Effects of oxytocin on emotion recognition in schizophrenia, a randomized double-blind pilot study

Abstract

Background:

Schizophrenia (SCZ) is a neurodevelopmental disorder that leads to poor social function. Oxytocin (OXT), a neuropeptide involved in social cognition, is a potential therapeutic agent for alleviating social dysfunction. Therefore, we investigated the effects of intranasal oxytocin (IN-OXT) on emotional processes in experimental interactive social contexts in individuals with SCZ.

Methods:

In a male-only parallel randomized placebo-controlled double-blind trial we investigated the effects of IN-OXT (24IU) on visual fixation on pictures of faces and emotion recognition in an interactive ball-tossing game that probed processing of social and non-social stimuli.

Results:

IN-OXT enhanced the recognition of emotions during an emotion-based ball-tossing game. This improvement was specific to the game that included social cue processing. IN-OXT did not affect eye gaze duration or gaze dwell time on faces in these patients.

Conclusions:

An acute low dose of IN-OXT had a modest effect on social cue processing and was limited to emotion recognition. Higher doses and long-term trials targeting emotional processing in SCZ may lead to improved social function.

Introduction

Schizophrenia (SCZ) is a mental disorder characterized by both positive symptoms (disorganized behavior, delusions, hallucinations, etc.) and negative symptoms (lack of motivation, anhedonia, social dysfunction, affective flattening, poor occupational function, etc.)¹. Patients with SCZ suffer from deficits in theory of mind, social perception and emotional processing². These deficits in emotional processing are correlated with decreased neural activity in the amygdala, hippocampus, fusiform gyrus, and frontal and visual associative cortices during facial emotion identification^3,4. Emotion recognition provides a basis for understanding the intentions and feelings of others and is therefore a key component of social cognition. Difficulties identifying emotional expressions⁵ and using social cues to make appropriate judgments lead to poor functional outcomes, unemployment and social isolation⁶. While current medications can ameliorate positive symptoms, they, unfortunately, fail to improve the negative symptoms and social deficits commonly seen in SCZ⁷.

A large and rapidly growing literature indicates that the neuropeptide oxytocin (OXT) plays an important role in social cognition in animals and humans⁸. In animals, OXT is essential for social recognition, maternal attachment, pair bond formation and empathy-based consolation^9–12. In humans, intranasal administration of oxytocin (IN-OXT) promotes context-dependent prosocial behaviors¹³, feelings of empathy^14–16, social memory^17,18, gaze toward the eye region¹⁹, and emotion recognition^20,21. Also, several studies in healthy controls have shown that the effects of IN-OXT on behavior are more pronounced in the presence of social contexts^22,23. Previous research of this neuropeptide in autism spectrum disorder has found that IN-OXT improves social cognition, increases visual scanning of faces, and increases neural activity in brain regions involved in social salience and social reward^24–29.

The effects of IN-OXT on patients with SCZ has also been investigated with inconsistent results. Recent meta-analyses found that IN-OXT to patients with SCZ was not effective for treating positive, negative or general symptoms³⁰ and IN-OXT did not improve social cognition or neurocognition³¹. However, moderator analyses revealed that IN-OXT had larger effects on those with high-level social cognition compared to those with low-level social cognition in patients with SCZ³¹. In separate studies, acute intake of IN-OXT enhanced the capacity of patients with SCZ to make mental inferences about others’ intentions and judgments based on various scenarios of video clips of actors³² and to detect vocal intonations of affect³³. A recent randomized fMRI study has shown that IN-OXT (40IU) increases accuracy on theory of mind task and the BOLD activity in the temporo-parietal junction in SCZ³⁴.

Even with these findings, there is a lack of studies examining the effect of IN-OXT on visual fixation on facial stimuli in SCZ. There is also a lack of knowledge regarding the effects of IN-OXT on socio-emotional processes during interactive social contexts that mirror everyday life. Therefore, we hypothesize that IN-OXT will enhance visual fixation on facial stimuli, improve emotion recognition (better emotion discrimination) following social interactive contexts, and will not affect identification of non-social cues.

Materials and Methods

Participants

We completed a double-blind placebo-controlled parallel group study with a comparison healthy control group (HC) who did not receive OXT or placebo, and an experimental and placebo group of SCZ (IN-OXT and IN-PL). The study was limited to men due to the potential side effects of oxytocin in women, namely the risk of miscarriage in an undetected pregnancy. The study was conducted at the Atlanta Veterans Affairs Health Care System (AVAHCS) in Decatur, Georgia. It was approved by the Emory University Institutional Review Board, the Atlanta VA Research and Development Committee, and by the Food and Drug Administration. All subjects read and signed an informed consent form and gave signed HIPAA authorization prior to participation in the study. The study was registered at clinicaltrials.gov (NCT01568528).

HCs were enrolled if they were between 18–65 years old, male, were not taking any psychotropic medication, and did not have a history of a psychiatric disorder as confirmed by the SCID-I³⁵.

