Effects of chronic intranasal oxytocin on visual attention to faces vs. natural scenes in older adults

Abstract

Aging is associated with changes in face processing, including desensitization to face cues like gaze direction and an attentional preference to faces with positive over negative emotional valence. A parallel line of research has shown that acute administration of oxytocin (OT) increases visual attention to social stimuli such as human faces. The current study examined effects of chronic OT administration among older adults on fixation duration to faces that varied in emotional expression, gaze direction, age, and sex. One hundred and twelve generally healthy older adults (aged 55–95 years) underwent a randomized, placebo-controlled, double-blind, between-subject clinical trial in which they self-administered either OT or placebo (P) intranasally twice a day for 4 weeks. The behavioral task involved rating the trustworthiness of faces (i.e., social stimuli) and natural scenes (i.e., non-social control stimuli) during eye tracking and was conducted before and after the intervention. Fixation duration to both the faces and the natural scenes declined from pre- to post-intervention, however this decline was less pronounced among older adults in the OT compared to the P group for faces but not scenes. Further, face cues (emotional expression, gaze direction, age, sex) did not moderate the treatment effect. This study provides first evidence that chronic intranasal OT maintains salience of social cues over time in older adults, perhaps buffering effects of habituation. These findings enhance understanding of OT effects on social cognition among older adults, and would benefit from follow up with a young adult comparison group to directly speak to specificity of observed effects to older adults and reflection of the aging process.

1. Introduction

The ability to read face cues is advantageous as it aids in the interpretation of the physical, mental, and emotional states of others. The face simultaneously integrates a rich supply of social information (Adolphs and Tusche, 2017), and cues found in faces such as emotion, gaze direction, age, and sex routinely guide social judgements (Pehlivanoglu et al., 2023).

Importantly, there are age-related changes in face processing. For instance, the ability to recognize faces declines in older adulthood (Martschuk and Sporer, 2018), and older adults are worse than young adults at identifying facial emotions (Gonçalves et al., 2018); an effect that has been found for still (Campbell et al., 2017) as well as dynamic (Grainger et al., 2017) stimuli. Age-related shifts in face processing were further modulated by cues such as emotional valence, gaze direction, and face age. For example, older adults generally exhibit a positivity bias when processing face emotion, evidenced by a positively skewed interpretation of ambiguous emotional face expressions (Petro et al., 2021) and visuo-attentional preference directed towards happy and away from angry faces (Isaacowitz et al., 2006). In addition, there is neural and behavioral evidence of age-related desensitization to eye gaze cues, such as for angry faces with direct gaze (Slessor et al., 2010; Ziaei et al., 2016), as well as improved emotion identification for direct-gazing older relative to young faces (Campbell et al., 2017), suggesting complex interactions of multiple biological/social face cues on face processing and subsequent social judgements in aging.

Growing evidence suggests that social cognition—including face processing—may be subject to individual differences in hormone levels (Puglia et al., 2015; Shamay-Tsoory and Abu-Akel, 2016) and can be enhanced via administration of biologically relevant compounds such as oxytocin (OT). OT is a naturally occurring neuropeptide that can be administered via nasal spray directly targeting central functions (Guastella et al., 2012) and is involved in the processing of socioemotional cues including faces (see Horta et al., 2020a). Single-dose (acute) administration of OT in young adults has been shown to improve facial recognition (Rimmele et al., 2009), emotion identification (Shahrestani et al., 2013), as well as attune visual attention to the eyes (Guastella et al., 2008). The effects of exogenous OT are socially selective, with studies demonstrating that OT increased recognition of human faces over non-social stimuli (Rimmele et al., 2009) and biased attention to emotional faces but not non-social stimuli (Xu et al., 2019). The social salience hypothesis posits that socioemotional effects of OT are context-dependent and contingent on individual factors such as age, personality, or sex (Shamay-Tsoory and Abu-Akel, 2016). In particular, OT may modulate overt and covert attention in early stages of attentional processing, thereby producing greater social salience and subsequent impact on socioemotional behavior (Pehlivanoglu et al., 2020). This potential function of OT is in line with findings of increased visuo-attentional processing of eyes following OT administration (Guastella et al., 2008).

