Acute effects of psilocybin on the dynamics of gaze fixations during visual aesthetic perception

Abstract

Serotonergic psychedelics are remarkable for their capacity to induce variable yet reproducible modifications to human consciousness. The acute effects of these compounds include perceptual alterations, predominantly in the visual domain, yet these alterations have been mostly documented only by subjective reports. We used eye-tracking to quantify the effects of low vs. high doses of psilocybin mushrooms on eye movements during the exploration of complex visual stimuli under semi-naturalistic conditions, focusing on the case of aesthetic perception. The experimental condition (high vs. low dose) was a priori unknown to participants and experimenters. High doses resulted in a more local visual exploration of paintings, and a less entropic distribution of fixations. While psilocybin altered gaze behavior and increased subjective emotional intensity and feelings of flow, it did not affect the aesthetic ratings of the stimuli, suggesting a dissociation between perceptual and evaluative aspects of aesthetic experience. These findings suggest that psilocybin may influence gaze fixation by altering the perception of low-level visual features, including textures, shapes, and colors. Our work highlights the possibility of investigating psychedelics by addressing their effect on behavior under complex naturalistic conditions, contributing to maintaining subject engagement while also increasing the ecological validity of the findings.

Introduction

Moderate to high doses of classic psychedelic substances (i.e. LSD, psilocybin, DMT and mescaline) are known to produce profound alterations in consciousness, including acute changes in perception, emotion, cognition, and the sense of self¹² The effects of serotonergic psychedelics on human subjective experience show consistency despite individual variation, drug dose and non-pharmacological inter-individual factors³. Perceptual alterations are among the most prominent and commonly reported acute effects induced by psychedelics. These alterations encompass visual and auditory distortions, changes in body schema and tactile perception, the onset or enhancement of synesthesia and disrupted perception of time¹. Changes to visual saliency processing are also frequently reported, with some items in the environment becoming more salient and personally relevant than usual⁴.

Early research attempted to provide insight into how psychedelics affect perception and cognitive functions (e.g., memory, attention, decision-making, etc.)⁵. However, several of these early studies present limitations such as unrefined methodological standards, inconsistent blinding procedures and control groups, non-standardized dosages and heterogeneous rating scales⁶. In addition, the tools used to study the effects of psychedelics in the human brain were limited by the technology available at the time. Researchers relied mainly on subjective self-reports from participants, observational data and animal models⁷. However, as psychedelic research declined in the 70 s and 80 s, new tools and technologies emerged. The development of neuroimaging techniques such as MRI and fMRI allowed researchers to visualize and study brain structure and function in vivo, while eye tracking technology improved, becoming more widely available and affordable, and allowing accurate measurements of gaze direction capable of informing cognitive processes underlying perception and attention⁸. Even though recent studies have begun to re-examine the psychophysical effects of psychedelics and their capacity to alter perception, most of this research attempts to correlate brain imaging data with self-reported sensory alterations^7,9. Thus, a gap in knowledge remains concerning the impact of psychedelic drugs on objective behavioral metrics, such as the dynamics of sensory engagement with the environment quantified using eye-tracking.

This study was conducted in a semi-natural setting, with experimental conditions aiming to retain the ecological validity of a real-world setting (e.g. familiar environment, self-selected time) while maintaining core experimental controls (e.g. randomization, self-blinding), with the experimental condition a priori unknown to participants and researchers. However, blinding experimental conditions is difficult in naturalistic settings. Therefore, we adopted a self-blinding procedure intended to reduce the influence of participant expectations prior to the experimental session¹⁰. We used eye tracking to measure the dynamics of eye fixations as participants freely explored a selection of classical paintings while under the acute effects of high and low doses of psilocybin present in Psilocybe cubensis mushrooms. Additionally, we recorded self-reported evaluations of emotional valence and beauty ratings for each painting. As a first attempt to investigate the effects of psychedelics on the dynamics of visual perception, our study is inherently exploratory. Our methodology was deliberately chosen to foster a high level of engagement while minimizing possible sources of stress and distraction¹¹. Moreover, the choice of classical paintings was based on scientific evidence and anecdotal reports indicating that psychedelics positively interact with aesthetic experience^12,14,15,16. Subjects participated in this experiment in natural settings, with the agreement to engage in an active control condition (i.e. a low dose of psilocybin not expected to introduce major changes in subjective experience) which was prepared and randomized by themselves¹⁰. Because of this semi-naturalistic design, participants benefited from a familiar and comfortable research setting, which could attenuate their overall distraction and contribute to mitigate confounding factors related to stress and anxiety, potentially amplified by the acute effects of the drug³.

To complement the exploratory nature of this study, we propose a hypothesis based on the Relaxed Brain Under Psychedelics (REBUS) model, a theoretical proposal of how psychedelics alter brain function to elicit changes in consciousness and cognition. According to this model, psychedelics bring about the revision of heavily weighted high-level priors¹³. During visual perception, this could be manifest as a reduced focus on salient features (i.e. informative priors for image recognition and categorization), and thus as a more unbiased and homogeneous sampling of the visual stimuli – or equivalently, as an increased entropy in the spatial distribution of gaze fixations^{14,15,16,17,18,19,20,21}. These predictions can be tested by comparing the probability distribution of fixations between the high and low dose conditions.

