Pupil dilation mimicry boosts trust in depression

(1)

Pupil dilation mimicry boosts trust in depression

Katharina Wehebrink (10608230) BCS Research Master Thesis

University of Amsterdam Supervisor: Mariska Kret Co-assessor: Katja Kölkebeck

(2)

Abstract

Background: Previous research has shown an important role of pupil mimicry during social interaction and in trust decisions. Trust is often impaired in individuals who suffer from depression. Therefore, the current study examined whether there is a relationship between depressed individuals’ trust deficit and a lack of pupil mimicry.

Methods: Forty-two depressed and thirty-nine healthy control subjects played trust games with virtual partners. Images of these partners’ eye region were presented to participants before they had to decide how much they wanted to invest in their partner. Partners’ pupils either dilated, constricted, or stayed static. To assess subjects’ pupil mimicry and its effect on decisions of trust, their pupil size was recorded with eye tracking equipment.

Results: Whereas controls trusted partners with dilating pupils more than partners with constricting pupils, depressed subjects distrusted partners whose pupils changed in size, and independent of whether their partners pupils dilated or constricted, invested less than in partners whose pupils remained static. This difference in investment behaviour could not be explained by pupil mimicry, as both groups mimicked the pupil size of their partners, and pupil dilation mimicry boosted trust decisions in the depressed group, whereas there was no relationship between pupil constriction mimicry and distrust in either of the groups.

Conclusions: The data suggest that depressed individuals do mimic the pupil size of a partner, but as opposed to healthy individuals, they show lower trust in general. Yet, mimicking dilating pupils seem to help depressed individuals to evaluate a partner’s trustworthiness.

These findings provide further evidence for the important role of pupil size and pupil mimicry in interpersonal trust.

(3)

Introduction

In many situations, whether we trust a person or not is a decision of milliseconds. Nearly every social interaction between humans is based on trust and reciprocation (Lewis & Weigert, 1985). To evaluate a counterpart’s trustworthiness, we rely on certain characteristics that signal a safe interaction such as emotional expressions, bodily gestures or group membership (e.g. Dunn & Schweitzer, 2005; Oosterhof & Todorov, 2008).

Psychiatric disorders such as major depressive disorder often come along with impaired social skills (Segrin, 2000; Silk et al., 2008), including deficits in emotion recognition (Kret & Ploeger, 2015): individuals with these disorders avoid eye contact and have difficulties in building up trustful relationships (Lester & Gatto 1990; Muris, Meesters, van Melick, &

Zwamback, 2001; Benhua & Yanxiang 2004). The ability to process signals of trust and translate them into behaviour seems to be impaired which might also be related to lower levels of

cooperation, egocentric behaviour and limited ability of perspective-taking (Brendan Clark, Thorne, Hardy & Cropsey, 2013; Cusi, Nazaronov, Macqueen, & McKinnon, 2013; Inoue, Tonooka, Yamada, & Kanba. 2004; Rochat, Billieux, & van der Linden, 2012). The mechanisms underlying these deficits are poorly understood. Previous studies that tried to explore these deficits have mainly focused on explicit signals, such as facial expressions (e.g. Rubinow & Post, 1992). Yet, recent research findings suggest that the pupil or more specifically pupil mimicry plays a significant role in social decision making (e.g. Hess, Kret, Tomonaga & Matsuzawa, 2014). Because pupillary responses are autonomic and not controllable (Loewenfeld, 1993), they can provide important reliable information. Therefore the current study tries to link the findings on pupil mimicry to social skill impairments in depression.

(4)

For humans and other animals, eyes do not only have a visual function but also serve as a reference point to be seen by others (e.g. Tomasello, Hare, Lehmann, & Call, 2007). Research has shown that the pupil dilates or constricts not only in response to different lighting conditions but also in response to emotions and thoughts (Goldwater, 1972; Loewenfeld, 1993; Laeng, Sirois, & Gredeback, 2012). By observing the pupil one can acquire significant information about the emotional, mental or cognitive state of another person (Mistry, 2005). Several

pupillometry studies demonstrated increasing pupil sizes when viewing pictures with arousing or emotional relevant stimuli (e.g. Hess & Polt, 1960; Peavler & McLaughlin, 1967; Kret,

Stekelenburg, Roelofs, & de Gelder, 2013). Furthermore, larger pupil sizes are associated with increased approach behaviour and attractiveness in humans (Laeng & Falkenberg, 2007; Wisemann & Watt, 2010).

More recently, the concept of pupil synchronization or pupil mimicry has gained rising attention. It has been shown that we implicitly mimic the pupil size of our interaction partners (Kret et al., 2014; Kret, Fischer & de Dreu, 2015) and that the amygdala, an area densely connected with the brainstem’s autonomic nuclei regulating the pupil, plays a key role in this phenomenon (Harrison, Singer, Rotshtein, Dolan, & Critchley, 2006; Demos, Kelley, Ryan, Davis, & Whalen, 2008). Moreover, a recent study by Kret, Fischer and de Dreu (2015) indicated that paying attention to other people’s pupils and mimicking their changes in size helps to

determine the trustworthiness of a partner. Healthy subjects trusted partners with dilating pupils more than partners with static pupils, and especially so when their pupils synchronized, i.e. when subjects’ pupils dilated along with the dilating pupils of the partner. Thus, in order to make a reliable decision about a counterpart’s trustworthiness people tended to use the pupil as an implicit reference point.

(5)

These findings by Kret et al. (2015) suggest that pupil mimicry can be important for the establishment of trust, and possibly, that a lack of pupil mimicry could account for a lack of trust. From that point of view, it can be hypothesised that the known lower levels of trust in depressed individuals (e.g. Lester & Gatto, 1990) could be associated with a failure to mimic interaction partner’s pupil sizes and missing out on the opportunity to implicitly infer trust from their own pupil sizes. The present study examined this hypothesis by investigating the relationship between pupil mimicry and trust conceptions in hospitalised patients with a clinically diagnosed major depressive episode as compared to a healthy control group. First, it was examined whether the participants’ pupil size assembles with a partner’s dilating or constricting pupils as compared to partner’s pupils that remain static and unchanged. Second, it was tested whether there is a difference in pupil mimicry between depressed individuals and healthy controls, and third, whether pupil mimicry influences trust decisions in the two groups.