In fact, older adults may particularly benefit from OT intervention due to age-related changes in processing face cues; however, the extant literature is limited regarding OT effects in older adults (Horta et al., 2020b). One of the few exceptions is Campbell et al. (2014) who found that older men, relative to older women and young adults, showed enhanced emotion identification abilities after intranasal OT relative to placebo (P) administration, supporting an age-by-sex interaction of OT effects on face cue processing (but see Grainger et al., 2019, Horta et al., 2019, and Lin et al., 2022 for non-significant age, or age-by-sex effects). Notably, these previous studies used a single-dose design, which is pervasive in the OT literature. Although single-dose studies have advanced scientific knowledge by demonstrating acute effects of OT on socioemotional functioning, the clinical relevance and practical use of exogenous OT are largely unknown–particularly for older adults–and must be informed by examining effects of OT administration over time (Horta et al., 2020a). Chronic (i.e., repeated) OT administration has been demonstrated generally safe and tolerable in older adults (Rung et al., 2021; Wright et al., 2023) and associated with reduced decline in physical function and fatigue compared to P after ten consecutive days of intranasal administration in older adults (Barraza et al., 2013).

In the present study, generally healthy older men and women self-administered intranasal OT over 4 weeks to determine the extent to which chronic OT modulated visual attention to human faces that varied in emotional expression, gaze direction, age, and sex. In line with the social salience hypothesis (Shamay-Tsoory and Abu-Akel, 2016), we hypothesized that chronic intranasal OT, relative to P, would enhance fixation duration to faces (i.e., social stimuli) over natural scenes (i.e., non-social control stimuli) from pre- to post-intervention. We also explored the extent to which these effects may be modulated by biological/social face cues (i.e., face emotion, gaze, age, and sex).

2. Method

2.1. Participants

The data were taken from a larger clinical trial (NCT02069431) on the effects of chronic intranasal OT administration on physical, cognitive, and socioemotional function in older adults (see also Rung et al., 2021; Polk et al., 2022; Valdes-Hernandez et al., 2021; Wright et al., 2023). Participants were recruited between February 2016 and February 2020 on a rolling basis via university participant registries and throughout the community in North Central Florida. Male and female participants were included if they were over 55 years old, willing and able to provide informed consent, safe to undergo OT administration according to the lab-internal Health Demographics Screener (e.g., no severe nasal obstruction, hypersensitivity to OT or vasopressin, history of hyponatremia, inappropriate antidiuretic hormone secretion, psychogenic polydipsia, low sodium (< 134 mEq/L) and high urine osmolality (> 1200 L), currently taking vasoconstrictors or antidiuretic medication), scored ≥ 30 points on the Telephone Interview for Cognitive Status (TICS; Brandt et al., 1988), and had a blood pressure under 180/100 mm Hg. Exclusion criteria included heavy smoking and/or drinking, serious brain damage, severe glaucoma, macular degeneration, and cataract (for more detail, see Rung et al., 2021). The current report examined data from participants with valid eye tracking in the Direct Averted Gaze Task at both pre- and post-intervention, resulting in a sample of 112 older adults (M = 71.07 years; SD = 7.67 years; 34% female; 88.39% white; see Table 1 for sample description). We conducted sensitivity analysis in G*Power (Faul et al., 2009), indicating that with the current sample size of N = 112 and p =.05 as the Type I error threshold, we were powered at 95% to detect a small effect (i.e., Cohen’s f² = 0.14) for the primary hypothesis (i.e., interaction between Treatment (between-subject), Timepoint (within-subject), and Stimulus Type (within-subject) on fixation duration (for a similar approach, see Pehlivanoglu et al., 2023).

Table 1.

Sample-descriptive characteristics and inference statistics for treatment group differences in demographics, health, and cognition at pre-intervention (N = 112).

	Oxytocin (N = 62) M (SD)/%	Placebo (N = 50) M (SD)/%	p-values
Demographics
Age	72.05 (7.61)	69.86 (7.65)	0.13
Education	16.54 (2.80)	16.06 (4.35)	0.09
Sex	35% female	32% female	0.85
Health
Physical	8.10 (1.20)	8.22 (1.20)	0.61
Mental	8.61 (1.18)	8.56 (1.03)	0.59
Vision	6.55 (2.16)	6.70 (2.22)	0.79
Cognition
Crystalized Composite	118.84 (12.14)	117.89 (10.22)	1.00
Fluid Composite	88.90 (9.16)	90.24 (9.61)	0.37