Materials and methods

Participants

Twenty-three participants (four females, 31 ± 4 years, 72 ± 15 kg [mean ± STD]) were recruited through word of mouth and social media advertising. To participate, individuals were instructed to contact the provided number via WhatsApp, and subsequently received a phone call from the researchers. During the call, the researchers provided a brief explanation of the details and purpose of the experiment, as well as the inclusion and exclusion criteria. Participants were then given a full written explanation of the study and a copy of the informed consent form. To determine eligibility, all participants underwent a psychiatric interview to screen for exclusion criteria, which are detailed in the supplementary material and summarized below. After this screening, participants and researchers agreed on a date for the start of the experiment. Participants reported having 15 ± 13 prior experiences with serotonergic psychedelics, of which 3.1 ± 2.4 were considered challenging [mean ± STD]. Participants had normal or corrected-to-normal vision. All participants completed the full experimental protocol, including self-report questionnaires. Two participants opted to stop the eye-tracking task during the high dose condition, and six were excluded from eye-tracking analyses due to insufficient data quality, yielding a final sample of 15 participants for eye-tracking analysis (four females, 31 ± 4 years, 70 ± 16 kg [mean ± STD]).

A priori power analysis assuming an expected effect size of d = 0.6, significance level α = 0.05, and power = 0.8 yielded a target sample size of 23 participants. After data quality exclusions, the final sample for eye-tracking analyses consisted of 15 participants. A sensitivity analysis showed that this sample size provides 80% power to detect an effect of d = 0.76 or larger (two-tailed, paired design).

This study was conducted in accordance with the Declaration of Helsinki and approved by the Research Ethics Committee at the Universidad Abierta Interamericana (Buenos Aires, Argentina), protocol number 0-1068. All participants gave written informed consent and received no financial compensation for their participation in the experiment. All data was collected from the participants in natural settings, i.e. those chosen by the participants without intervention from the researchers. The researchers did not provide nor administered psilocybin to the participants, nor instructed them in any way concerning drug use. Participants were briefed on the rationale of self-blinding, but were not required to follow any specific self-blinding procedure.

Inclusion and exclusion criteria

Participants were required to have at least two prior experiences with a dose equal to or exceeding 3 g of dried psilocybin mushrooms. In addition to minimizing the risk of adverse psychological reactions, including only individuals with prior psychedelic experience helped ensure that participants were familiar with the nature of the effects¹⁵. This was important to reduce the likelihood of distraction or anxiety due to the novelty of the experience, which could interfere with task performance. Our goal was to maintain focus and engagement during the experimental procedures, thereby improving data quality under both conditions.

To participate in this research protocol, subjects volunteered to partake in a series of tests under the effects of psilocybin mushrooms and in the presence of four members of the research team. Subjects who consumed serotonergic psychedelics during the 15 days prior to the dosing day were not included in the study to avoid tolerance induced by recent use of psilocybin or other serotonergic psychedelics¹⁶. The same applied for all psychoactive substances (including alcohol, caffeine and tobacco) for a period of 24 h prior to the dosing day, with the objective of avoiding potential pharmacological interactions as well as drug-induced interference in the acute effects of psilocybin¹⁵. To participate in the experiment, subjects declared their willingness to abstain from using psychedelics between measurement sessions.

Subjects who fulfilled DSM-5 criteria for the following disorders were excluded from the experiment: schizophrenia or other psychotic disorders, type 1 or 2 bipolar disorder (including first- and second-degree relatives), personality disorders, dissociative identity disorder, post-traumatic stress disorder, substance abuse or dependence in the past 5 years, depressive disorders, recurrent depressive episodes, obsessive-compulsive disorder, generalized anxiety disorder, dysthymia, panic disorder, bulimia or anorexia, as well as subjects with a history of neurological disorders. Pregnant women and subjects under psychiatric medication of any kind were excluded. Subjects exhibiting potential dysfunctional states as measured by the Depression Anxiety Stress Scale (DASS)¹⁷ (with scores > 4 for depression, > 3 for anxiety and > 7 for stress) were subsequently reviewed by the clinical interviewer for confirmation.

Experimental design and setting

Experiments were conducted in private homes chosen by the participants with researchers present at the moment. Conditions (high vs. low dose) were a priori unknown to participants and researchers as subjects implemented a self-blinding procedure, following the design introduced by Szigeti and colleagues¹⁰. The experiment was divided into two parts, one corresponding to the dosing condition ($\:\approx\:$3 g of ground and homogenized dried psilocybin mushrooms in gel capsules, provided by the participants) and one corresponding to an active control condition ($\:\approx\:$0.5 g of ground and homogenized dried psilocybin mushrooms mixed with edible mushrooms to match the weight of the gel capsules between conditions). Conditions were separated by an interval of one month to attenuate potential tolerance effects. As stated above, while researchers explained participants the rationale and objective of the self-blinding procedure (i.e. randomly dividing the two doses in capsules, while keeping the identity of the capsules unknown until the data analysis stage), they did not require nor instruct participants to follow any specific self-blinding procedure. During interviews conducted after completion of the experiment, we confirmed that participants implemented the self-blinded assisted by a third party of their choice. In all cases, psilocybin intake and experiments took place in the comfortable setting of a house and in the presence of the participant and the team of researchers. Two participants opted to stop the eye tracking task during the high dose condition and were therefore excluded from the following analysis; however, no adverse effects were reported.

