In schizophrenia, an abnormal reduction in neuronal gamma oscillations (30-100 Hz) is associated with negative symptoms such as cognitive dysfunction. The literature suggests that rescuing gamma oscillations through non-invasive brain stimulation may be an accessible and safe add-on strategy to mitigate negative symptoms. Here, a stimulation protocol based on gamma visual stimulation will be tested. This pilot study will follow an uncontrolled clinical trial design: A minimum of ten patients diagnosed with schizophrenia or a schizoaffective disorder and predominant negative symptoms will be recruited at Klinikum rechts der Isar. They will undergo a multisession stimulation protocol, consisting of one hour of 40 Hz visual stimulation per day over five consecutive days, during which they will be encouraged to fall asleep. An equal number of patients will be recruited for a treatment-as-usual group without intervention. Pre- and post-assessments will include EEG, a cognitive test battery (THINC-IT), a mood scale (PANAS), and a schizophrenia symptom scale (PANSS). This study's results will inform on the feasibility of gamma visual stimulation as a potential add-on intervention in schizophrenia.

SAMPLE Inward patients at Klinikum rechts der Isar, Munich, Germany diagnosed with schizophrenia (F20) or a schizoaffective disorder (F25) will be recruited. Treating physicians will screen and pre-select patients who are primarily affected by negative symptoms and on stable medication and/or psychotherapy (treatment as usual) throughout the study period. The target sample size for this pilot is at least N = 10 patients who undergo the experimental study protocol, and an equal number of patients who only complete the scales as a treatment-as-usual control group.

PROTOCOL - EXPERIMENTAL GROUP The protocol for the experimental group will be as follows. After written informed consent was obtained by their treating physician, patients will undergo session 0 for pre-tests five to one day(s) before the first stimulation session. For a fair comparison with the post-tests in the last session, patients will be asked not to drink coffee in the three hours beforehand, to rest at the lab for one hour, and then to complete the mood and cognition scales (PANAS and THINC-IT). The psychiatric symptom assessment (PANSS) will be carried out by their treating physician before the first stimulation in session 1.

The stimulation protocol will take place the week after, from Monday to Friday, at approximately the same time of day as session 0. Patients should again avoid consuming coffee three hours before each session start, to facilitate sleep. In session 1 (Monday), the EEG will first be set up. The session will start with a five-minute resting-state EEG recording; then, patients will lay down and undergo the visual stimulation protocol with their eyes closed for one hour. They will be encouraged to relax and fall asleep during this time. Afterward, we will inquire about adverse events and the EEG will be removed. In sessions 2 to 4 (Tuesday to Thursday), no EEG will be recorded. Patients will undergo one hour of visual stimulation and subsequently be asked about adverse events in the same way as in session 1. In the last session 5, the protocol will be the same as in session 1, including EEG, five minutes of resting-state recording, and one hour of visual stimulation. Afterward, as in session 0, patients will also complete the mood and cognition scales (PANAS and THINC-IT). The psychiatric symptom assessment (PANSS) will be carried out by their treating physician zero to three days after the last stimulation session.

PROTOCOL - CONTROL GROUP As for the control group, after written informed consent was obtained by their treating physician, patients will undergo session 0 analogously to the experimental group for the pre-tests. This procedure will be repeated 7-10 days later for the post-tests.

VISUAL STIMULATION Visual stimulation parameters will be analogous to a previous study (Hainke et al., 2025). A customized sleep mask with inbuilt LEDs externally linked to a microcontroller will be used to deliver visual stimuli. Its high-wavelength LEDs with a narrow spectral peak at 605 nm and an illuminance of 80 lux will flicker at 40 Hz in a square wave pattern at a 50 % duty cycle. Light will be faded in at the beginning and out at the end for 10 seconds, respectively. Patients will be asked to keep their eyes closed and encouraged to fall asleep for the full stimulation duration of 60 minutes.

After every stimulation session, the experimenter will verbally inquire about adverse events, by asking the patient: "Have you experienced any undesired effects during or after the stimulation?". If the patient answers yes, they will be asked to describe the adverse event and then rate its severity as mild, moderate or severe and whether it was unlikely, likely or certainly related to the stimulation.

For the resting-state measurement, the LEDs will be flickering at 40 Hz as during the stimulation, but they will be covered with black tape ("blackout"). This controls for the possibility of electrical interference from the mask on EEG data (Hainke et al., 2025). Here, patients will remain awake.