Patients in the SCZ group also had to be between 18–65 years old, male, have a diagnosis of SCZ or schizoaffective disorder as confirmed by the SCID-I and were required to not have a history of bipolar disorder. They also needed to have a primary deficit syndrome as validated by the Kirkpatrick Schedule for the Deficit Syndrome³⁶, be psychiatrically and medically stable for eight weeks prior to informed consent, maintained on stable treatment with antipsychotic medication and/or other concomitant psychotropic treatment for at least six weeks prior to consent, have no more than a moderate severity rating (≤ 4) on hallucinations and unusual thought content on the Positive and Negative Symptoms Scale (PANSS³⁷), and have normal blood chemistry. SCZ patients were excluded if they were too depressed (score greater than 10) as measured by the Calgary Depression Scale (CDSS³⁸), were displaying Parkinsonian side effects to antipsychotic medications as measured by the Simpson-Angus scale (score greater than 6), had been hospitalized in the eight weeks prior to consent, had homicidal ideation in the previous six months, or had active suicidal ideation with a specific plan and intent or demonstrated suicidal behavior within one month of consent as determined by the Columbia-Suicide Severity Rating Scale C-SSRS³⁹.

Subjects (HC or SCZ) were excluded if they had active substance dependence within the prior thirty days (cigarette smoking was allowed) as assessed by urine toxicology screen, inadequate visual acuity as assessed by an eye chart, a history of mental retardation or pervasive developmental disorder, a history of neurological disorder (such as traumatic brain injury, seizure disorder, Parkinson’s Disease, dementia), a loss of consciousness for more than ten minutes due to head trauma, a known HIV infection or AIDS, or treatment with a benzodiazepine in the two weeks prior to consent.

Cognitive Assessment

We administered the Measurement and Treatment Research to Improve Cognition in Schizophrenia Consensus Cognitive Battery (MCCB) to all participants prior to administration of IN-OXT or IN-PL. The MCCB is a consensus battery of neurocognitive tests assessing seven key cognitive domains relevant to SCZ: processing speed, attention/vigilance, working memory, verbal learning, visual learning, reasoning and problem solving, and social cognition⁴⁰. The composite and domain scores generated from the MCCB are all T-scores adjusted for sex and age.

Study design

The IN-OXT group (n=11) received intranasal Syntocinon Spray (Novartis, 24IU) whereas the IN-PL group (n=9) received intranasal placebo in a double-blind placebo-controlled parallel group design with an allocation ratio of 1:1. The Syntocinon Spray was imported from the Internationale Apotheke pharmacy in Bern, Switzerland. The placebo consisted of nasal saline that was prepared locally at the research pharmacy service at Emory University. Recruitment and study procedures were completed during the period of 2013–2017.

HCs completed the test procedures on a single day without receiving intranasal oxytocin or placebo. HCs were included in the study to determine a baseline on study procedures compared to SCZ who did and did not receive OXT (IN-PL). SCZ subjects were required to attend two visits. The first visit included all screening procedures and completion of the MATRICS Consensus Cognitive Battery⁴⁰ to assess cognitive status in the verbal memory, visual memory, attention, reasoning and problem solving, and social cognition domains. The second visit included the administration of IN-OXT or IN-PL under close supervision by E.D. Treatment assignment was by random allocation in blocks of six. Both experimenters and subjects were blind to the treatment assignment. The IN-OXT dose was three 4 IU puffs per nostril for a total dose of 24 IU. This dose has been used in many similarly designed challenge studies examining the effects of a single dose of OT^18,19,25,41. Normal saline spray was delivered intranasally (IN-PL) in a similar manner to the IN-OXT group. All the behavioral tests were performed 45 minutes after administration of IN-OXT or IN-PL in a set order: eye tracking tasks followed by social interaction tasks.

Behavioral tasks

Facial Emotion Identification Test (FEIT)

We utilized the Facial Emotion Identification Test (FEIT) to assess emotion recognition in this study. The stimuli are 19 standard black and white pictures of faces showing one of six different emotions (happy, sad, angry, surprise, disgusted, ashamed). The pictures are shown for 15 sec, with 10 sec between each face. After the presentation of each face the participant is asked to choose which of the six emotions was displayed. The score on the test is the sum of correct responses⁴².

Visual scanning of faces:

All eye-tracking tasks were previously used in our work on oxytocin and autism spectrum disorders²⁵. Subjects were seated with their chin resting on a chin rest and their forehead resting against a bar to minimize head movements. The Applied Science Laboratories (ASL), Model 5000 Eye tracking system (Applied Science Laboratories, Bedord, MA) tracked and quantified the eye movements of subjects at a sampling rate of 60 Hz. Calibration of eye position was performed according to standard procedures recommended by ASL. The participants then performed a gender identification and gaze direction identification task while their eye movements were monitored.