The eye-tracking task was part of a larger experimental session that included EEG recordings and additional tasks (e.g., language production and music listening), all conducted under the same dosing condition. The session began approximately one hour after capsule ingestion, with the art viewing task consistently placed near the beginning of the sequence. A preparatory period of about 30–45 min was used to allow participants to acclimate and to complete the EEG setup. The total duration of the experimental session was approximately 3.5 to 5 h.

Acute effects

During the acute effects, subjects used a visual analogue scale (VAS) to provide their subjective intensity report for the following items: “Sounds influence what I see”, “My sense of size and space is distorted”, “I feel unusual bodily sensations”, “I see geometric patterns”, “Edges seem warped”, “I see movement in things that aren’t really moving”, “Things look strange”. The VAS used in this experiment is the abbreviated form of a previously used version¹⁸ which is restricted to include items that focus on the effects of the drug on perception. The VAS was completed four times, the first one hour after dose ingestion and then every one hour until the effects of the drug subsided, as determined by the subjective report of the participants.

Self-reported scales and questionnaires

Two days prior to the dosing day of each condition, subjects completed the following self-reported scales designed to assess different psychological traits: State-Trait Anxiety Inventory (STAI-Trait), Big Five Inventory (BFI)¹⁹.Tellegen Absorption Scale (TAS)²⁰ and Short Suggestibility Scale (SSS)²¹. A battery of self-reported scales was administered both before dosing and immediately after the acute effects, including the State-Trait Anxiety Inventory (STAI-State)²²the Positive and Negative Affect Schedule (PANAS), and the Psychological Well-being Scale (BIEPS)²³. Additionally, before dosing, subjects completed scales assessing expectations (EXP) and contextual factors prior to psychedelic experiences (PRE)²⁴and after the acute effects they completed the Mystical Experience Scale (MEQ) and the Altered States of Consciousness Scale (5D-ASC)²⁵. Participants also completed the Aesthetic Experience Questionnaire (AEQ)²⁶ on the dosing day immediately after completing the eye-tracking task, which is the primary questionnaire that is reported in this work. While the AEQ questionnaire refers to how the respondents usually perceive artworks, we clarified to the participants of our study that their responses should reflect their aesthetic perception during the acute effects of the drug. See the supplementary material for a detailed description of the other questionnaires.

Stimuli

A total of 30 paintings were selected from a collection of 100 paintings published by the BBC (https://www.bbc.co.uk/programmes/p009ypzq), ranging from 12th century China to the 1950s. The artworks in the BBC collection were selected by an expert art critic to represent multiple styles, themes and artistic periods. We randomly selected 30 of these to ensure a diverse yet standardized stimulus set. The curator of this collection avoided particularly popular or famous works, contributing to mitigate the familiarity of participants with the stimuli; nevertheless, some paintings in the final subset may still have been familiar to participants. We did not assess prior familiarity systematically during data collection, which constitutes a limitation of our study. A table with the author, title, year, artistic period and main themes for each artwork can be found in the supplementary material (Table S2). Images had a minimum resolution of 768 $\:\times\:$ 536 pixels, a maximum resolution of 1280 $\:\times\:$ 768 pixels, and were presented in front of a gray background.

Eye tracking

The measurements reported below were conducted as part of a larger experiment designed to investigate the effects of psychedelics on perception, creativity, language and music production, as well as the associated electrophysiological correlates measured with EEG, which will be presented in future reports.

For the eye tracking task, participants were seated 60 cm from a centrally located monitor with their heads supported by a custom-made chin rest to prevent head movement. Experiments were conducted using a Gazepoint GP3 HD portable eye-tracker, which is optimized to measure eye fixations under diverse and natural conditions and could be easily transported to the experimental setting. The monitor used for image presentation had a resolution of 1920 $\:\times\:$ 1080 pixels and a size of 31 $\:\times\:$ 17.5 cm. While presenting the visual stimuli, the spatial and temporal coordinates of the gaze location were recorded using the Gazepoint GP3 HD portable eye tracker, with a temporal resolution of 150 Hz and a visual angle accuracy of 0.5–1.0 degrees. Both the eye tracker control and the presentation of the different stimuli were performed with a custom script written in Python using the open-source library PsychoPy version 2021.1.0 (https://www.psychopy.org/).

The experiment started one hour after the participants consumed the capsules that contained either the low dose or the high dose of active mushroom material. The outline of the experiment is presented in Fig. 1. Each block (T) began with the presentation of a fixation cross for 2 s; next, the stimulus was presented for 30 s. Two self-reported VAS were then presented to assess the subject’s emotional valence and beauty rating of the previously presented painting. This process was repeated until all stimuli were shown (30 blocks). The same stimuli were shown for each experimental condition, with the order of appearance randomized for each subject and experimental condition. To avoid changes in the measurement accuracy over time, a standard eye tracker 5-point calibration procedure was performed at the end of each fifth block.