In sum, previous research has shown that a primary characteristic of depression is the impairment in social functioning including a lack of trust in others (Lee, Harkness, Sabbagh, & Jacobson, 2005; Muris et al., 2001). Even though these difficulties in depressed individuals are pervasive, a detailed understanding of the cognitive mechanisms underlying these deficits has not been reached. Because recent research findings point to an important role of pupil mimicry in social interaction and trust decisions in particular (i.e. Kret et al., 2015), the pupil is introduced as a possible candidate in explaining the impairments in depression.

The current study investigated whether there is a difference between depressed

individuals and healthy controls in pupil mimicry and whether there is a link with investments. Specifically, it was tested whether depressed subjects mimic their partners’ pupils less than controls and therefore in their decision, rely less on the pupillary changes in the partner and the

(6)

pupillary changes in themselves. Trust was measured via a trust game where participants could invest in a virtual partner of whom they could only see the eye region with therein the dilating, constricting or static pupils. Participants’ pupils where tracked with an eye tracker to measure pupil mimicry.

Method and Materials Participants

A total of 106 participants were recruited at the University Hospital Münster, Germany. The sample included 64 patients with depression (32 males) and 42 healthy controls (24 males). While inpatients volunteered, outpatients and controls received a compensation of 20 EUR, following standard guidelines of the institute. All participants took part in the structured clinical interview for DSM-IV (SCID-I; First, Spitzer, Gibbon, & Williams, 2012) in order to confirm patient’s diagnoses of major depressive disorder and to rule out any other psychiatric disorders in patients and control subjects. Furthermore, the Hamilton Depression Rating Scale was conducted to rate depressive symptoms and their severity in both groups (Hamilton, 1960). For patients, all Axis-I or Axis-II psychiatric disorders other than depression formed exclusion criteria. For control subjects, any history of psychological disease constituted an exclusion criterion. Nineteen patients and two controls were excluded because they met the exclusion criteria. Three other subjects had to be excluded because they could not complete the task due to technical

difficulties. Therefore, the statistical analyses are based on a sample size of 41 depressed patients (mean age = 39.26, SD = 11.45) and 39 healthy controls (mean age = 39.56, SD = 11.00).

The two groups differed on Hamilton scores (t(78) = 42.818, p < .001), level of arousal (stimulus-unrelated pupil size; F(1,4309) = 14.051, p < .001) and appropriate use of Theory of Mind (ToM) vocabulary in the Moving Shapes paradigm (t(79) = -1.245, p = .017). There were

(7)

no other significant differences between the two groups (see Table 1). Moreover, bivariate correlations between all continuous variables were calculated (see Appendix A). All subjects had (corrected to) normal visual acuity, as assessed with a Snellen chart (Snellen, 1862).

The study was approved by the local ethics committee (2012-495-f-S) and the internal review board of the University of Amsterdam (2012-WOP-2159). All participants were informed in detail about the aims and the protocol of the study and provided written informed consent. Table 1. Demographics and test scores of included participants.

Mean (SD) Mean differences

Depression group (N = 42) Control group (N = 39) _χ2_{/t / F test values p value}

Gender (m/f) 23/20 22/17 _χ2 _{(1) = 0.071} _0.791 Age 39.26 (11.45) 39.56 (11.00) t (79) = - 0.061 0.805 Hamilton Score 14.61 (5.83)1 _{0.67 (1.15)} _{t (78) = 42.818} _{< .001} EQ score 47.40 (40.89) 42.69 (38.32) t (79) = 0.534 0.595 Baseline corrected pupil size 83.93 (155.86) 89.26 (157.50) F(1, 105813) = 0.017 0.898 Arousal 2383.27 (901.98) 1775.13 (597.21) F(1,4309) = 14.051 < .001 Reaction time 1.421 (72.561) 1.285 (76.191) t (4.422) = 1.290 0.197 ToM Intentionality 2.98 (0.97) 3.23 (0.79) t (79) = -1.245 0.217 ToM Appropriateness 1.24 (0.49) 1.49 (0.35) t (79) = - 2.449 0.017 GD Intentionality 2.57 (0.54) 2.72 (0.48) t (79) = -1.326 0.189 GD Appropriateness 1.54 (.32) 1.59 (0.44) t (79) = - 0.688 0.493

1_{Scores of 2 participants are missing.}

EQ Empathy Quotient

Baseline corrected pupil size The average pupil size 500ms prior to the start of change in partner’s pupils was subtracted from the final filtered pupil size for each participant and served as the baseline corrected pupil size for all analyses

Arousal Participant’s stimulus-unrelated pupil size, i.e., participant s average pupil size 200-400ms prior to stimulus onset. The stimulus-unrelated pupil size might indicate participants’ general level of arousal unrelated to the stimulus material

ToM = Theory of Mind, which includes the ability to think about mental states such as thoughts and beliefs in oneself and others (Baron-Cohen, 1989).