Gender identification

Subjects were instructed to identify the gender of a face presented on the computer screen. The stimuli were 24 colored pictures of neutral faces (half males and half females) from the NimStim database⁴³. Each stimulus was presented on the screen for 2s, followed by two labels appearing below the same face (“male”/ “female”) that remained on the screen until the subject responded with one of the two keys on the button box to indicate their judgement of the sex of each person in the pictures.

Gaze direction identification

Subjects were instructed to identify the direction of gaze of the face presented on the screen. The stimuli consisted of 24 grayscale pictures of male faces (generously given by the Wicker B. database)⁴⁴. The facial stimuli have either averted eye gaze (half) or straight-ahead eye gaze (half). Each stimulus was presented for 2s, followed by a picture with the same face and two labels below the face (“Looking at me” / “Not looking at me”) that remained on the screen until the subject responded with one of the two keys on the button box.

All stimuli were presented on a computer screen and responses were recorded by a keypad. We collected the dwell time and the fixation count for subjects’ eye gaze on the face as a region of interest. The dwell time corresponds to the duration of fixations within the region of interest (face area). The fixation count is equal to the total number of fixations in a particular region of interest. We also collected the dwell time and fixation count on the eye region, but it did not yield significantly different results from the face region. Therefore, no further details about the eye region were provided in this manuscript.

Emotion recognition via a ball-tossing game

Emotion-based Social Ball-Tossing Game (ES-BTG)

We have developed a new version of the original cyber-ball game⁴³ that includes the presentation of different facial expressions of the players as a social feedback to subject’s actions throughout the game. Subjects were engaged in a computerized multi-round ball-tossing game with three other players. Players were depicted by a correspondent static picture that changed in terms of facial expression throughout the game. The subject received the ball from the other players. Upon catching the ball, the subject was instructed to throw the ball to others by selecting the appropriate keys on the keyboard (left, right, top). Subjects were told that these players are real players. In actuality, these players were fictitious, and their responses (actions and facial expressions) were scripted with specific algorithms. In this game, the subject had a probability of 1/3 of reception of the ball from each of the players throughout the entire game. However, we manipulated the facial expression that each of the players displayed during the game upon reception of the ball. The game consisted of three phases. During the first phase all three players displayed neutral expressions. In the second phase, two of the other players displayed emotional expressions upon receipt of the ball, either positive or negative, in 80% of the trials while one player throughout the second phase displayed neutral emotions during the whole game. In the third phase, two players displayed emotional expressions in 100% of the trials and the neutral player maintained a neutral expression throughout the entire game. We asked the subjects to answer written questions about the emotional expressions of each player before the game, after the second phase of the game (mid-point, early detection) and at the end of the third phase of game (end point, late detection). We added players’ pictures (neutral facial expressions) to each of the questions on the questionnaire to help the subject remember players’ face identity. Before the game, we asked the subject if each of the players’ faces (that should be neutral before the game) expressed any particular emotion (1 = not at all; 7 = very much). At mid- and end-points of the game, subjects were asked what expression, if any, each player expressed on his face (1 = negative, 4 = neutral, 7 = positive). These questions were designed to infer the subject’s capacity to recognize the emotional expressions of the fictitious players whose faces they saw during the ball game.

Valence-based non-Social Ball-Tossing Game (VnS-BTG)

This task aims to study the effects of IN-OXT on the identification of positive and negative valence of non-social stimuli presented as feedback upon receipt of the ball throughout a ball-tossing game. This game followed the same procedures as the ES-BTG except that the subject played with three computers and the feedback from these computers consisted of non-social stimuli. During the first phase, the three computers were depicted on the screen by three random pictures of geometric shapes. During the second phase of the game, two of these computers displayed positive (beach scenes) and negative (fire) valence pictures (International Affective Picture System, IAPS) upon receipt of the ball in 80% of the trials. The third computer was depicted by emotionally neutral pictures of random roads. During the third phase, emotional stimuli were displayed in 100% of the cases. During this session the participant was asked similar questions to those seen in the ES-BTG.

Statistical analysis

Statistical analyses were performed with SPSS v.24 (IBM, Armonk, NY). For the demographic comparison between the two groups of patients and healthy controls, we conducted univariate ANOVAs on age, race. To determine if there were any significant differences between the treatment groups (IN-OXT v IN-PL) we conducted univariate ANOVAs on PANSS total scores, PANSS negative, PANSS positive scores, and the severity of Deficit Syndrome individually.

The MCCB was initially analyzed in a univariate ANOVA split by diagnosis (SCZ vs. HC) to determine whether there were any significant differences in domains of cognition. A subsequent analysis was conducted between those who received IN-OXT compared to those who received IN-PL to determine whether there was an initial difference in the cognitive profile of the two treatment groups prior to administration of the drug.

For the face perception task we utilized two regions of interest, one on the entire face and one on just the eyes during the gender identification and gaze direction tasks. We performed univariate ANOVAs on dwell duration and fixation count individually as dependent measures and treatment group among patients as a fixed factor. Primary analyses focused on the faces of the stimuli presented during both the gender and gaze tasks. A similar analysis was performed on dwell duration and fixation count on the eye region of interest for both tasks. Secondary analyses were conducted in a similar manner to above, with univariate ANOVAs including dwell duration and fixation count individually as dependent measures and HC versus IN-PL as a fixed factor.