Note: Age and Education were indicated in years. Years of education were missing for two participants in the oxytocin (OT) group and seven participants in the placebo (P) group. Cognition was assessed via the NIH toolbox (Akshoomoff et al., 2013) and represented by the mean unadjusted scores; cognition scores were missing for two participants in the OT group. Subjective physical health (Please rate your general physical health), mental health (Please rate your general mental health/mood), and vision (Please rate your vision (without contacts or glasses if used regularly)) were assessed using one item each with a response scale ranging from 1 = poor to 10 = excellent. The p-values were computed using independent samples t-test for normally distributed demographics (i.e., age), Wilcoxon signed-rank test in case of non-normality (i.e., education, physical health, mental health, vision, and crystalized and fluid cognition), and chi-square test for the categorical variable of sex, using treatment group as the independent variable for all tests.

2.2. Study procedure

All procedures were approved by the University of Florida Institutional Review Board (Protocol #: 201300801). To ensure eligibility, participants completed a pre-screening phone call prior to enrollment, followed by an in-person screening visit during which all participants provided written informed consent. The clinical trial consisted of one screening visit, three pre-intervention visits, a 4-week intervention, and three post-intervention visits that mirrored the corresponding pre-intervention visit (see Rung et al., 2021 for details and a session flow-chart). During the 4-week intervention, participants were instructed to self-administer the nasal spray to achieve a total dosage of 24 international units (IU) of either OT or P twice daily (once in the morning: 7–9AM; once in the evening: 5–7PM) by dispensing one puff containing 12 IUs in each nostril, for a daily dose of 48 IUs over the course of 4 weeks. The OT and P sprays (IND 100,860) were identical in every respect, except P did not include the active OT ingredient; the compounding, randomization, and distribution process were directed by the Investigational Drug Service at the University of Florida (Rung et al., 2021). Post-intervention visits occurred during the final week of the intervention, and participants were instructed to abstain from self-administering the spray the morning of the session to avoid acute administration effects.

Participants were debriefed and compensated upon study completion. Data under consideration in this study were collected in the second pre- and post-intervention visits as described next.

2.3. Measures

2.3.1. The Direct Averted Gaze Task

2.3.1.1. Task procedure.

The Direct Averted Gaze Task was developed for this study and completed by all participants during both their pre- and post-intervention visit. The task comprised four runs in which participants rated the trustworthiness of each face or natural scene presented on the computer screen while their eye movements were tracked. Fig. 1 depicts sample stimuli and the trial design. In each run, participants viewed and rated 72 faces and 12 scenes on a 4-point scale from Not at all trustworthy to Very trustworthy. Each trial began with a jittered fixation cross with a mean duration of 500 ms at the center of the screen. The stimulus was then presented in the upper center of the screen for 2000 ms before the rating scale appeared below the picture for 1500 ms, prompting participants to rate the trustworthiness of the face or scene. The entire task lasted 40 minutes.

Fig. 1.

Sample stimuli and trial design. A) Each face identity was presented once in each emotion-by-gaze direction combination. B) At the start of each trial, participants focused on the fixation cross for an average jittered duration of 500 ms; then they viewed a face for 2000 ms, after which a rating scale appeared for 1500 ms below the face image; participants rated the trustworthiness of each face (or scene, in the control condition) by pressing the corresponding button on the joystick which matched the location of the answer choices on the screen. Photos used with permission.

Eye movements were recorded via an SR EyeLink 1000 Plus eye tracker (SR Research Ltd., Canada) with a temporal resolution of 2000 Hz and a spatial resolution of 0.02°. Participants removed bifocals or trifocals as well as any eye makeup before undergoing eye tracking and used a mounted chin rest to reduce head motion during the task. Standard 9-point calibration was conducted prior to the start of each of the four task runs. If worse than “fair,” calibration was repeated at 5-point, and at 3-point if necessary.

2.3.1.2. Stimuli.

For the face stimuli, 72 distinct face identities were selected from the FACES database (Ebner et al., 2010), with equal numbers of young-male, young-female, old-male, and old-female faces (i.e., 18 in each category). The FACES database is a set of naturalistic face images that have been validated in terms of emotional expression, perceived age, attractiveness, distinctiveness, and trustworthiness (Ebner et al., 2018; Pehlivanoglu et al., 2023). Each face identify was presented four times across the task, once in each of the four task runs, with angry or happy expressions paired with direct or averted eye gaze in equal numbers (Fig. 1A; see Ziaei et al., 2016) for a fully counter-balanced design. The presentation order of faces in each run was pseudorandomized to ensure that the same emotion expression or eye gaze direction did not occur consecutively more than three times. The same faces were presented at pre- and post-intervention, but in reversed order, to control for order effects.