(8)

Stimuli

To create stimuli for the virtual trust game, photographs of nine men and nine women were standardized by using Adobe Photoshop®. This included resizing the images into life size, turning them into grayscale, cropping them to reveal only the eye region, retouching distracting blemishes, and standardizing contrast levels and mean luminosity. The eyes (pupil, iris, and parts of the sclera) were removed from the images and replaced by a video layer using Adobe After Effects that featured artificial eyes with dilating, constricting, or static pupils. Pupil diameter was 5mm in the beginning; adjusting pupils dilated to 7mm diameter or constricted to 3mm diameter. In total, this yielded 54 stimuli – 18 faces for each of the three conditions of partner’s pupil (dilation, constriction, and static). These stimuli constituted the virtual partners in the trust game task. Scrambled versions of the stimuli were created by applying a Fourier transformation with a Gaussian filter to each image Using MATLAB fft-function. This yielded 18 unique pre-stimulus images with the same luminosity and level of detail as the respective pre-stimulus, but lacking any semantic content (e.g., Hoffman, Gothard, Schmid, & Logothetis, 2007). Procedure

All participants were tested individually. Before the actual eye-tracking task started, participants received a brief explanation about the study and signed the informed consent. In addition, they completed some questionnaires (see Table 1). Subjects were then familiarized with the eye tracking procedure (including checking for sufficient eye-sight and the dominant eye) and the trust game task (Berg, Dickhaut, & McCabe, 1995). After that, they were seated in front of a computer screen (1280x1024px) with a distance of ~50cm. To minimize movement, the head was placed on a head stabilizer. The room was darkened, with the only light sources

(9)

being the stimulus screen and the Eyelink screen (which was visible only to the experimenter). The trust game started after all settings to track the pupil correctly had been checked.

The trust game was programmed in Presentation (v16.2, Neurobehavioral Systems). Each trial started with a scrambled version of the stimulus. After 3000ms, a grey fixation cross

appeared in the middle of the scrambled image. To verify that the eyes of the subject were directed towards the relevant part of the screen, the stimulus only appeared after participants’ gaze fell within an area of interest around the fixation cross (70x70px) for 1000ms. The pupils of all stimuli stayed static for 1500ms. Then, the pupils dilated, constricted, or remained static for another 1500ms (see Figure 1 for an example of a trial). In all three conditions, partner’s pupils were static during the final 1000ms of stimulus presentation. After subjects observed the partner for 4000ms, they used two response boxes (with two buttons each) to decide how much money they wanted to transfer to their partner (0, 2, 4, or 6 EUR). They were told that the transferred amount gets tripled and credited to their partner, who could then decide how much he wants to retransfer back to the participant. No feedback was given in between trials. After a button press, the next trial started with its respective scramble. Figure 1 gives an overview of the trial outline.

After participants had finished the trust game, they were asked to participate in an additional task: the moving shapes paradigm (Abell, Happé, & Frith, 2000). This often-used paradigm assesses ToM abilities by showing silent animations of two triangles that are

interacting with reference to mental states (ToM), displaying goal-directed behaviour (GD), or moving randomly (RM). Participants were asked to describe the interactions. The descriptions were recorded for later evaluation. The answers were scored on a) the frequency of ToM vocabulary (intentionality), b) appropriate understanding of the story line (appropriateness) and c) length of answers. In the end, participants were informed about the actual goal of the study

(10)

and the SKID-I interview was conducted.

Figure 1. To create experimental stimuli, the eyes from pictures of the eye regions of faces were removed and then added to the same eye white, iris, and pupil to each stimulus. In each trial, a scrambled image of a stimulus was presented for 4,000ms. This image was then replaced by the stimulus previously prepared stimulus of the eye region. In all conditions, the stimulus remained static for the first 1,500ms. In the dilation and constriction conditions, the pupils gradually changed in size over the following 1,500ms and then remained at that size during the final 1,000ms. Finally, a screen appeared asking participants to decide to transfer 0, 2, 4, or 6€ to their partner.

Data Preparation

Participants’ pupil size and gaze were continuously recorded with 500Hz using an EyeLink 1000 eye tracking system (SR Research). Gaps smaller than 250ms were interpolated. Data was smoothed by applying a 10th order low-pass Butterworth filter with a cut-off frequency of 10Hz. Trials were excluded if more than 50% of the data within that trial were missing (i.e., because the eye tracker lost the pupil). Outliers were removed with an algorithm using the

(11)

differences between a subject’s subsequent pupil size samples: If this difference was larger than the mean difference plus 1.5 times the standard deviation of the differences, that sample was deleted. Furthermore, typical overshoot-like artefacts around blinks were removed over an area of -40ms to +80ms on both sides of the blink. This procedure led to the exclusion of 2,97% of trials. In order to improve efficiency of upcoming computations, the data was reduced to 50 Hz by averaging 50 subsequent samples to form one value of pupil size. Visual inspection of line plots confirmed that the reduced amount of data was indeed sufficient to accurately represent the growth curve of pupil size.

The average pupil size of 500ms of every participant and every trial (thus 5 values) before the partners’ pupils started to change (10000ms-1500ms after stimulus onset), served as a baseline and was subtracted from all remaining pupil size values (1600ms-4000ms).

Statistical Analysis

Because of the nested structure of the data, the most appropriate method to analyse the data is through multilevel modelling (Bagiella, Sloan, & Heitjan, 2000; Hox, 2002). In order to acquire the model with the best fit to the data, we started with a full model and deleted non-significant higher level terms step by step. Every extinction was checked with the Akaike information criterion (AIC; Doherty, White, & Burnham, 2012).

For the participant’s pupil analysis, a three-level regression model, defined by pupil size samples (repeated measures) nested within trials, and trials nested within subjects, was used (Linear Mixed Model, implemented in SPSS Version 22.0; Bagiella et al., 2000; West, Welch, & Galecki, 2006). Time was added as a repeated factor with a First-Order Autoregressive

covariance structure (AR1) in order to control for auto-correlation over time points. The baseline corrected pupil size served as the dependent variable. The factors Partner’s Pupil (constriction

(12)

vs. static vs. dilation, coded as -1, 0 and 1 respectively) and Group (depression vs. control, coded as -1 and 1) and their interactions served as predictors. Furthermore, three orthogonal

polynomials were included to account for linear, quadratic and cubic trends in the growth curves. A random intercept and random linear, quadratic and cubic terms accounted for individual differences. Furthermore, a random effect of actor was examined. In the main results section only effects of group and partner’s pupil will be described, but all effects can be found in the supplementing result section.