For the ball-games performance and questionnaires we first performed an univariate ANOVA to determine whether there were any significant differences between the IN-OT, IN-PL or HC groups in their performance during the game. Secondarily, repeated measures ANOVAs were conducted comparing IN-OT to IN-PL, and IN-PL to HC to determine whether participants in a given group threw the ball to one player (or non-human avatar) more than any other. Post-hoc pairwise comparisons with Bonferroni corrections are reported to account for any significant differences in ball tossing from the subjects to the positive, neutral and negative players. Effect sizes are reported with eta-squared (η²) to quantify the actual difference found between two groups being compared. Questionnaires were administered throughout the ball tossing games to determine whether perceptions of the players (or non-human avatars) shifted as they played the game. For the ES-BTG, questionnaires were administered to determine baseline levels of attractiveness, emotional intensity, and trustworthiness of each player in the game. The baseline ratings of attractiveness, emotional intensity and trustworthiness were averaged and utilized in one-way ANOVAs to determine any significant differences between groups. Non-parametric testing of the perception of the other players (or non-human avatars) was conducted using Mann-Whitney U tests to determine differences in emotional recognition attributed to IN-OT among those with SCZ or to determine whether there was a difference between HC and IN-PL. We then utilized Friedman Tests to determine whether there were any within-group differences when comparing IN-OT to IN-PL and IN-PL to HC. Lastly, Wilcoxon ranked sum tests were used to determine whether there were any post-hoc differences between the perception of each of the three players (or non-human avatars).

Results

Demographics

We recruited a total of 54 SCZ and HC patients. We recruited 28 patients who had SCZ however, only 20 met the inclusion criteria and completed the study. We also recruited 26 HC patients, however only 19 met inclusion criteria and completed the study (Table 1.) There were no significant differences in age or race among all the groups. Between the IN-OXT and IN-PL there were no significant differences in PANSS positive or negative scores, and severity of the Deficit Syndrome (F < 2.25, P > 0.1, η² < 0.11).

Table 1.

Demographic characteristics of the sample.

	Schizophrenia		Controls (N=19)	Fisher Exact Test	p-value
	IN-OXT (N=11)	IN-PL (N=9)
Race*	10 (AA)	9 (AA)	13 (AA)		0.121
	1 (Cauca)	0 (Cauca)	6 (Cauca)
Medication*
Atypical Antipsychotics	7	7	NA		0.622
Typical Antipsychotics	2	2	NA
No Antipsychotics	2	0	NA
				T-Test/F-value
Age (mean ± SD)	50.82 ± 9.16	47.78 ± 7.41	51.36 ± 9.25	0.630	0.539
PANSS (mean ± SD)
Total	49.18 ± 8.15	47.78 ± 7.41	NA	0.399	0.695
Negative	13.63 ± 4.10	13.55 ± 4.24	NA	0.043	0.966
Positive	12.09 ± 2.62	11.55 ± 2.50	NA	0.463	0.649
SDS Symptom Score (mean ± SD)	14.73 ± 4.10	11.78 ± 4.66	NA	1.510	0.149
MATRICS (mean ± SD)
Composite	31.73 ± 13.59	27.56 ± 13.26	40.21 ± 10.01	4.031	0.026
Attention/Vigilance	40.73 ± 14.71	33.22 ± 11.32	45.84 ± 10.70	3.365	0.046
Reasoning and Problem Solving	42.36 ± 10.98	42.56 ± 9.21	46.26 ± 7.37	0.893	0.418
Social Cognition	32.18 ± 7.82	35.89 ± 10.06	45.42 ± 12.46	5.864	0.006
Speed of Processing	37.27 ± 9.55	36.33 ± 13.74	46.79 ± 12.66	3.280	0.049
Visual Learning	41.91 ± 14.11	36.78 ± 12.49	43.42 ± 9.77	0.990	0.382
Verbal Learning	40.45 ± 6.22	37.33 ± 5.15	40.79 ± 7.61	0.853	0.435
Working Memory	37.18 ± 16.66	31.56 ± 12.95	40.47 ± 11.17	1.378	0.265

^*Denotes use of Fisher Exact test given small sample size.

Cognition

To confirm that there were significant differences in the cognitive ability of HC as compared to those with SCZ we ran univariate ANOVAs on the MCCB composite score and each domain (attention/vigilance, reasoning and problem solving, social cognition, speed of processing, visual learning, verbal learning, and working memory) individually (Table 1). As expected, the HC on the whole performed better on the MCCB than the SCZ (F=7.53, P=0.009, η² =0.169). More specifically, the HC were significantly better than the SCZ in the attention/vigilance domain (F=4.70, P=0.037, η² =0.113), social cognition domain (F=11.28, P=0.002, η² =0.234), and speed of processing domains (F=6.71, P=0.014, η²=0.153). To determine whether there were any significant differences in the cognitive profiles of the group receiving IN-OXT compared to the IN-PL similar univariate ANOVAs were conducted. There were no significant differences in MCCB scores in the group receiving IN-OXT compared to the group receiving IN-PL (F < 1.75, P > 0.2).