Scene stimuli were selected from Johnson et al. (2007). The scenes were photos of natural landscapes, such as forests or bodies of water (with no humans or human construction, like houses), to assure they served as non-social control. The scenes were not specifically manipulated to convey positive or negative atmospheres. Twelve unique scenes were randomly intermixed with 72 face images in each of the four runs. Participants thus viewed a total of 288 faces and 48 scenes over the course of the task.

2.4. Analysis

2.4.1. Preprocessing eye tracking data

Preprocessing of the raw eye tracking data was conducted in EyeLink Data Viewer (SR Research Ltd., version 4.2.1). Areas of interest (AOIs) were defined for the faces (i.e., an oval shape beginning slightly above the hairline, extending slightly below the chin and a little past the edges of the face stimuli) and the scenes (i.e., a rectangular shape spanning the full image). After AOIs were specified, fixation duration was extracted from each trial for each participant. Eye tracking from the 2000 ms period when the stimulus was presented without the rating scale was used for analysis (Fig. 1B). A fixation was defined as sustained gaze in an AOI occurring with 1° of visual angle and lasting for a minimum of 100 ms (Isaacowitz et al., 2006); fixation duration was defined as the sum of the length of fixations occurring within each AOI per trial.

2.4.2. Statistical analysis

Using Stata 17 (StataCorp, 2021), multilevel random intercept models were used to account for the nested structure of the data. Specifically, we conducted cross-random effects analysis with cross--classification of participants and stimuli and a nested structure for repeated observations within participants. This approach allowed eye movements from the same participant to be correlated across different face identities and accounted for dependencies of eye movements of different participants viewing the same face identity (for a similar approach, see Pehlivanoglu et al., 2023).

To address our first hypothesis examining whether chronic intranasal OT compared to P would differentially impact fixation duration to faces (social) over natural scenes (non-social control) at post- relative to pre-intervention, we tested for a three-way interaction between Treatment (Between-subject: OT vs. P), Stimulus Type (Within-subject: face vs. scene), and Timepoint (Within-subject: pre-intervention vs. post-intervention) on fixation duration. A separate multilevel model further explored whether Face Emotion (Within-subject: happy vs. angry), Face Gaze (Within-subject: averted vs. direct), Face Age (Within-subject: young vs. old), and Face Sex (Within-subject: male vs. female) moderated a Treatment by Timepoint interaction on fixation duration to the face.

As non-hypothesis-driven analysis, we also ran two multilevel models with identical structure to the models described above on perceived trustworthiness (i.e., participants’ subjective ratings of faces and scenes, with higher scores indicative of greater perceived trustworthiness) as outcome variable. Data from 121 participants (M = 71.10 years; SD = 7.66 years; 29.75% female; 50.41% OT group) with trustworthiness ratings from both pre- and post-intervention were considered in this analysis.

Significant interactions were interpreted via z tests for pairwise comparisons and corrected for multiple comparisons (Bonferroni). All results remained the same when participant age and sex were entered as covariates in all models. Researchers remained blinded to treatment group assignment until analyses were completed.

3. Results

The first model resulted in a significant three-way interaction between Treatment, Stimulus Type, and Timepoint on fixation duration (χ²₍₁₎ = 4.32, p =.04; Fig. 2). As depicted in Fig. 2, both treatment groups decreased in fixation duration to faces and scenes from pre- to post-intervention. However, in line with our hypothesis, this decline in fixation duration to faces was more pronounced in the P (z = −15.11, p <.001) than the OT (z = −7.14, p <.001) group, while for scenes, fixation duration from pre- to post-intervention declined comparably for both the OT (z = −4.40, p <.001) and the P (z = −4.48, p <.001) group. Treatment differences in fixation duration to faces or scenes were not significant at either pre-intervention (z < 1.37, p >.68) or post-intervention (z < 1.66, p >.39). Also, the covariate participant age was not significant (z = 1.77, p =.08), but participant sex was significant with males exhibiting longer fixation durations overall than females (z = 2.04, p =.04).