EEG SETUP EEG will be measured in the experimental group at sessions 1 and 5 using the Neurofax system at the clinic's sleep laboratory, supported by Polaris.One software (Nihon Kohden Europe GmbH, Rosbach v.d.H., Germany). The sampling rate will be 1000 Hz. Gold cup electrodes will be positioned at A1, A2 (mastoids reference), FpZ (ground), left EOG, left EMG, C3, C4, O1, Oz, O2, PO3, POz, and PO4. Sleep scoring will be performed automatedly and offline using the Python library YASA based on the C4, EOG, and EMG electrodes, and participants' sex and age (Vallat & Walker, 2021).

EEG PROCESSING

Data will be pre-processed using MNE Python as in Hainke et al. (2025):

Band-pass filtering (0.16-300 Hz) Bad channel rejection by visual inspection Electrode averaging by Region of Interest: central (C3, C4) and occipital (O1, Oz, O2, PO3, POz, PO4) Re-referencing to the mastoid average (A1, A2) Dividing data into 30-second epochs Bad epoch rejection (<50% sleep scoring algorithm certainty or stimulation duration <25 seconds) For frequency domain analyses, epochs with a peak to peak amplitude > 1 mV will be rejected. All 30-s epochs will be subjected to a Fast-Fourier Transform using a Hamming window to obtain the Power Spectral Density per epoch. For time domain analyses, data will be further divided into 25 ms segments length of one cycle of a 40 Hz oscillation); segments with a peak-to-peak amplitude >100 μV will be rejected. 30-s epochs and 25-ms segments will be averaged by region of interest (central / occipital), condition (blackout / stimulation), session (1 / 5), and state (awake / light sleep / deep sleep). Light sleep is defined as NREM1 and NREM2; deep sleep is defined as NREM3.

EEG OUTCOMES The two resulting EEG outcomes describe the Steady-State Visually Evoked Potentials (SSVEPs), i.e., the magnitude of the neuronal response to 40 Hz visual stimulation, from complementary perspectives. EEG signals can be interpreted in the time domain or the frequency domain; analysing both allows for maximal information gain about underlying neuronal processes (Hainke et al., 2025). In the time domain, SSVEP magnitude is quantified as the peak-to-peak amplitude of the 25-ms segment average in microvolts. In the frequency domain, SSVEP magnitude is quantified as the signal-to-noise ratio of power at 40 Hz, i.e., the power spectral density value at 40 Hz in dB divided by the surrounding values [38 to 39.5 Hz] + [40.5 to 42 Hz] in dB.

COGNITIVE & PSYCHIATRIC OUTCOMES The following scales will be completed by both the experimental and control group patients, in the respective first and last session. Cognition will be assessed with the THINC-integrated tool test battery (THINC-IT; Harrison et al., 2018), presented on a Windows laptop. It has previously been used to assess cognition in schizophrenia (Szmyd et al., 2023) and has five components: Spotter (Choice Reaction Time), Symbol Check (1-back test), Trails (Trails Making Test B), Codebreaker (Digit Symbol Substitution Test), and a self-reported cognitive function questionnaire (5-item Perceived Deficit Questionnaire). Mood will be measured with the Positive And Negative Affect Schedule (PANAS; Watson et al., 1988). The Positive and Negative Syndrome Scale (PANSS; Kay et al., 1987) will be administered by a trained physician to quantify schizophrenia-related psychiatric symptoms.

An interim analysis will be conducted when a sample of 10 patients in the experimental group has been reached.

Inclusion Criteria:

• Medical diagnosis of schizophrenia (F20) or schizoaffective disorder (F25)

Exclusion Criteria:

• Age < 18 years
• Any history of seizures
• Acute suicidality assessed with the Columbia-Suicide Severity Rating Scale (C-SSRS; Brent et al., 2008)
• Any other relevant axis 1 disorder
• Red-green colour blindness or current ocular disease
• Alcohol, cannabis, or illicit drug addiction within the last 3 months

Anonymised IPD that underlie results in a publication will be shared, including EEG, THINC-IT, PANAS, PANSS, and adverse events reports data.

IPD and supporting information will be made available immediately after the publication of study results, for an unrestricted duration.

The de-identified individual patient data, i.e., all IPD that underlie results in a publication, will be made accessible after its publication for non-commercial academic projects that have a legitimate research topic and a clearly stated hypothesis. In the event that the application is accepted, researchers will be asked to get the study approved by their institution's ethics board. The study principal investigator will subsequently provide the de-identified data sets via a safe data transfer system.