Visual Scanning and Facial Emotion Identification Test (FEIT):

Means and standard deviations of the dwell duration, fixation count, and FEIT are included in Table 2. Regarding focus on the face region of interest during both the gaze and gender tasks, administration of IN-OXT did not lead to a significant increase in dwell duration or fixation count when compared to IN-PL administration (F < 0.60, P > 0.4). Similarly, regarding focus on the eyes alone, administration of IN-OXT did not lead to a significant increase in dwell duration or fixation count when compared to IN-PL administration (F < 0.60, P > 0.4).

Table 2.

Descriptive Statistics of Cognitive and Emotional Recognition Variables

	Schizophrenia		Controls (N=19)	F-value	p-value
	IN-OXT (N=11)	IN-PL (N=9)
Eye tracking
Gaze Task
Dwell Duration, Face	1.01 ± 0.57	1.18 ± 0.52	1.11 ± 0.75	0.182	0.834
Fixation Count, Face	9.19 ± 4.86	9.88 ± 4.83	6.79 ± 3.74	1.767	0.187
Dwell Duration, Eyes	0.2 ± 0.17	0.27 ± 0.22	0.39 ± 0.38	1.502	0.238
Fixation Count, Eyes	1.73 ± 1.34	1.93 ± 1.28	1.92 ± 1.39	0.077	0.926
Gender Task
Dwell Duration, Face	1.05 ± 0.66	0.99 ± 0.78	1.26 ± 0.64	0.573	0.569
Fixation Count, Face	8.34 ± 4.45	7.38 ± 6.24	7.94 ± 4.9	0.085	0.918
Dwell Duration, Eyes	0.15 ± 0.13	0.19 ± 0.25	0.21 ± 0.18	0.322	0.727
Fixation Count, Eyes	1.25 ± 0.86	1.12 ± 1.02	1.36 ± 1.08	0.171	0.844
FEIT Score	11.00 ± 3.58	9.44 ± 4.01	12.16 ± 2.63	2.123	0.134
Emotion-based Social Ball-Tossing Game
Tosses, % towards player, mean ± SD
Positive	41.96 ± 17.66	49.57 ± 26.12	39.68 ± 12.63	0.951	0.396
Neutral	31.47 ± 10.58	26.5 ± 14.95	29.56 ± 8.98	0.51	0.605
Negative	26.57 ± 13.04	23.93 ± 14.62	30.77 ± 11.75	0.968	0.389
Valence-based non-Social Ball Tossing Game
Tosses, % towards player, mean ± SD
Positive	36.36 ± 12.91	48.72 ± 21.76	40.49 ± 10.21	1.904	0.164
Neutral	33.57 ± 20.72	30.77 ± 17.62	29.55 ± 8.21	0.254	0.777
Negative	30.07 ± 13.08	20.51 ± 12.16	29.96 ± 7.2	2.93	0.066

Regarding focus on the face region of interest during the gaze task, there was no significance in the dwell duration, but a trend significance in fixation count during the gaze task, such that IN-PL had a higher number of fixations during the session compared to HC (F=3.192, P=0.087, η²=0.122). There were no significant differences in either the fixation count or dwell duration in the face region during the gender task (F < 1.00, P > 0.3). When focused on the eye region of interest there was no significant difference in dwell duration or fixation count between those who received IN-PL and HC on either task (F < 1.00, P > 0.3).

In a one-way ANOVA comparing the IN-OT and IN-PL, there were no significant differences in the FEIT score (F=0.822, p=0.38). We did find a significant difference between IN-PL and HC (F=4.518, p=0.043) such that HC were able to more successfully identify facial emotions as compared to IN-PL.

Socio-emotional recognition: Emotion-based Social Ball Tossing Game (ES-BTG) and questionnaire

ES-BTG performance

To assess whether there were any significant differences in throw pattern during the ES-BTG we ran a one-way ANOVA comparing proportions of throws to the negative, neutral and positive fictitious players between groups. During the ES-BTG, there was no difference in throw pattern between the two groups of individuals with SCZ (IN-OXT, IN-PL) and the HCs as they threw the ball proportionally about the same to the negative player (F=0.968, p=0.389), neutral player (F=0.510, p=0.605) and positive player (F=0.950, p=0.396) (Fig. 1). To determine whether participants differentially threw the ball to the three fictitious players we ran repeated measures ANOVAs comparing IN-OXT to IN-PL and comparing IN-PL to HC. We found a significantly different within group effect of player type such that patients (IN-OT and IN-PL) differentially passed the ball to the three separate players (positive, neutral and negative; F = 5.71, P = 0.007, η² = 0.241). In post-hoc testing we found that individuals in the IN-PL group preferentially threw to the positive player (49.6%) more than to the neutral or negative player (26.5%; 23.9%) (Bonferroni corrections, P= 0.05; P=0.037). There were no significant differences between the factors within the IN-OT group.