Trusting behaviour was analysed with a series of two level models defined by the different trials that were nested within participants. Trust decisions served as the dependent variable in all models. Group, partner’s pupil and their interaction served as predictors. In addition, in order to further investigate the source of trust, participants pupil response was also investigated through the division of Mimicry/No-Mimicry trials, Dilation Mimicry trials or Constriction Mimicry trials, based on a median split in each participant. For example, a trial was categorized as a ‘dilation mimicry trial’ or ‘a constriction mimicry trial’ when the mean pupil size of the trial was higher than the median pupil size of a participant when viewing a partner with dilating pupils or when viewing a partner with constricting pupils. The statistical models included the factors group, type of mimicry (Mimicry/No-Mimicry, Dilation Mimicry or Constriction Mimicry) and the interaction between the two.

To further investigate whether possible differences between depressed patients and controls are related to the depression or to other unforeseen mediating factors, seven additional candidates being reaction times, level of arousal, Hamilton scores, EQ scores, and ratings of the moving shapes animations (ToM Intentionality/ Appropriateness and GD

(13)

In the results section, we will only describe effects of group and partner’s pupil, as well as important additional candidates that changed the interpretation of the results. For all full models see Appendix D.

Results Trust investments

Results showed that depressed subjects tended to trust their partners less than control subjects (F(1,4.392)=3.907, p=.051). Moreover, partners with static pupils were trusted more than partners with dilating or constricting pupils (F(2, 4.392) = 27.06, p < .001). This main effect was however driven by the patient group, as shown by a significant Group x Partner’s Pupil interaction (F(2, 4.392) = 4.139, p = .016). As predicted, control participants trusted partners with constricting pupils less than partners with static pupils (t(4.391) = 4.501, p < .001) and partners with dilating pupils (t(4.392) = 3.334, p = .001). The difference between dilating and static pupils was not significant (t(4.391) = 1.159, p = .246). Participants in the depressed group displayed a different pattern: they trusted partners with dilating or constricting pupils less than partners with static pupils (dilating- static pupils: t(4.392) = 5.258, p < .001; constricting- static pupils: t(4.392) = 5.849, p < 001, see Figure 2A).

A crucial question is whether depression is causing this lack of trust in partners whose pupils change in size, or whether other factors are at play. In an attempt to better understand this effect, below, we will explore possible mediating factors. First, an analysis of the reaction times shows that depressed participants take longer to make a decision about partners with static pupils than dilating (t(4.392) = 2.832, p = .005) and constricting pupils (t(4.392) = 2.923, p = .003). The reaction times in the control group are unaffected by the changes in partners pupil sizes (ps ≥ .86). See Figure 2B for a graphical overview. The fact that reaction times differ, suggests that

(14)

patients have different strategies or recruit different networks, while taking the decision. Yet, when including reaction time as an additional covariate, the interaction between group x partner’s pupil is still significant (F(2, 4421) = 4.424, p = .012). Thus, reaction time is not mediating this effect.

Second, because patients and controls differ in their level of arousal (F(1,4309) = 14.051, p < .001), we included level of arousal as a covariate in the model. This inclusion eliminated the significant main effect of group (F(1, 4390) = 1.919, p = .166). Yet, the significant group x partner’s pupil interaction remained unchanged, meaning that level of arousal is not mediating this effect.

Third, although the groups differed on some ToM abilities, these differences could not explain the different investment pattern between the two groups (see Appendix D Table D1 G-J).

Figure 2. Bar chart A illustrates the mean trust level of depressed subjects and controls in the

constriction, dilation, and static condition. Bar chart B shows the mean reaction times of depressed subjects and controls in the three within-subject conditions. Error bars indicate ±1 SE. (*p < .05, **p <

.01).

(15)

Pupil mimicry

The concept of pupil mimicry involves that a participant’s pupil size should be larger when viewing dilating pupils than when viewing static pupils and larger when viewing static than when viewing constricting pupils. In addition, pupil mimicry might be reflected in the Linear Trend × Partner’s Pupil Size interaction and the Quadratic Trend × Partner’s Pupil Size interaction.

Results show that participants synchronized with their partners pupil size, as indicated by a significant effect of Partner’s pupil (F(2, 4705.17) = 6.086, p = .002). In addition, a Partner’s Pupil x Linear trend interaction was found (F(2, 50257.49) = 10.943, p < .001). Visual inspection shows indeed that participants’ pupils were largest and increased fastest when partners’ pupils dilated. See Table D2 in Appendix D for all statistical models.

Pupil Dilation mimicry

Pupil Dilation mimicry was revealed in effects of partner pupil size, F(1, 70.570) = 10.39, p = .001 and a Partner Pupil Size × Linear Trend interaction, F(1, 70.570) = 17.872.80, p < .001. Participants’ pupils were larger, dilated faster, and showed a greater peak when

participants observed partners with dilating versus static pupils. Even though there was no significant Partner’s Pupil × Group interaction (F(1, 70.570) = 2.156, p = .142), further

exploration shows that the pupil size of control participants significantly differs when watching partner’s with static vs. dilating pupils (t(70.570) = -3.266, p = .001) whereas this is not the case for depressed participants (t(70.570) = -1.249, p = .214). Graphical inspection supports this trend: Figure 3A shows that the control group has bigger pupil sizes when watching partners dilating pupils vs. static pupils. In contrast, pupil sizes in the depressed group do not seem to

(16)

differ within the first half after the pupil of the partner starts to change. See Table D3 in Appendix D for all statistical models.

Pupil Constriction mimicry

None of the groups displayed pupil constriction mimicry. See Appendix C for a graphical overview and Table D4 in Appendix D for all statistical models.

Dilation mimicry and trust

Results showed a significant effect of dilation mimicry (F(1, 1.432) = 3.873, p = .049) showing that participants invested more in partners with dilating pupils if their own pupils mimicked them as compared to when they did not. Furthermore, a trend towards a significant dilation mimicry x group interaction was observed (F(1, 1.432) = 3.523, p = .061), showing that trust in no dilation mimicry trials is significantly lower in the depressed group than in the control group (t(1472) = -2.800, p= .005, see Figure 3) and that the depressed group invests more in mimicry trials than in non-mimicry trials (t(1472) = -2.651, p = .008), whereas the control group does not show such difference (t(1472) = -0.085, p = .933; see Figure 3B). As we observed differences between the two groups in their reaction times, ToM appropriateness and level of arousal, it is possible that including these variables mediate the trend towards a group x dilation mimicry interaction. We explored that possibility, but including these variables did not affect the interpretation of the results. See Table D5 in Appendix D for all statistical models.