Fig. 2.

The interaction between Treatment, Stimulus Type, and Timepoint was significant (p =.04). Individuals in the oxytocin (OT; black line) and placebo (P; gray line) groups declined in fixation duration from pre- to post-intervention; for the face stimuli, this decline was more pronounced among the P (p <.001) than the OT (p <.001) group, while for the scenes fixation duration declined comparably for the OT and the P group. Note: the full range of the y-axis is 0 – 2000 ms, and error bars reflect standard errors.

Our second model showed that the Treatment by Timepoint interaction was not moderated by Face Emotion, Face Gaze, Face Age, or Face Sex (χ²s < 1.87, ps >.17). The covariates of participant age and participant sex were not significant in this model (zs < 1.90, ps >.06).

For the additional, non-hypothesis-testing analysis, the Treatment, Stimulus Type, and Timepoint interaction on perceived trustworthiness was not significant (χ²₍₁₎ = 1.47, p =.23). Further, the Treatment, Timepoint, and Face Emotion, Face Gaze, Face Age, or Face Sex on perceived facial trustworthiness also did not reach significance (χ²s < 3.00, ps >.08).

4. Discussion

Older adults experience declines in socioemotional functioning, including the processing and integration of face cues which are vital for successful social interactions (Slessor et al., 2010; Ziaei et al., 2016). This study investigated the extent to which chronic intranasal OT administration enhanced visual attention to faces, over natural scenes, in generally healthy older adults. The findings were in line with our study prediction that chronic OT administration would benefit face processing among older adults in that administration of OT relative to P ameliorated the decline in fixation duration to faces, relative to natural scenes, from pre- to post-intervention. This finding corroborates and extends previous research that OT effects are specific to socially salient information (Pfundmair et al., 2017; Shamay-Tsoory and Abu-Akel, 2016; Xu et al., 2019). Further, face cues of emotion, gaze direction, age, and sex did not moderate these effects. Results from this study suggest that chronic intranasal OT may heighten the social salience of faces, independent of varying social/biological face cues, over non-social scenes in older adults.

As noted, we observed a significant decline in fixation duration to both faces and scenes from pre- to post-intervention for both the OT and the P groups. One speculation regarding the mechanisms underlying this decline in visual processing may be habituation to, or familiarity with, the task upon repeated exposure (Rankin et al., 2009). Importantly, however, our effect was stronger in the P than the OT group for the faces, and was not present for the natural scenes, suggesting that OT may have selectively attenuated habituation to social cues in our older adults. This finding is in line with fMRI data from the same clinical trial demonstrating that chronic OT administration ameliorated decrease in functional connectivity to social stimuli in an animacy perception task after repeated exposure (Valdes-Hernandez et al., 2021). This pattern of findings aligns with the social salience hypothesis (Shamay-Tsoory and Abu-Akel, 2016) that effects of exogenous OT on social cognition are due to OT regulating social cue salience. That is, 4-week OT administration may have maintained the salience of faces over time, partially countering decline in visual attention due to habituation; keeping older adults more alert to relevant social cues.

In contrast, fixation duration to the scenes declined from pre- to post-intervention comparably for individuals in the OT and the P group. This finding aligns with previous literature demonstrating that OT administration selectively facilitates social over non-social processes, including improved memory (Rimmele et al., 2009) and longer fixation duration (Pfundmair et al., 2017) for social than non-social information. Our finding further extends previous evidence to chronic oxytocin administration and older adults, a developmental population often understudied when investigating the effects of OT on socioemotional functioning but that may particularly benefit from OT’s treatment potential (Horta et al., 2020b; Huffmeijer et al., 2012).