Figure 1.

Distribution of throws to each of the three avatars (positive, negative, and neutral) during the Emotion-based Social Ball-Tossing Game (percentage of total throws with standard error).

When comparing IN-PL compared to HC, there was no between group effect of type of player, however there was a significant within-subject effect (F = 7.715, P=0.001, η² = 0.23). Utilizing Bonferroni corrections, we found that the percentage of ball throws in the IN-PL group was more preferentially directed to the positive player (49.6%) as compared to the neutral (26.5%; P=0.016) and negative player (23.9%; P=0.013). There were no significant differences in the distribution of ball throws in the HC group.

Pregame Questionnaires

Prior to playing ES-BTG, subjects were asked to rate the level of attractiveness, emotional intensity, and trustworthiness of each of the players’ pictures that was used during the game. After giving each of the players’ pictures a rating of attractiveness, emotional intensity, and trustworthiness those ratings were averaged for use in a one-way ANOVA. When comparing IN-OXT to IN-PL we found no significant difference in their ratings of attractiveness (F=0.01, p=0.921). On the other hand, we found a significant difference in their ratings of emotional expression prior to the game (F=6.02, p=0.025) such that IN-PL rated the pictures as being more expressive as compared to the IN-OXT. There was also a trending difference in their ratings of trustworthiness (F=3.77, p=0.068) between IN-OXT and IN-PL such that IN-OXT rated the pictures as more trustworthy than IN-PL.

In a similar analysis between IN-PL and HC we found that there were no significant differences in their ratings of attractiveness (F=2.045, p=0.165), emotional expression (F=0.876, p=0.358), or trustworthiness (F=2.493, p=0.126).

Emotion recognition (mid and post game)

After playing the second phase of the ES-BTG (mid-game), subjects were asked what expression, if any, each of the players displayed on a scale of 1 to 7 (1 = negative, 4 = neutral, 7 = positive). We ran a Mann-Whitney U test to determine whether there were any significant differences in perception of the three players among those in the IN-OT as compared to the IN-PL in the middle of the game. We found that there was no effect of IN-OXT at this stage of the game (U>40, Z<1.5, p>0.25). Among those with SCZ (both IN-PL and IN-OT) we found a significant difference in perception towards the three players using the Friedman Test (χ²(2) = 11.48, p = 0.003). In post-hoc testing using the Wilcoxon Ranked Sum test we found that significantly higher ratings were given to the positive player compared to the negative player (Z=−2.487, p=0.013) and to the neutral player (Z=−3.18, p=0.001) when compared to the negative player.

At the end of the game, utilizing the same testing procedures, we found that those who were administered IN-OT rated the positive player more favorably than the IN-PL group (U=23.5, Z=−2.056, p=0.04) and rated the negative player as more negative than the IN-PL group (U=26.0, Z=−1.963, p=0.050; Figure 2). There were no significant differences between the IN-OT and IN-PL group in the ratings of the neutral player. When assessing within group differences among those with SCZ we found a significant difference in perception of the three players (χ2(2) = 15.60, p < 0.001) such that the positive player was rated as more positive compared to the neutral and negative players (Z=−2.435, p=0.015; Z=−3.13, p=0.002), and the neutral player was rated higher than the negative player (Z=−2.92, p=0.003).

Figure 2.

A. Player ratings of the negative player during the ES-BTG both at the middle of the game and the end of the game. The * indicates a significant difference between the end of game ratings of the IN-PL and IN-OT at p=0.05.

B. Player ratings of the positive player during the ES-BTG both at the middle of the game and the end of the game. The * indicates a significant difference between the end of game ratings of the IN-PL and IN-OT at p=0.04.

In separate analyses comparing HC to IN-PL at mid-game we found a significant difference in the perception of the positive player (U=49.5, Z=−1.976, p=0.048). Further, there was a significant within group difference of perception towards the three players at mid-game (χ2(2) = 11.48, p = 0.003) such that the positive player was rated higher than the neutral and negative players (Z=−2.604, p=0.009; Z=−3.58,p <0.001) and that the neutral player was rated higher than the negative player (Z=−3.084, p=0.002). At the end of the game, the HCs were seen to rank the positive player more favorably than the IN-PL (U=44.5, Z=−2.118, p=0.034). Further, there was a significant within group difference of perception towards the three players at the end of the game (χ2(2) = 26.8, p < 0.001) such that the positive player was rated higher than the neutral and negative players (Z=−3.882, p<0.001; Z=−3.887, p <0.001) and the neutral player was rated higher than the negative player (Z=−2.437, p=0.015).