(17)

Constriction mimicry and distrust

No significant effects of constriction mimicry were observed (ps ≥ .118). See Table D6 in Appendix D for all statistical models.

Figure 3. Graph A shows pupil size in depressed and control subjects in reaction to

partners with dilating and static pupils. It can be seen that the two groups display a different pattern: the pupil size in the control group is constantly bigger when watching dilating pupils, whereas depressed participants’ pupil size does not differ when watching partners with dilating and static pupils within the first half after stimulus presentation. Graph B displays the further exploration of the dilation mimicry x group interaction. Whereas the depressed group shows significantly higher trust investments in dilation mimicry trials than in no dilation mimicry trials, the control group displays similar trusting behaviour for dilation mimicry trials and no dilation mimicry trials. Furthermore, the depressed group shows significantly lower trust investment in no mimicry trials compared to the control group. This indicates that especially depressed participants benefit from

mimicking partners pupil size. (** p < .01).

(18)

Discussion

The current study tested whether there is a relationship between pupil mimicry and depressed patients’ lack of trust. Results demonstrated that depressed participants trusted

partners with dilating pupils less than controls, but that both groups mimic the pupil size of their partners to the same extent. Most remarkably, whereas pupil mimicry helped depressed

participants to trust partners with dilating pupils more, this was not the case for healthy control participants.

The perception of partners’ pupil size

To avoid betrayal and allow trust, individuals must understand the intentions and emotions of their group members. Research indicates that people with depression have limited social abilities including deficits in emotion recognition (Kret & Ploeger, 2015): They avoid eye contact and have trust building problems (Benhua & Yanxiang 2004; Lester & Gatto 1990; Muris et al., 2001).

The human eye region - or more specifically the pupil - seems to play an important role in social decision making. Previous research in healthy participants has shown that people generally hold positive associations with large or dilating pupils (Amemiya & Ohtomo, 2011, Kret et al., 2014). For instance, Hess (1975) showed that if people are instructed to draw eyes for an angry face and a smiling face, they generally draw big pupils in the happy and small pupils in the angry face. In addition, the study by Kret et al. (2015) indicated that the dilation of a

partner’s pupil induces trust.

Even though there is no comparable study that explored pupil mimicry in depressive participants, previous studies have shown that social functioning in general and social decision-making in particular is abnormal in depression (i.e. Destoop, Schrijvers, De Grave, Sabbe, & De

(19)

Bruin, 2011, Lee et al., 2005). In the current study we predicted that depressed individuals demonstrate less trusting behaviour than healthy controls and fail to link dilating pupils to trustworthiness as has been shown in control participants in previous research. Our results support this hypothesis: depressed participants trusted partners with dilating pupils less than healthy controls. In fact, depressed patients trusted partners with dilating pupils just as little as they trusted partners with constricting pupils. Thus, whereas control participants differentiate between dilating and constricting pupils, depressed participants do not show this behaviour, but trust partners with static pupils most. In general, pupil diameter reflects state of arousal and emotional interest (i.e. Janisse, 1977; Granholm & Steinhauer, 2004). Yet, depressed individuals trust partners who do not show any arousal or emotional interest most. They seem to avoid any kind of social interaction. In line with this are the findings of earlier research that demonstrated that unlike healthy controls depressed patients do not exhibit dominant action tendencies towards happy or angry expressions (Radke, Güths, Andre, Müller, de Bruijn, 2014). Depressed patients might focus more on the avoidance of rejection and punishment (Destoop et al., 2011) and have a lack of implicit mutual trust or openness (Ratcliffe, 2014). Therefore, depressed participants might not differentiate between partners with dilating or constricting pupils but distrust both.

Further exploration of reaction times showed that patients in the depressed group are particularly slower when making decisions about partners with static pupils. This might indicate that static pupils are processed differently in patients versus controls. Changing pupil sizes reflect changes in partner’s inner state and are therefore supposed to be more socially salient cues than pupils that do not change in size (Harrison, Gray, & Critchley, 2009). In contrast, static pupils do not change in size and might not be as socially relevant. Depressed individuals might thus immediately react to dilating or constricting pupils to prevent social interaction but take

(20)

some more time to make a decision about partners with static pupils. Interestingly, when comparing the investments on partners with static pupils, there is no significant difference between the control group and the depressed group. Thus, depressed individuals invest the same amount but take longer for this decision, indicating that the amount of trust that they offer in static pupils is ‘normal’. Yet, they need more time than controls to reach this decision. Depression is characterized by high levels of automatic avoidance especially across social domains (Ottenbreit, Dobson, & Quigley, 2014; Radke et al., 2014) and linked to general low levels of interpersonal trust (Kim, Chung, Perry, Kawachi, & Subramanian, 2012). Therefore, a possible explanation for the slower reaction times could be that the automatic response of avoidance or distrust has to be down-regulated when recognizing a non-emotionally loaded stimulus.

