPFC r	31, 25, −8		3.66
		T-o fusiform gyrus l	−44, −62, −17	77	4.17
		Inferior temporal gyrus, t-o l	−53, −59, −14		4.14
		Intraparietal sulcus l	−29, −62, 49	89	3.99
		Inferior parietal lobule l	−35, −71, 49		3.83
		Superior parietal lobule WM l	−26, −47, 46		3.59

R, right; l, left; WM, white matter; Ant, anterior; Post, posterior; Mid, middle; T-o, temporo-occipital part; LOC, lateral occipital complex; VLPFC, ventrolateral prefrontal cortex; OPFC, orbitoprefrontal cortex; DLPFC, dorsolateral prefrontal cortex; SMA, supplementary motor area.

Main column 1: participant number. Main column 2: phenomenology of Visual Hallucinations (VH) during fMRI scanning, as reported by the participants directly after scanning. Upper part: number of VH during scanning and their total duration. Lower part: descriptions of the reported VH, in chronological order. Main columns 3 and 4: anatomical brain regions showing increased (column 3) or decreased (column 4) activity during VH periods versus baseline (periods without VH; voxel-wise threshold: P < 0.001 (uncorrected); on cluster level: 0.05 family wise error rate). The start and finish of VH periods were indicated by pressing a button with the right index and middle finger, respectively. Reported are x, y, and z values in MNI space, K (cluster size in number of voxels), and the T_max and P-value of the corresponding cluster. Lower part of columns 3 and 4: visualization of activated brain regions during VH, rendered on a standard glass brain using ParaView.³⁴

Individual Results

Table 2 shows the phenomenology of VH during scanning and the activated and deactivated areas during VH. The activated higher-order VIS areas in participant 1 largely paralleled the content of the reported VH; human bodies and body parts are processed in the MTGp,³⁵ faces are processed in the LOC inferior (LOCi) and the STSp,³⁶ and animals in the LOC.³⁷ Furthermore, the DMN (hippocampus and medial frontal cortices,³⁸ VAN (amygdala and VLPFC) and DAN (DLPFC) were activated. Activity of the supramarginal gyrus (SMG) may relate to higher somatosensory processing or to the auditory component of the VH^39,40; superior colliculus activity is related to vertical eye movements, target selection and attention.⁴¹ Button presses may have caused putaminal activity,⁴² while communication between different cortical areas may have led to thalamic involvement.⁴³ In participant 2, increased activation of higher VIS areas also corresponded to the complex character of the VH. Again, the VLPFC (VAN) and SMG were activated during VH. The button presses likely caused precentral activation.^44,45 It is not clear why the right gyrus precentralis and postcentralis showed decreased activation. Participant 3 also showed activity in higher-order VIS areas as the LOC, the MTG temporal-occipital part (MTGto), which processes motion,⁴⁶ and the Angular Gyrus (AG; dorsal part of the VIS).^9,47 The DMN (precuneus), the VAN (left Cerebellum VI⁴⁸) and the DAN (the intraparietal sulcus, IPs)) were also active during VH. The button presses may have activated frontal motor areas^44,45 and the right cerebellum VI.⁴⁸ Superior parietal lobe (SPL) activity may indicate spatial attentional switching,⁴⁹ whereas activation around the right postcentral sulcus may indicate ipsilateral sensory connections related to touching the button. Participant 4 showed decreased activity in V1 and the right cuneus. Again the MTGto, AG, and SMG were activated. Participant 5 only showed decreased activation. Interestingly, her VH included a tilt of the visual axis, which is associated with IPL hypofunction.⁸ This corresponds with our found IPL deactivation. This specific type of hallucination might also have induced other sensory experiences. Other regions, including higher VIS areas, the DAN (the DLPFC), and the DMN (the precuneus, cerebellum crus I) also showed decreased activity. However, it is unclear how these sensory experiences may have led to an overall decrease of activity in the visual networks. Participant 6 also showed activation of the higher VIS areas corresponding to the VH content. The DAN (dorsolateral prefrontal cortex [DLPFC], IPs), the VAN (ventrolateral prefrontal cortex [VLPFC]) and the DMN (cerebellum crus I^19,48) were also activated. Activation of frontal motor areas was most likely caused by the button presses, although supplementary motor area (SMA) activity may also indicate responding to false predictions related to the VH.⁵⁰ In participant 7, early visual areas including V1 were active during VH, as were higher ventral and dorsal VIS areas. Furthermore, the DAN (DLPFC) was activated, whereas some areas of the DMN were activated (the medial frontal gyrus, cerebellum crus I) and others deactivated (the bilateral cerebellum vermis IX/X⁴⁸). Frontal motor areas and the SPL also showed activation.