Non-social recognition: Valence-based non-Social Ball Tossing Game (VNS-BTG)

VNS-BTG performance

To assess whether there were any significant differences in throw pattern during the VNS-BTG we ran a one-way ANOVA comparing proportions of throws to the negative, neutral and positive avatar (each designated by non-social pictures) between groups. During the VNS-BTG, there was no statistically significant difference in throw pattern between the two groups of individuals with SCZ (IN-OXT, IN-PL) and the HCs as they threw the ball proportionally about the same to the negative avatar (F=2.93, p=0.066), neutral avatar (F=0.254, p=0.777) and positive avatar (F=1.904, p=0.164; Figure 3). To determine whether participants differentially threw the ball to the three avatars we ran repeated measures ANOVAs comparing IN-OXT to IN-PL and comparing IN-PL to HC. We found a significantly different within group effects of avatar type such that patients differentially passed the ball to the three avatars (positive, neutral and negative; F = 3.55, P = 0.039, η² = 0.165). In post-hoc testing we found that individuals with IN-PL sent more throws toward the positive avatar (48.7%) as compared to the negative avatar (20.5%) (Bonferroni corrections, P=0.002). However, there were no significant differences in the distributions of ball tosses to any of the avatars in the IN-OT group.

Figure 3.

Distribution of throws to each of the three avatars (positive, negative, and neutral) during the Valence Non-Social Ball-Tossing Game (percentage of total throws with standard error).

When comparing IN-PL compared to HC, there was no between group effect of type of player, however there was a significant within-subject effect (F = 11.21, P<0.001, η² = 0.301). Utilizing Bonferroni corrections, we found that the percentage of ball throws in the IN-PL group was more preferentially directed to the positive avatar (48.7%) as compared to the neutral avatar (30.8%; P=0.043) and the negative avatar (20.5%; P=0.001). Within the HC there was a significant difference in the number of throws towards the positive avatar (40.5%) as compared to the negative player (30%; P=0.041).

Valence recognition (mid and end of game)

During the VNS-BTG (mid-game), subjects were asked what expression, if any, each of the avatars displayed on a scale of 1 to 7 (1 = negative, 4 = neutral, 7 = positive). We ran a Mann-Whitney U test to determine whether there were any significant differences in perception of the three players among those in the IN-OT as compared to the IN-PL in the middle of the game. We found that there was no effect of IN-OXT at this stage of the game (U>40, Z<1.5, p>.25). Among those with SCZ (both IN-PL and IN-OT) we found a significant difference in perception towards the three avatars using the Friedman Test (χ²(2) = 15.7, p < 0.001). In post-hoc testing using the Wilcoxon Ranked Sum test we found that significantly higher ratings were given to the positive avatar compared to the neutral (Z=−2.311, p=0.021) and negative avatars (Z=−3.295, p=0.001) and that higher ratings were given to the neutral avatar compared to the negative avatar (Z=−2.885, p=0.004).

Similarly, at the end of the game there was no effect of IN-OXT on the ratings of the avatars when compared to the IN-PL (U>32, Z<1.4, p>0.20; Figure 4). Among those with SCZ (both IN-PL and IN-OT) at the end of the game there was a significant difference in perception towards the three avatars using the Friedman Test (χ²(2) = 15.22, p < 0.001). In post-hoc testing using the Wilcoxon Ranked Sum test we found that significantly higher ratings were given to the positive avatar compared to the neutral (Z=−2.814, p=0.005) and negative avatars (Z=−3.236, p=0.001) and that higher ratings were given to the neutral avatar compared to the negative avatar (Z=−2.331, p=0.020).

Figure 4.

A. Player ratings of the negative player during the VNS-BTG both at the middle of the game and the end of the game.

B. Player ratings of the positive player during the VNS-BTG both at the middle of the game and the end of the game.

In separate analyses comparing HC to IN-PL at mid-game of the VNS-BTG there was no significant difference in the perception of the avatars between the HC and the IN-PL (U>55, Z<1.7, p>0.100). However, there was a significant within group difference of perception towards the three players at mid-game (χ²(2) = 32.65, p < 0.001) such that the positive player was rated higher than the neutral and negative players (Z=−3.8, p<0.001; Z=−4.325,p <0.001) and that the neutral player was rated higher than the negative player (Z=−3.745, p<0.001). At the end of the game, there was a significant difference in the ratings of the negative avatar between the HC and the IN-PL (U=41.5, Z=−2.532, p=0.011) such that the HC rated the avatar as more negative compared to the IN-PL. At the end of the game, the HC also rated the positive avatar as more positive than the ratings given by the IN-PL (U=48.5, Z=−2.288, P=0.022). There was a significant within group difference of perception towards the three players at the end of the game (χ2(2) = 34.14, p < 0.001) such that the positive player was rated higher than the neutral and negative players (Z=−3.980, p<0.001; Z=−4.430, p <0.001) and that the neutral player was rated higher than the negative player (Z=−3.658, p<0.001).