Pupil Mimicry in Depression

Mimicry of another persons’ behaviour, gestures or facial expressions grounds positive feelings and impacts the bond between individuals (van Baaren, Holland, Kawakami & van Knippenberg, 2004). Pupil mimicry might serve a similar purpose as demonstrated by different studies showing a link to empathy and trust (Simms, 1967; Harrison et al., 2006; Kret et al., 2015). Based on these findings, it was predicted that depressed participants fail to mimic their partners’ pupil size and this might be the cause of their impaired social abilities. Even though results showed intact general pupil mimicry in depressed individuals, there seems to be a

dampened effect of pupil dilation mimicry in the depressed group. The ability to quickly respond to salient external cues has been attributed to an evolutionarily old subcortical route for

processing of emotional information (Tamietto et al., 2009). Possibly, pupil mimicry works via this route as well (Kret et al., 2014). However, in depressed participants, this process seems to be

(21)

decelerated. Compared to healthy controls, depressed patients pupils increase in size when much later when watching partners with dilating pupils. This could indicate that the subcortical route that might be involved in pupil mimicry is intact in depression, because pupil mimicry takes place, but that the process is slowed down. This implies a dysfunctional coupling between pupil mimicry and the actual trusting behaviour, a topic that will be elaborated on in the next

paragraph.

Relationship between Pupil Mimicry and Trusting Behaviour

In general, both groups invested more when mimicking dilating pupils than when

mimicking constricting pupils. Furthermore, whereas control participants did not show different trust behaviours for dilation mimicry and no dilation mimicry trials, depressed participants trusted more when they mimicked a partner’s dilating pupil.

Kret et al. (2015) suggest that the human decision to trust someone with dilating pupils is induced by our own pupil size: by mirroring another’s pupil size, the mimicker catches the emotions of others more strongly, which facilitates emotional understanding. Although depressed individuals synchronized with their partners’ pupils, they trusted partners with dilating pupils just as much as partners with constricting pupils, which suggests that depressed individuals’ own pupil size does not help them to differentiate between trustworthy and less trustworthy partners as much as it helps control participants. Yet, when only looking at dilation mimicry trials, it can be seen that even though trust levels in the depressed group were generally lower than in the control group, depressed participants show increased trust behaviour in dilation mimicry trials. This in turn indicates that although depressed participants trust less, mimicking a partners pupil size helps them to infer trust and eventually trust a partner more.

(22)

One possible explanation for the dissociation between physiological reaction to trusting signals and actual psychological trusting behaviour could lie in a dysfunctional cerebral stimulus processing due to altered activation pattern in specific brain regions that are important for social judgement, emotion recognition and perspective taking. Previous research in healthy individuals has shown that the neural systems supporting the observation of changes in pupil size and

subsequent mimicry involves somatosensory systems as well as brain regions that are involved in social and cognitive appraisal (i.e., Harrison et al., 2006). Even though it has not been

investigated in depression yet – there might be a link between the behavioural shortcomings in depression and dysfunctional activation of networks that are related to social and emotional decision making. Several brain areas show an abnormal activation pattern in depressive individuals (i.e. Drevets, Price, & Furey, 2008; Johnstone, van Reekum, Urry, Kalin, & Davidson, 2007). For example, the amygdala and the mPFC show abnormally high activation levels, which might be a default signal not to trust a partner (Drevets, 1998; Davidson,

Pizzagalli, Nitschke, & Putnam, 2002; Koenigs & Grafmann, 2009). Furthermore, Johnstone et al. (2007) propose an ineffective contribution of prefrontal regulatory circuitry in depression because trusting someone involves the suppression of amygdala activity. Moreover it is noteworthy, that regions that contribute to Theory of Mind (ToM) are involved in pupil

processing as well (e.g. Harrison et al., 2006). Again, activation in these regions is abnormal in depression, which might be the cause of overly self-focused behaviour with a limited ability to empathise and perspective taking (Cusi et al. 2013; Inoue et al., 2004; Rochat et al. 2012).

Therefore, it could be speculated that this dysregulation could also contribute to the difference in trust interference from pupils between the depressed and the control group because depressed individuals fail to predict their partners’ intentions. Earlier studies indicating that

(23)

people with depressive disorder are not impaired in their ability to decode mental states per se, but rather fail to draw valid conclusions about the mental states of others and their possible intentions, support this hypothesis (Wolkenstein, Schönenberg, Schirm, & Hautzinger, 2011). Alternatively, it could be imagined that people with depression can interpret the intentions of others very well, but do not want social interaction in order to avoid negative outcomes or failure. A number of different kinds of studies found that individuals with major depressive disorder report a lack of confidence in their relationships and a belief that relationships will fail (i.e. Hammen & Brennan, 2001; Lewinshohn, Rohde & Seeley, 1998; Clark, Thorne, Hardy & Cropsey, 2013).

Even though these brain abnormalities could be one possible explanation for the lower trust investments in the depressed group while there is intact general pupil mimicry, it does not fully explain why the link between dilation mimicry and trust is greater in depressed individuals than in healthy controls. As mentioned before, the brain structures that are necessary to mimic a partner’s pupil size seem to be preserved but the processes seem to be slowed down. It might be that when mimicking the dilating pupil of a partner, the immediate reaction to distrust gets down-regulated and helps them to trust more. Other than healthy controls, they need their own pupil size as a signal to trust. In line with this assumption is the indication that dilation mimicry in depressed participants starts later. As there are no differences in reaction times between dilation mimicry and no-dilation mimicry trials, the underlying process must be so fast that it is not visible in reaction times. Future research will be needed to see whether these findings can be replicated and if so what the underlying processes are.

(24)

A major limitation of the current study is the fact that most depressed subjects used antidepressant medication that can affect norepinephrine levels, which is closely linked with pupil dilation (e.g., Gould, Altamirano, Javors, & Frazer, 2006; Nieuwenhuis & Jepma, 2011, Gabay, Pertzoy, & Henik, 2011), and amygdala activation which is crucial for both pupil

mimicry and trust inference (Sheline et al., 2001; Harrison et al., 2009). Therefore, future studies may try to include non-medicated depressed patients (cf. Sheline et al., 2001) or a group with sub-clinical depressive symptoms.