Discussion

Our main finding is the absence of V1 activity during VH in the majority of our participants. In general, participants experienced complex VH, and showed activation of the higher-order visual areas and regions belonging to the visual-attention-related networks (DAN and VAN) and memory (DMN).

Disconnection of V1 During VH, Involvement of Higher Visual Areas

During VH, V1 showed increased activation in only 1 out of 7 participants; in 1 other participant the early visual areas showed decreased activation. This corroborates recent studies on AH and VH that found that involvement of the primary sensory cortices is not necessary for conscious perception.^5–7,16 In another study, two patients lacking a functioning V1 still reported seeing phosphenes in their blind hemifield when Transcranial Magnetic Stimulation (TMS) was applied to the ipsilateral IPs.⁵¹ Moreover, another patient with a complete unilateral V1 lesion had some subjective phenomenal awareness in the blind field. Our data showed activation of higher visual areas during VH. These activity patterns correspond closely to the complex content of the VH, similarly to previous studies.^5,6 A functional dissociation between V1 and higher visual areas was reported previously in patients with psychosis: although the occipital network functioned as a single integrated network in healthy controls, in patients with psychosis it was split into four “subnetworks”.⁵² Together with the caudal part of the inferior temporal gyrus, V1 acted as an independent subnetwork. Altered anatomy of the visual cortex may play a role in the V1 disconnection. Compared to controls, patients with psychosis had thinner cortices in early visual areas and the LOC.⁵³ However, another study only found reduced thickness of visual association areas, not of V1.⁷

Widespread and Variable Brain Region Activation, Covering All Vision-Related Networks

The widespread activation across all brain areas during VH and the large variability between participants confirm our hypotheses about the diffuse involvement of the vision-related networks. Activation of DMN structures is in line with VH involving memory retrieval and self-reference^38,54–56 and has been reported before.^5,6,57 VH in psychosis are associated with an increased structural connectivity between visual areas and the hippocampus.⁵⁷ However, one participant in our study showed deactivation of a cerebellar DMN structure, and another study reported DMN withdrawal during psychotic VH.⁷ This suggests a complex role for memory in psychotic VH. In accordance with previous studies,^5,7 our results showed widespread activation of DAN and VAN.¹⁹ Activation of the amygdala was seen in one participant who experienced frightening VH. Interestingly, in patients with psychosis and VH the amygdala, part of the VAN, is hyperconnected with the visual cortex.⁵⁸

A Model for the Cause of Psychotic VH: The Dissociated V1

Our data confirm the hypotheses of Shine et al and Menon et al^19,20: instead of focal pathology we found widespread involvement of all the vision-related networks in psychotic VH. Accordingly, an effective connectivity study in patients with psychosis showed that during the transition into hallucinations (which were mostly auditory), destabilization of the connections between the salience network (closely related to the VAN), the frontoparietal Central Executive Network (a task-positive network like the DAN) and the DMN occurred.⁵⁹ We also confirmed that V1 is mostly excluded from VH activation. Combining these results, we suggest that activity during VH represents a widespread feedback loop, across the visual network (except for V1) and the vision related-networks on memory and attention the DAN, DMN, and VAN (see figure 1). The question is where the loop starts. The above-mentioned effective connectivity study found that during the actual hallucinations information was transferred from the hippocampus to the salience network.⁵⁹ We postulate that the dissociation of V1 from the rest of the brain plays a key role in psychotic VH generation. Normally, V1 provides feed-forward information to the vision-related networks, leading to it being identified as coming from the outer world. Moreover, V1 receives inhibitory feedback projections from higher visual areas.^60,61 The postulated V1 dissociation may lead to an impairment in both feedforward and feedback signaling. Moreover, V1 dysconnectivity may also lead to increased spontaneous activity of the higher visual areas that V1 projects onto, as in Charles Bonnet Syndrome.¹⁷ Either way, hallucinating subjects consequently interpret internally-generated visual signals as having a cause in the outer world. However, our finding that V1 is generally not activated during VH could also indicate that V1 is simply not involved in—and is thus irrelevant to—the occurrence of psychotic VH, although our other study showed V1 dysconnectivity in VH-prone psychotic patients.¹³ Because V1 plays a very important role in normal visual processing, we can hypothesize that V1 is key to the generation of psychotic VH. Regardless of the relevance of V1, we can postulate that impaired co-functioning of the vision-related networks contributes to the lack of inhibition and therefore to continuation of these visual information loops. The LOC likely functions as a hub in this loop. In a recent fMRI study we found a strong functional connection between this area and other parts of the brain during rest (submitted¹³). Moreover, in the current study the LOC was consistently activated during VH. The hub function of the LOC may be compensatory for the disconnected V1. Such a central role for the LOC is consistent with the phenomenology of psychotic VH, which mostly includes human or human-like beings and animals. However, data from our lab also showed more widespread occipital dysfunction, with reduced functional connectivity within the VIS (submitted¹³), and reduced occipital responsiveness around image recognition of genuine percepts (submitted⁶²) in VH-prone patients. This suggests that many more visual areas are involved in the “hallucinatory” loop, which can give rise to various VH about objects and landscapes, as well as geometric and simple VH. This model also explains other clinical characteristics of VH in psychosis. Activity across all vision-related networks during VH may explain why VH resemble real visual percepts.^3,12 Their sometimes bizarre character may be caused by aberrant salience mapping by the VAN, while their personal character is consistent with the involvement of memory-related networks (DMN), both of which show activation during VH.