Statistical fixation metrics

Due to insufficient data quality (more than 30% of stimuli presenting more than 30% of the fixations missing or outside the screen, suggestive of lack of engagement with the task) the data of 6 participants were excluded from the analysis presented in this and the following section. This threshold excluded all subjects who failed to engage with the task (i.e. subjects with a majority of off-screen fixations), while simultaneously resulting in a final number of participants that was adequate for statistical analysis. Fixations outside the images were removed from the subsequent analysis. We did not find a significant difference in the number of removed fixations between conditions (see Figure S1 of the supplementary material).

Fig. 1

Outline of the eye tracking task. After an initial calibration, the first block (T1) begins. Each block consists of the following steps: a fixation cross is displayed for 2 s, followed by the presentation of the stimulus for 30 s, during which participants freely explore the painting while their gaze is being recorded by the eye tracker. Emotional valence and beauty ratings were assessed after each painting. This procedure was repeated for the 30 paintings, with intermediate eye tracker calibration every 5 blocks.

After recording the coordinates of the gaze position, fixations were detected automatically using the built-in velocity-based algorithm provided by the Gazepoint GP3 HD software, accessed through the iohub module in PsychoPy²⁷with minimum fixation durations of 150 ms²⁸. This resulted in the average horizontal and vertical positions ($\:{x}_{i}$, $\:{y}_{i}$), as well as the time of the i-th fixation ($\:{t}_{i}$), with the positions recorded in pixels and the time in seconds. Since the dimensions of the selected paintings varied, all fixations were standardized to a common dimension of 1280 × 768 pixels to allow straightforward comparison of distances. For each image, participant and condition, the following measures were calculated: the total number of fixations within the image ($\:N$), the mean distance between fixations ($\:ds$), the mean time between fixations ($\:dt$) and the mean distance between all pairs of fixations ($\:{ds}^{2}$). These metrics are outlined in Fig. 4a, and are computed as follows:

$$\:ds=\:\frac{1}{N-1}{\sum\:}_{i=1}^{N-1}\sqrt{{{(x}_{i+1}-{x}_{i})}^{2}+{{(y}_{i+1}-{y}_{i})}^{2}}$$

$$\:dt=\:\frac{1}{N-1}{\sum\:}_{i=1}^{N-1}{t}_{i+1}-{t}_{i})$$

$$\:{ds}^{2}=\:\frac{1}{N(N-1)}\:{\sum\:}_{i=1}^{N}{\sum\:}_{j=1,j\ne\:i}^{N}\sqrt{{{(x}_{i}-{x}_{j})}^{2}+{{(y}_{i}-{y}_{j})}^{2}}\:$$

(1)

Additionally, the standard deviation was also calculated for each image for the distances between consecutive fixations ($\:ds$) and the distances between all pairs of fixations ($\:{ds}^{2}$), as follows:

$$\:\sigma\:\left(ds\right)=\:\sqrt{{\frac{1}{N-2}{\sum\:}_{i=1}^{N-1}{({{x}_{i+1}-{x}_{i})}^{2}+{{(y}_{i+1}-{y}_{i})}^{2}-ds)}^{2}}^{}}$$

$$\:{\sigma\:(ds}^{2})=\:\sqrt{{\frac{1}{N(N-1)-2}{\sum\:}_{i=1}^{N-1}{\sum\:}_{j=1,j\ne\:i}^{N}{({{x}_{i}-{x}_{j})}^{2}+{{(y}_{i}-{y}_{j})}^{2}-ds2)}^{2}}^{}}$$

(2)

These measures were then averaged across images, resulting in a mean value for each participant and dosing condition.

Shannon entropy

The Shannon entropy (H) of the 2D fixation probability distribution was calculated by first discretizing the standardized fixation coordinates into 2D histograms, using square bins of four different sizes: 40 × 40, 60 × 60, 80 × 80, and 100 × 100 pixels. For each participant and condition, we computed the relative frequency of fixations within each bin, and then calculated entropy according to Shannon’s formula²⁹:

$$\:SE=-{\sum\:}_{i=1}^{{N}_{i}}{\sum\:}_{j=1}^{{N}_{j}}{P}_{ij}\:log\left({P}_{ij}\right)$$

(3)

where $\:{P}_{ij}\:$is the probability of a fixation falling within bin $\:i,j$. Entropy values were computed separately for each bin width to assess the robustness of the effects across spatial resolutions.

Statistical analyses

The results of the questionnaires from both dosing conditions were compared using Student’s t-test for paired samples, as implemented in Python’s scipy library version 1.11.4 (https://scipy.org). We reported uncorrected p-values and also indicated instances where p-values were significant when corrected using Bejamini-Hochberg’s method for False Discovery Rate (FDR) with $\:\alpha\:$=0.05, except in cases where the multiple tests did not meet the assumption of statistical independence (for instance, results obtained from computing entropy using different bin widths). Effect sizes were estimated using Cohen’s d and the rank-biserial correlation (RBC) otherwise for the result of parametric and non-parametric tests, respectively. All effect sizes are reported in Table 1 and Table S1 of the supplementary material.