Discussion

We investigated the effects of IN-OXT on subjects’ visual fixation of faces during gender identification and during detection of eye-gaze direction. We did not find a difference in dwell time or fixation count between subjects who received IN-OXT and those who received IN-PL. This lack of difference could be due to our small sample size which would not allow us to detect a significance with a moderate effect size of drug on eye-tracking behaviors. In line with the recent meta-analysis³¹, IN-OXT seems to have larger effects on high-level social cognition compared to simple social perception. Given that scanning faces and attending to social cues is an embedded part of social perception, it is possible that IN-OXT does not have effects on visual scanning in SCZ.

We also investigated the effects of IN-OXT on subjects’ capacity to recognize emotions in others during an interactive social context. Prior to the initiation of the ES-BTG we found a significant difference in their ratings of emotional expression such that IN-PL rated the pictures as being more expressive as compared to the IN-OXT. This could be due to problems with emotion recognition and attribution, stating that the images presented were showing emotions that actually were not present given the neutral stimuli before the start of the game. There was also a trending difference in their ratings of trustworthiness (F=3.77, p=0.068) between IN-OXT and IN-PL such that IN-OXT rated the pictures as more trustworthy than IN-PL. This could be indicative of the positive effect of oxytocin on trustworthiness and interactions with others. During the ES-BTG, patients who received placebo had a similar distribution of ball throws toward the 3 players compared to those who received IN-OXT. They reciprocated more often with the positive player during the social interaction than with the negative and neutral players. However, at the end of the game, when asked about players’ emotional expressions, those who received IN-OXT rated the picture that represented the positive player as more positive than the ones representing negative and neutral players. They also rated the negative player as more negative compared to the IN-PL. Interestingly, IN-OXT did not affect emotion recognition of non-social cues during the valence-based non-social ball tossing game.

These results are in line with other studies which show that IN-OXT enhances emotion recognition and theory of mind in SCZ^46–49, although it should be noted that other studies did not find significant effects^50.51. Emotion recognition requires first-level perception of social cues, emotional feedback, comprehension of social norms and the ability to recall and attribute different emotions to each individual. Therefore, emotional recognition within social interactive contexts cannot be equated to a simple social perception. Thus, the findings with IN-OXT could be related to the interpretation of social cues that require emotion recognition and evaluation. This finding corroborates the hypothesis that IN-OXT has more prominent effects on higher order cognitive processing as compared to bottom up social perception in SCZ.

In line with OXT’s prosocial effects that are observed in the literature, we found that IN-OXT enhanced the rating for positive players. These results are in line with previous literature showing that IN-OXT enhances social approach behavior in the context of positive social cues⁵². Another study assessing IN-OXT in healthy subjects found that IN-OXT increased the participants’ desire to play with others in a cyber-ball game⁵². We also found that IN-OXT reduced the rating for negative players, which supports the hypothesis that IN-OXT effects observed herein are related to emotion recognition and attribution of appropriate emotional valence to social cues.

While several studies have shown that IN-OXT enhances prosocial behaviors in healthy subjects, IN-OXT is also known to enhance prosocial behaviors in patients with social and psychiatric disorders^16,25,53. We previously found that IN-OXT led to more prosocial behaviors in patients with autism spectrum disorder whose social clinical symptoms are less pronounced²⁵. Another recent study by Woolley et al. (2017) showed that IN-OXT increased facial expressivity in individuals with schizophrenia in response to emotional stimuli, suggesting that oxytocin may have a potential role in treating the blunted emotions in SCZ⁵⁴.

Our findings show that IN-OXT has limited and small effects on socio-emotional processing. IN-OXT enhanced emotion recognition, but not attention to social cues. A recent dose-dependent study of oxytocin using neurophysiological measures showed that mid-range dose (36–48IU) is optimal to enhance social processing in SCZ⁵⁵. Thus, it is possible that lower acute doses of IN-OXT (such as 24IU used herein) are not optimal. Future long-term trials using mid-range doses of oxytocin and targeting socio-emotional deficits in SCZ may potentially lead to more positive outcomes. Moreover, the limited effects seen in this study could be attributed to limits of the brain penetrance of IN-OXT. However, this hypothesis is less plausible given the recent findings of labeled intranasal oxytocin in the cerebrospinal fluid and brains of macaques^56,57, and an increase in brain oxytocin concentrations after administration of intranasal oxytocin in oxytocin knockout mice⁵⁸. Our study is limited by its small sample size and the between-subject study design, both of which limit the power of the study. The between-subject design and the absence of the use of baseline measurement (pre-treatment) are limitations because they do not rule out the possibility of unspecified differences mediating treatment effects. However, both groups were matched in terms of symptom severity and clinical characteristics. Another limitation of the study is the use of static facial stimuli during interactive social games. The use of more dynamic and ecologically valid stimuli would improve the assessment of drug effects on social behavior. We would also like to acknowledge that this study includes only male gender patients. Among these, 95% are of African American ethnicity. Therefore, our results may not generalize to women with SCZ or to individuals of a different race. To generalize our findings future studies should include a larger sample size, both genders, and greater racial diversity among the participants.