Fig. 1.

A model for the cause of psychotic VH: the disconnected V1. In our model, complex psychotic visual hallucinations are the result of a dissociation of higher-order visual processing areas from the primary visual cortex (V1). Functionally, the dissociation manifests itself by an (curious) absence of primary visual cortex activity and a simultaneous looping of brain activity across the vision-related functional networks on memory and attention. These vision-related networks include the outside-world focused Dorsal Attention Network (DAN), the inner-world focused and memory-related Default Mode Network (DMN), and the saliency-focused Ventral Attention Network (VAN), which functions as a switch between DAN and DMN.¹⁹ Shine et al¹⁹ and Menon et al²⁰ postulated that widespread dysfunction would lead to poor visual processing by the VIS, suboptimal assignment of relevance to visual stimuli by the VAN, and inadequate switching between the outer world (DAN) and inner-world (DMN) representations. Consequently, this biases conscious visual perception away from information derived from the outer world and towards internally generated percepts, i.e. visual hallucinations. Figure created with BioRender.com.

Clinical Implications

Our model suggests that psychotic VH could be treated by enhancing the feed-forward signals from V1. As such, increasing light levels or watching videos or pictures may diminish VH, analogous to how listening to music reduces the duration and severity of AH in psychosis.^63,64 Applying stimulatory repetitive TMS (rTMS) to V1 may also enhance feed-forward signaling. However, the medial position of V1makes it hard to reach by rTMS, and it can induce phosphenes.⁶⁵ Alternatively, applying inhibitory rTMS (irTMS) to the hub region LOC may reduce VH by perturbing signal transmission and looping within the vision-related networks. In schizophrenia, irTMS on higher auditory areas is already used for reducing AH, with moderate effect.⁶⁶ IrTMS to the right occipital cortex significantly reduced the severity and complexity of VH (n = 1).⁶⁷ Moreover, irTMS substantially reduced the VH severity in a patient with combined PD and severe visual impairment.⁶⁸ That study targeted the bilateral SMA because Eigenvector Centrality Mapping of fMRI signals had indicated that both SMA’s served as hub regions during VH. Similarly, irTMS on the LOC or other VH-related hubs may be a promising approach for treating VH in psychosis. Defining individual targets for rTMS may require patients to first undergo an fMRI scan, which could create a barrier to participation. Therefore, generic irTMS treatment of the LOC may be considered first.

Limitations

This study had a relatively small group size, which reduced its statistical power. However, the majority of patients (6/7) did show the same patterns and our study has substantially expanded the number of participants on whom fMRI data were acquired while hallucinating. A second limitation is that during scanning, our participants could choose to keep their eyes closed or open, which may have affected results. Nevertheless, visual comparison of the activation patterns in both groups did not show marked differences. Finally, our use of fMRI does not allow statements on causality. Nevertheless, combining our data with those of previous studies has provided new insights into the possible causes of psychotic VH, as summarized in our model. Future studies could use Eigenvector Centrality Mapping and effective connectivity modeling to infer causality and determine if there are common final pathways in VH. Future treatments could focus on these common pathways.

In conclusion, during psychotic VH, V1 dissociates from the vision-related networks, while there is a simultaneous and widespread increase in the activation of higher-order-visual areas and visual attention and memory-related networks. We postulate that this dissociation affects conscious perception by biasing it away from the visual information about the external world and towards internally generated percepts, thereby resulting in VH.

Conflict of Interest statement

None.

Funding

This project has received funding from: MD/PhD grant for MvO from the Graduate School Medical Sciences (GSMS) Groningen, the department of Neurology and the Laboratory of Experimental Ophthalmology of the University Medical Center Groningen, the Rob Giel Research Centre Groningen. The funding organizations had no role in the design, conduct, analysis, or publication of this research.