Frequentist methods were complemented using Bayesian statistics to compare the evidence in favor of the null hypothesis with that in favor of the alternative hypothesis. We computed the Bayesian statistic BF10 (Bayes factor in favor of the alternative hypothesis over the null hypothesis) as implemented in Python’s pingouin library version 0.5.3 (https://pingouin-stats.org)³⁰. Eye-tracking measurements and VAS scores from both dosing conditions were compared using non-parametric paired Wilcoxon signed-rank tests, using Python’s scipy library. The results obtained using a complementary approach based on linear mixed effects model are shown in Table S3 of the supplementary material.

In the figures, all boxplots extend from the lower to upper quartile values with a line at the median; the whiskers extend from the upper/lower quartiles up to 1.5 times the interquartile range. Single points scattered over the boxplots represent data points from individual subjects.

Results

Participants correctly determined the experimental condition in 42 of the 46 measurement days (91% accuracy), with the same percentage for high dose and the active control condition. The results of all the self-reported scales and questionnaires appear in the supplementary material (Table S1). As expected, the high dose condition resulted in significant increases across all items of the 5D-ASC and MEQ30 questionnaires. Moreover, most of the negative results also presented BF10 factors below 1/3, interpretable as evidence in favor of the null hypothesis (i.e. no effect of the psilocybin). We fitted a linear mixed effects model to the data including the ordering of the conditions (high dose first, and vice-versa) as a factor, and confirmed the lack of significant interactions (p < 0.05, FDR-corrected for multiple comparisons) between the experimental conditions, as well as the randomized order in which they were completed by the participants.

Acute effects

To obtain an index representing the overall intensity of the acute effects experienced by subjects, we computed the sum of all the VAS items at each measurement time point. These results are summarized in Table 1. As expected, all four VAS total scores were significantly higher for the high dose of psilocybin vs. the active control (low dose). Figure 2 shows the temporal evolution of the total VAS score, as well as the scores for each individual item, averaged across participants for both dosing conditions. For all VAS scores, a peak in the intensity of the effects appeared between 2 and 3 h after drug intake, with a tendency to decrease after 4 h. This figure also indicates the time interval when the eye tracking task was performed, as well as the time point when the AEQ was completed by the participants.

Fig. 2

Acute effects measured using individual VAS items and the overall intensity of the experience given by the sum of all the items. Results are shown for each measurement time point, with consecutive measurements separated by one hour. The points indicate the mean across participants and the vertical lines the standard error of the mean (SEM). The time during which the eye tracking task (ET) took place and when the AEQ was completed are both indicated in the x-axis. Statistical significance is indicated using asterisks (Wilcoxon signed-rank tests). *p < 0.05 (Benjamini-Hochberg FDR correction).

Subjective ratings during the perception task

The scores for the six factors of the AEQ were compared between the dosing conditions using Student t-tests, revealing that the emotional response factor and the flow of experience factor scored significantly higher in the high dose condition (Fig. 3a; Table 1). The emotional valence and perceived beauty VAS scores were averaged across the images for each participant and compared using Wilcoxon signed-rank tests, showing no significant differences between dosing conditions (Fig. 3b; Table 1). Importantly, the AEQ emotional response subscale reflects participants’ overall emotional engagement with the aesthetic experience across the entire session, whereas the VAS emotional valence score corresponds to their immediate reaction to each individual painting. These two measures thus capture different aspects of emotional processing, which may explain the absence of significant effects in the trial-wise VAS scores despite higher global AEQ ratings in the high dose condition.

Table 1 Acute effects outcomes grouped by domain, presented as mean ± standard deviation, together with statistical significance and effect size (*p < 0.05, **p < 0.01, ***p < 0.001).

Fig. 3

Results of the subjective reports obtained during the aesthetic perception task. (a) Mean AEQ scores for each sub-scale, averaged across stimuli. (b) Emotional valence and perceived beauty VAS scores, averaged across stimuli. The p-values computed using Wilcoxon signed-rank tests are shown as insets. *p < 0.05 (Benjamini-Hochberg FDR correction).

Statistical fixation metrics

As described in the Methods, the total number of fixations ($\:N$), the average distance between fixations ($\:ds$), the average time between fixations ($\:dt$) and the average distance between all pairs of fixations ($\:{dt}^{2}$) were calculated for each stimulus and averaged across all stimuli for each participant and dosing condition. The resulting values were compared between dosing conditions using non-parametric paired Wilcoxon signed-rank tests (Fig. 4b; Table 1). A significant effect of the dosing condition was found for the metrics $\:ds\:$and $\:{ds}^{2}$, in both cases indicative of lower values for the high dose condition. Figure 4c presents example fixation heatmaps showcasing the difference between metrics $\:ds$ and $\:{ds}^{2}$.

The standard deviation of the distances between consecutive fixations ($\:\sigma\:\left(ds\right)$) and the standard deviation of the distances between all pairs of fixations ($\:\sigma\:\left({ds}^{2}\right)$) were calculated and averaged across images for each participant and dosing condition. For both metrics, the statistical analysis also revealed significantly lower values for the high dose condition (Fig. 5; Table 1).