The finding that depressed individuals actually do trust more when mimicking dilating pupils shows that the ability to infer trust from their own pupil size is not impaired in general. This could have important implications for possible clinical interventions. For example, depressed individuals might benefit from interventions that stimulate eye contact. Previous studies already highlighted the role of poor social skills in the development and course of

depression (i.e. Segrin, 2000). Yet, including the pupil in such interventions and helping them to mimic others pupils, has not been done yet. Therefore, future studies are needed to tests this

possibility. Furthermore, in order to confirm the proposed relationship between the dissociation

of pupil mimicry, trusting behaviour and different brain regions, it will be crucial to explore the actual neural background of pupil mimicry in depressed individuals as well as other

psychological disorders such as anxiety disorder. Conclusion

The current study explored the link between a lack of pupil mimicry and the deficit in interpersonal trust in depressed individuals. Results further confirm lower trust levels in depressed individuals. Yet, the link between pupil dilation mimicry and trust is stronger in depressed individuals than in control individuals: mimicking dilating pupils helps depressed

(25)

participants to evaluate a partner’s trustworthiness. These findings further confirm the important role of pupil size and pupil mimicry in interpersonal trust and social interaction.

(26)

References

Abell, F., Happe, F., & Frith, U., (2000). Do triangles play tricks? Attribution of mental states to animated shapes in normal and abnormal development. Cognitive Development, 15, 1-16. Amemiya, S., & Ohtomo, K. (2011). Effect of the observed pupil size on the amygdala of the

beholders. Social Cognitive and Affective Neuroscience, 7, 332-41.

Bagiella, E., Sloan, R.P., Heitjan, D.F. (2000) Mixed-effects models in psychophysiology. Psychophysiology, 37, 13–20.

Baron-Cohen, S. (1989). The autistic child’s theory of mind: a case of specific developmental delay. Journal of Child Psychology and Psychiatry 30, 285–297.

Benhua, X., Yanxiang, P., 2004. Research on the relationship between interpersonal trust and depression of undergraduate. Journal of Clinical Psychosomatic Diseases, 2, 106-108.

Berg, J., Dickhaut, J., & McCabe, K. (1995). Trust, reciprocity, and social history. Games and

economic behavior, 10, 122-142.

Brendan Clark, C., Thorne, C.B., Hardy, S., & Cropsey, K.L. (2013). Cooperation and depressive symptoms. Journal of Affective Disorders, 150, 1184-1187.

Clark, C. B., Thorne, C. B., Hardy, S., & Cropsey, K. L. (2013). Cooperation and depressive symptoms. Journal of Affective Disorders, 150, 1184–1187. doi:10.1016/j.jad.2013.05.011 Cusi, A.M., Nazarov, A., Macqueen, G.M., & McKinnon, M.C., (2013.) Theory of mind deficits

in patients with mild symptoms of major depressive disorder. Psychiatric Research, 210, 672-674.

Davidson, R. J., Pizzagalli, D., Nitschke, J. B., & Putnam, K. (2002). Depression: perspectives from affective neuroscience. Annual Review of Psychology, 53, 545–574.

(27)

Demos, K.E., Kelley, W.M., Ryan, S.L., Davis, F.C. & Whalen, P.J. (2008). Human amygdala sensitivity to the pupil size of others. Cerebral Cortex, 18, 2729–2734.

Destoop, M., Schrijvers, D., De Grave, C., Sabbe, B., & De Bruijn, E. R. A. (2012). Better to give than to take? Interactive social decision-making in severe major depressive disorder. Journal of Affective Disorders, 137, 98–105. doi:10.1016/j.jad.2011.12.010.

Doherty, P. F., White, G. C., & Burnham, K. P. (2012). Comparison of model building and selection strategies. Journal of Ornithology, 152, 317–323. doi:10.1007/s10336-010-0598-5.

Drevets, MD, W. C. (1998). Functional neuroimaging studies of depression: the anatomy of melancholia. Annual review of medicine, 49, 341-361.

Drevets, W. C., Price, Æ. J. L., & Furey, M. L. (2008). Brain structural and functional abnormalities in mood disorders  : implications for neurocircuitry models of depression. Brain Structure and function, 213, 93–118. doi:10.1007/s00429-008-0189-x

Dunn, J.R. & Schweitzer, M.E. (2005). Feeling and believing: The influence of emotion on trust. Journal of Personality and Social Psychology, 88, 736-748. DOI:

10.1037/0022-3514.88.5.736

First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. (2012). Structured Clinical Interview

for DSM-IV® Axis I Disorders (SCID-I), Clinician Version, Administration Booklet.

American Psychiatric Pub.

Gabay, S., Pertzov, Y., & Henik, A. (2011). Orienting of attention, pupil size, and the norepinephrine system. Attention, Perception & Psychophysics, 73, 123–9. doi:10.3758/s13414-010-0015-4.

(28)

Goldwater, B. C. (1972). Psychological significance of pupillary movements. Psychological Bulletin, 77, 340–355. doi:10.1037/h0032456.

Gould, G. G., Altamirano, A. V., Javors, M. A., & Frazer, A. (2006). A comparison of the chronic treatment effects of venlafaxine and other antidepressants on serotonin and norepinephrine transporters. Biological psychiatry, 59, 408-414.

Granholm, E., & Steinhauer, S. R. (2004). Pupillometric measures of cognitive and emotional processes. International Journal of Psychophysiology, 52, 1–6.

doi:10.1016/j.ijpsycho.2003.12.001

Hamilton, M. (1960). A rating scale for depression. Journal of Neurosurgery, 23, 56-63.

Hammen, C. & Brennan, P.A. (2001). Depressed adolescents of depressed and non- depressed mothers: tests of an interpersonal impairment hypothesis. Journal of Consulting and Clinical Psychology 69, 284.

Harrison, N. a, Singer, T., Rotshtein, P., Dolan, R. J., & Critchley, H. D. (2006). Pupillary contagion: central mechanisms engaged in sadness processing. Social Cognitive and Affective Neuroscience, 1, 5–17. doi:10.1093/scan/nsl006

Harrison, N.A., Gray, M.A, & Critchley, H.D. (2009). Dynamic pupillary exchange engages brain regions encoding social salience. Social Neuroscience, 4, 233–243.

doi:10.1080/17470910802553508

Hess, E. H., & Polt, J. M. (1960). Pupil size as related to interest value of visual stimuli. Science, 132, 349-350.