Regarding our second aim, we did not find support for a moderation of face emotion, gaze direction, face age, or face sex on chronic OT effects on fixation duration in older adults despite the social and biological relevance of these face cues (Adolphs and Tusche, 2017; Pehlivanoglu et al., 2023). Previous research has shown that acute OT administration enhances attention to subtle face cues that promote social interaction, such as positive emotional expression and greater attention to the eyes (Domes et al., 2013; Guastella et al., 2008). The present study is the first to examine visual attention to face cues in the context of a chronic OT regimen, similar to a treatment approach (Horta et al., 2020a) and in older adults. It is possible that acute administration, leading to rapid increase in OT concentrations (Gossen et al., 2012), attunes attention to even subtle socially relevant cues (e.g., gaze direction) whereas a chronic approach may not have the same psycho-pharmacodynamic effects. Although studies have not yet investigated how acute compared to chronic OT may differentially impact processing of social cues like faces, there is evidence of dose-dependency on OT’s effects on socioemotional functioning more broadly. For example, preclinical studies have shown an increase in social behavior after acute intranasal OT and a decrease after chronic intranasal OT in mice (Huang et al., 2014). Similarly, Bales et al. (2013) found that acute intranasal OT in voles increased social behaviors directed towards familiar partners while chronic OT resulted in deficits in partner preference formation. Further, in humans, chronic OT administered every other day, relative to every day, for five days resulted in reduced amygdala reactivity to fearful faces as well as decreased arousal and intensity ratings, possibly reflective of dose-dependent reduced anxiolytic effects (Kou et al., 2022). Further research is necessary to determine ideal dosage and administration regimen to produce beneficial social effects (Horta et al., 2020a); systematically investigate the extent to which visual processing of faces varies across acute vs. chronic OT administration; and across adulthood and aging, e.g., via including a young adult comparison group.

Moreover, environmental (e.g., task design, use of naturalistic vs. computer-generated social stimuli; Auyeung et al., 2015; Grainger et al., 2017) and individual (e.g., age, sex; Campbell et al., 2014) factors have been shown to impact the effectiveness of OT administration on socioemotional functioning. An emerging literature supports a critical role of endogenous OT related levels and processes for understanding exogenous OT effects on social cognition and social behavior (Chen et al., 2020). For example, OT plasma concentration, oxytocin receptor gene (OXTR) polymorphisms, or OXTR methylation may benefit from OT administration (Shamay-Tsoory and Abu-Akel, 2016). In particular, OXTR methylation at several ‘5-Cytosine-phosphate-Guanine-3’ (CpG) sites in the OXTR promotor region MT2 has been inversely linked to OT gene expression (Kusui et al., 2001). Greater OXTR methylation has been associated with impairments in social cue processing such as emotional expression (Puglia et al., 2015); and higher levels of OXTR methylation have been linked to psychiatric and neurodevelopmental disorders (Siu et al., 2021). Future research should examine the role of the endogenous OT system on chronic intranasal OT effects on visual attention to social cues in aging.

4.1. Limitations and future research

The present study has some limitations. First, we did not include a young adult comparison group and therefore cannot conclude that the effects found in this study are specific to older adults and reflective of the aging process. In addition, the small sample size precluded examination of sex differences despite evidence of sex-dimorphism in research on OT and social cognition (Caldwell, 2018; Campbell et al., 2014).

A fruitful avenue for future research would be to go above behavioral assessment and capture the neurocognitive processes underlying chronic OT administration effects on face processing via fMRI or ERP in combination with eye tracking; these approaches would allow insight into mechanisms of brain action and temporal dynamics of OT in social cognition (e.g., perception, selection, and evaluation; Pehlivanoglu et al., 2020). Inclusion of objective physiological measures such as skin conductance response could further inform the interpretation that OT attenuates habituation to faces.

Also, future research should investigate whether chronic OT-enhanced sustained fixation duration to faces is associated with better social cognition, supporting OT’s therapeutic potential. Finally, innovative study designs that use dynamic (Grainger et al., 2017) and interactive (i.e., virtual social environment with interactive avatars) social paradigms will shed more light on OT’s impact on real-world social-cognitive phenomena among older adults.

4.2. Conclusion

In sum, this paper investigated whether chronic administration of OT enhanced visual processing to faces as prominent social cues (relative to natural scenes), using cues that systematically varied face emotion, gaze direction, age, and sex. The study went beyond previous work by employing a chronic OT administration design and in a sample of generally healthy older adults, a developmental group currently understudied in this line of research despite well-known age-related changes in socioemotional behavior. Results support that 4-week intranasal OT ameliorated decline in visual attention to faces, while not natural scenes, regardless of face cues, in line with the notion of the role of OT in gating social salience.

Data statement

All data and code for the analyses associated with this paper can be located on the Open Science Framework at this link: https://osf.io/cepux/?view_only=15ba25026e8e4ef898561092a3945fb4