Fig. 4

Changes in fixation metrics in high vs. low dose conditions. (a) Illustration of the statistical fixation metrics that were calculated for each image. From left to right: number of fixations ($\:N$), distance between consecutive fixations ($\:ds$), time between consecutive fixations ($\:dt$) and distance between all pairs of fixations ($\:{ds}^{2}$). (b) Values of the metrics $\:N$, $\:ds$, $\:dt\:$and $\:{ds}^{2}$, averaged across stimuli for each subject, and compared between the high dose of psilocybin (PSI) vs. the active control (AC) condition. The p-values (Wilcoxon signed-rank tests) are shown as insets. *p < 0.05 (Benjamini-Hochberg FDR correction). (c) Example heatmaps showcasing the difference between metrics $\:ds$ (left) and $\:{ds}^{2}\:$(right). Additional examples are provided in supplementary figure S4.

Fig. 5

Average values of the standard deviations of the distances between consecutive fixations $\:\sigma\:\left(ds\right)$ (left) and of the distances between all pairs of fixations $\:\sigma\:\left({ds}^{2}\right)$ (right). Results are shown for the high dose of psilocybin (PSI) and active control (AC) conditions and compared using Wilcoxon signed-rank tests (p-values shown as insets). *p < 0.05 (Benjamini-Hochberg FDR correction).

We conducted Spearman correlation analyses between total VAS scores (reflecting subjective intensity) and both eye-tracking metrics and aesthetic ratings, using data from all participants across placebo and psilocybin conditions. The analyses were performed separately for each combination of VAS score and behavioral measure. To control for multiple comparisons, we applied False Discovery Rate (FDR) correction. None of the correlations remained significant after this correction (results shown in supplementary figures S2 and S3). We also performed repeated measures correlation analyses between each eye-tracking metric and subjective ratings of emotion and beauty, separately for the high and low dose conditions. No significant differences were found between conditions for any of the correlations (results are summarized in supplementary table S4).

Shannon entropy

Fig. 6

shows the Shannon entropy calculated for both conditions using four different bin widths. For each panel, the resulting fixation probability distribution is shown on the left, while the boxplots with the comparison between high and low dose conditions is shown on the right. Significant results were obtained for all bin widths, indicating lower entropy values for the high dose condition (Table 1).

Figure 6. Psilocybin reduces the Shannon entropy of the fixation probability distribution. Panels show results obtained for the following bin width ($\:bw$) values: 40, 60, 80 and 100. For each panel, the resulting fixation probability distribution is shown on the left, while the boxplots with the comparison between high (PSI) and low dose (AC) conditions is shown on the right. The p-values computed using Wilcoxon signed-rank tests are shown as insets. *p < 0.05 (uncorrected).

Finally, we investigated whether correlations existed between the subjective reports (Fig. 3) and the fixation metrics assessed using eye tracking (Fig. 4). For this purpose, we computed the Pearson linear correlation coefficient between the metrics $\:N$, $\:ds$, $\:dt$, $\:{ds}^{2}$ and the AEQ dimensions, both for the high and low dose conditions. After correcting the significance of these correlations for multiple comparisons using Bejamini-Hochberg’s method for False Discovery Rate (FDR) with α = 0. 05, we found 4 significant correlations in the control condition and none in the dose condition (all correlations are provided in supplementary table S5). We then conducted statistical tests to compare the correlation coefficients between conditions for these correlations^31,32. This analysis revealed only one significant difference for the correlation coefficients between $\:ds$ and the emotional response sub-scale of the AEQ. As shown in Fig. 7, a significant correlation of r = 0.71 was found for the low dose condition; however, no correlation between these variables existed during the high dose condition.

Fig. 7

Scatter plot of the fixation metrics vs. the emotional sub-scale of the AEQ, including the best linear fit in the least squares sense for the active control condition (AC) and high dose (PSI) conditions. The correlation coefficient obtained for the high dose condition was r = 0.079, while the value for the low dose condition was r = 0.71, resulting in a significant increase of the Pearson correlation from the low to the high dose condition.

Discussion

Following a self-blinded paradigm under semi-naturalistic conditions with an active control (low dose), we investigated the modulation of perception by psilocybin, a natural serotonergic psychedelic of major contemporary interest in basic and clinical neuroscience. The experiment was designed to be self-blinded, with the experimental condition a priori unknown to participants and researchers. Nevertheless, there was a considerable unblinding rate in the participants. Instead of focusing solely on the measurement and analysis of subjective reports, we used eye tracking to quantify the behavior of participants regarding the statistics of their visual fixations when presented paintings from different art periods. These stimuli are naturally engaging and diverge from everyday visual scenes, thus factoring in an element of uncertainty in the experiment; moreover, their use represents an opportunity to further our knowledge of how psychedelic compounds influence aesthetic perception, a long-standing field of research which has gained attention recently^12,33. Our results suggest that participants under the high dose acquire information through a more local visual exploration, with shorter distances between all fixation pairs, as well as between consecutive fixations. Psilocybin also reduced the variability in these two metrics, while decreasing the entropy of the fixation probability distribution, which is consistent with narrower spatial distribution of attention under the high dose condition. Finally, we found that psilocybin induced a decorrelation between the consecutive fixation distance and the emotional response dimension of the aesthetic experience, as measured with the AEQ. With regards to the aesthetic experience itself, participants reported heightened emotional response and state of flow under the influence of the high dose relative to the control condition.

Ample anecdotal evidence suggests a link between the output of successful artists and the use of psilocybin, LSD and other serotonergic psychedelics, contributing to the emergence of the movement known as “psychedelic art”³⁴. However, few rigorous studies attempted to clarify the relationship between the acute effects of psychedelics, artistic perception and expression, and creativity. A recent naturalistic study addressed the effects of psychedelic microdosing on aesthetic perception¹² failing to find an effect of drug intake on feelings of awe elicited by paintings, thus partially contradicting anecdotal reports of a deeper and more profound aesthetic perception caused by these substances. Even though we did not explicitly address the construct of awe, we found increased scores of the emotional response dimension of the AEQ during the high dose condition. Interestingly, we did not find differences in the VAS item assessing the emotional valence of each individual painting. This discrepancy suggests that individuals may have experienced a heightened emotional response to the artworks overall in the high dose condition, however, this heightened emotional response may not necessarily translate into consistently more positive or negative evaluations of individual artworks, as reflected by the VAS scores. This interpretation implies that while the overall emotional experience may be intensified under the high dose condition, it may not significantly alter the valence of emotional responses to specific individual artworks.

We also found a marginally significant effect of psilocybin on the flow of experience dimension. Interestingly, previous work suggests some degree of overlap between the constructs of flow and absorption³⁵the latter being originally defined as “a disposition for having episodes of total attention that fully engage one’s representational resources”²⁰. It is known that trait absorption and the acute effects of psychedelics can exert mutual influences³⁶ and that absorption plays an important role in the processing of aesthetic stimuli from different modalities^{37,38,39,40,41,42}. However, the lack of studies using eye tracking hinders a direct comparison between these studies and our findings.

Our results show a discrepancy in how participants evaluate their general predisposition towards aesthetic perception (indicated by the AEQ questionnaire) and how they perceive and judge the aesthetic content of specific stimuli (indicated by the VAS). Previous work has addressed the effects of psychedelics on affective states and feelings of flow, showing acute and long-term increases of positive emotions³⁸enhanced emotional response to music³⁹and a state characterized by heightened abstract flow⁴⁰. Thus, the ratings from the AEQ questionnaire may reflect general aspects of the psychedelic experience that are not specifically related to aesthetic perception.

Recent work from Aday and colleagues³³ investigated changes in aesthetic experiences elicited by ayahuasca, adopting an open label naturalistic design. Measured with the AEQ, participants reported increased levels of aesthetic experience in two follow-up sessions (one week and one month after the ayahuasca retreat). These changes occurred for all dimensions of the questionnaire, while in our study we only found differences for two facets. This discrepancy could be attributed to expectation effects in the study conducted by Aday et al., given the lack of a control condition, or due to the more intense and sustained nature of psychedelic use in that study, which consisted of retreats totaling between 2 and 7 ayahuasca sessions. It is possible that the use of an active control in our experiments could have resulted in alterations to aesthetic perception, thus reducing the difference between conditions. However, van Elk and colleagues did not find any significant effect of a comparable dose of psilocybin⁴¹suggesting that our results were related to the effects of the high dose. Another factor to be considered is a possible effect of the setting on aesthetic perception, as natural environments could favor aesthetic experiences more than urban contexts. In contrast to the experiment by Aday et al., we administered the AEQ during the acute effects of the drug and we did not investigate whether the psilocybin resulted in long-term modifications. This choice was also motivated by the possibility of false beliefs during the acute effects⁴² and by the potential interference with the formation of memories⁴³which could be triggered by the drug and thus diminish the confidence in the questionnaire scores. Finally, neither our experiment nor the one conducted by Aday and colleagues revealed significant correlations between acute effects (e.g. mystical type experiences) and the changes in the AEQ scores. Further research is required to understand how aesthetic perception relates to other idiosyncratic aspects of the psychedelic experience.

The reported changes in fixation metrics could stem from the effects of psilocybin on low-level visual perception, top-bottom attentional processes, or a combination of both. According to a model proposed by Chatterjee⁴⁴the first stage of aesthetic perception corresponds to the identification of lower-level variables (color, form, texture, etc.), which are subsequently integrated to form a holistic view, directing visual attention to specific areas of interest depending on visual composition and complexity. The well-established effects of psychedelics on low-level visual perception² could influence the first stage of this process, while the acute effects on attention could influence the second stage^3,13. Previous studies established that more global fixation distributions are characteristic of the judgment of aesthetic appeal, while more local distributions relate to the processing of visual complexity in the painting⁴⁵suggesting that our findings could stem from the interaction between psychedelics and the perception of low-level visual features in the stimuli. The positive correlation between $\:ds$ and the AEQ sub-scale assessing the emotional aspect of the aesthetic experience suggests that semantic content is gathered by exploring the painting more globally when the participant is more emotionally engaged with the painting. Conversely, the decorrelation between these variables observed for the high dose condition could indicate that the emotional involvement with the artwork, which is overall greater in this condition, does not affect the way it is explored, suggesting that the increase in emotional engagement is not directly related to the semantic content gathered during visual exploration. However, it must be noted that the immediate VAS scales administered after each painting included a rating of subjective aesthetic appeal, yet no significant effects of psilocybin were found on this variable. Also, the exclusive use of classical paintings as visual stimuli in our study raises the question of whether the observed effects on gaze dynamics are specific to aesthetic experiences or reflect a more general alteration in visual exploration under psilocybin. Future studies could consider including non-artistic stimuli (e.g., natural scenes, geometric patterns, or everyday objects) to determine whether the reduction in fixation entropy and increased local exploration observed under psilocybin are specific to the aesthetic context or represent a broader change in visual attention mechanisms.

Although we attempted to minimize prior familiarity by selecting paintings from a curated set that aimed to avoid overly recognizable artworks, some of the selected paintings may still have been well known to participants. Familiarity with a given artwork could influence patterns of gaze or subjective evaluation. We did not systematically assess participants’ prior knowledge of the stimuli, which constitutes a limitation of the study. Nevertheless, since the same stimuli were used for both conditions of the experiment, it is unlikely that familiarity affected the results of the high vs. low dose comparison. Also, familiarity with the paintings is expected to increase in the second condition of the experiment; however, we did not find any effects of the order in which conditions were completed by participants.

According to the REBUS theory proposed by Carhart-Harris and Friston, psychedelics act by relaxing the precision of high-level priors or beliefs, thereby liberating bottom-up information flow¹³. In visual perception, this relaxation could manifest as a more entropic exploration of the visual scene, as the priors fail to inform gaze direction to regions where most of the relevant semantic information is represented. To date, this hypothesis has received evidence from multiple experiments, including complex tasks such as the production of natural speech^13,46. In contrast, our findings indicate a less entropic and more confined distribution of fixations. The liberation of bottom-up information flow could increase demands on cognitive processing, resulting in a slower and/or more local exploration of the visual scene. The relaxed priors might also lead to fixations failing to target relevant regions of the stimuli, increasing attention to low-level visual features and causing an exaggerated interest in this aspect of the paintings at the expense of a more global appreciation⁴⁵. In “The Doors of Perception”, Aldous Huxley famously narrated how his attention was lost at seemingly insignificant details of the visual scene: “I looked down by chance, and went on passionately staring by choice, at my own crossed legs. Those folds in the trousers—what a labyrinth of endlessly significant complexity! And the texture of the grey flannel—how rich, how deeply, mysteriously sumptuous!”⁴⁷. Such mode of perception seems compatible with the more local exploration revealed by our analysis of the eye tracking data.

The interpretation of our results faces some limitations, mainly due to specific choices concerning the experimental design. In this study, we prioritized the inclusion of tasks which are either natural and/or spontaneous (e.g. production of language), or capable of engaging the participants for relatively long periods of time, as is the case for the visual perception and the aesthetic appreciation of artworks. We also opted to pursue this experiment under conditions that are less constrained and more natural than those of a typical lab-based study. The merits and drawbacks of this approach have been discussed with detail elsewhere^3,29,50, yet here it is important to highlight that distractions during an eye tracking task can easily introduce a large amount of missing data points, potentially rendering the gathered data useless for further analysis. This issue could be expected to become more problematic in settings where subjects can develop feelings of fear and/or anxiety, including research laboratories and hospital settings³. Another characteristic of our study is the inclusion of a large variety of paintings, as opposed to presenting a reduced repertoire for a longer duration. This choice could have limited the processing of the stimuli to that of low-level features, as discussed above. We decided to present paintings for comparatively short durations to maintain the attention and engagement of the participants. A potential limitation could arise due to the large unblinding rate of the conditions, compatible with previous studies⁴¹ even when adopting an active condition as control (i.e. the low dose). While it is impossible to rule out a contribution of placebo effects to the eye tracking metrics, previous studies suggest that self-reported measures are typically more sensitive to placebo effects than behavioral or physiological measures⁴⁸. Another limitation to the generalizability of our findings is the predominantly male group of participants, which may represent a source of bias as gender could influence aesthetic perception. Finally, a major limitation of our study is its exploratory nature and the lack of pre-registration of the experimental design, hypothesis and analysis methodology. In this regard, we consider that exploratory studies are necessary when previous literature is scarce - in this case, due to the lack of reports addressing the sensitivity of eye tracking to capture behavioral changes elicited by psychedelics⁴⁹. A drawback of exploratory studies is the increased number of degrees of freedom, which can undermine the statistical significance of the findings. However, we attempted to minimize the degrees of freedom by focusing only on the basic statistical descriptors of the eye movement time series, and on the entropy of the fixation probability distribution, motivated by the entropic brain hypothesis and other related theories of psychedelic action^13,50. Future work by our group and others should build upon this first effort, capitalizing the present results for hypothesis generation in a pre-registered study.

In conclusion, we achieved a quantitative description of fixation statistics during aesthetic perception and their modulation by mushrooms of the Psilocybe genus, while also shedding light on the interaction between acute psychedelic effects and the aesthetic experience elicited by a selection of paintings. Guided by theoretical efforts positing a relaxation of priors during the acute effects of serotonergic psychedelics, we first hypothesized a more entropic distribution of fixations. However, we found the opposite result, which could be compatible with the idiosyncratic effects of psilocybin on the perception of low-level visual information. Future studies are required to address this and other possibilities raised by our work, a first attempt to quantify the acute effects of psychedelics on the complex eye movement behaviors underlying visual perception.