Hess, E. H. (1975). The role of pupil size in communication. Scientific American, 233, 110-2. Hoffman, K. L., Gothard, K. M., Schmid, M. C., & Logothetis, N. K. (2007). Facial-expression

(29)

Hox, J. J. (2002). Multilevel analysis: Techniques and applications. West Sussex, England: Routledge.

Inoue, Y., Tonooka, Y., Yamada, K., & Kanba, S. (2004). Deficiency of Theory of Mind in patients with remitted mood disorder. Journal of Affective Disorders, 82, 403-409. Janisse, M.P. (1977). Pupillometry. Washington, DC: Hemishpere.

Johnstone, T., van Reekum, C. M., Urry, H. L., Kalin, N. H., & Davidson, R. J. (2007). Failure to regulate: counterproductive recruitment of top-down prefrontal-subcortical circuitry in major depression. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 27, 8877–84. doi:10.1523/JNEUROSCI.2063-07.2007

Kim, S. S., Chung, Y., Perry, M. J., Kawachi, I., & Subramanian, S. V. (2012). Association between interpersonal trust, reciprocity, and depression in South Korea: A prospective analysis. PLoS ONE, 7. doi:10.1371/journal.pone.0030602

Koenigs, M., & Grafman, J. (2009). The functional neuroanatomy of depression: distinct roles for ventromedial and dorsolateral prefrontal cortex. Behavioural brain research, 201, 239-243.

Kret, M. E., Fischer, A. H., & de Dreu, C. K. W. (2015). Pupil mimicry correlates with trust in in-group partners with dilating pupils. Psychological Science, 1-10.

doi:10.1177/0956797615588306.

Kret, M. E., & Ploeger, A. (2015). Emotion processing deficits: A liability spectrum providing insight into comorbidity of mental disorders. Neuroscience and Biobehavioral Reviews, 52, 153–171. doi:10.1016/j.neubiorev.2015.02.011

(30)

pupil size and pupil mimicry in the development of trust- an fMRI study. Manuscript in preparation.

Kret, M. E., Stekelenburg, J. J., Roelofs, K., & de Gelder, B. (2013). Perception of face and body expressions using electromyography, pupillometry and gaze measures. Frontiers in

Psychology, 4, 1-12. doi:10.3389/fpsyg.2013.00028.

Kret, M. E., Tomonaga, M., & Matsuzawa, T. (2014). Chimpanzees and humans mimic pupil-size of conspecifics. PloS One, 9, e104886. doi:10.1371/journal.pone.0104886

Laeng, B. & Falkenberg, L. (2007). Women’s pupillary response tp sexually signifcant others during hormonal cycle. Hormones and Behaviour, 52, 520-530.

Laeng, B., Sirois, S., & Gredeback, G. (2012). Pupillometry: A Window to the Preconscious? Perspectives on Psychological Science, 7, 18–27. doi:10.1177/1745691611427305 Lee, L., Harkness, K. L., Sabbagh, M. A., & Jacobson, J. A. (2005). Mental state decoding

abilities in clinical depression. Journal of Affective Disorders, 86, 247–258. doi:10.1016/j.jad.2005.02.007

Lester, D. & Gatto, J.L. (1990). Interpersonal trust, depression, and suicidal ideation in teenagers. Psychological Reports, 67, 786-790.

Lewinsohn, P.M., Rohde, P., & Seeley, J.R., (1998). Major depressive disorder in older

adolescents: prevalence, risk factors, and clinical implications. Clinical Psychology Review, 18, 765–794.

Lewis, J.D. & Weigert, A.J. (1985). Trust as a social reality. Social Forces, 63, 967-985. Loewenfeld, I. (1993). The pupil: Anatomy, physiology, and clinical applications.Detroit,

(31)

Mistry, N.H. (2005). Looking beyond the eye: Pupillary dilation and ist relationship to

conversational behaviour in computer-mediated conversation and face-to-face conversation. Unpublished doctoral dissertation. doi:10.1017/CBO9781107415324.004.

Muris, P., Meesters, C., van Melick, M., & Zwambag, L. (2001). Self-reported attachment style, attachment quality, and symptoms of anxiety and depression in young adolescents. Personality and Individual Differences, 30, 809-818.

Nieuwenhuis, S., & Jepma, M. (2011). Investigating the role of the noradrenergic system in human cognition. Decision Making, Affect, and Learning: Attention and Performance, 23, 1–25. doi:10.1093/acprof:oso/9780199600434.003.0017

Oosterhof, N.N. & Todorov, A. (2008). The functional basis of face evaluation. Proc Natl Acad Sci USA, 105, 11087-92. doi: 10.1073/pnas.0805664105

Ottenbreit, N. D., Dobson, K. S., & Quigley, L. (2014). An examination of avoidance in major depression in comparison to social anxiety disorder. Behaviour Research and Therapy, 56, 82–90. doi:10.1016/j.brat.2014.03.005

Peavler, W.S. & McLaughlin, J.P. (1967). The question of stimulus content and pupil size. Psychonomic Science, 8, 505-506.

Radke, S., Güths, F., André, J. a, Müller, B. W., & de Bruijn, E. R. a. (2014). In action or inaction? Social approach-avoidance tendencies in major depression. Psychiatry Research, 219, 513–517. doi:10.1016/j.psychres.2014.07.011.

Ratcliffe, M. (2014). The phenomenology of depression and the nature of empathy. Medicine, Health Care and Philosophy, 17, 269–280. doi:10.1007/s11019-013-9499-8.

(32)

Rochat, L., Billieux, J., & Van der Linden, M. (2012). Difficulties in disengaging attentional resources from self-generated thoughts moderate the link between dysphoria and maladaptive self-referential thinking. Cognition and Emotion, 26, 748-757.

Rubinow, D. R., & Post, R. M. (1992). Impaired recognition of affect in facial expression in depressed patients. Biological Psychiatry, 31, 947–953. doi:10.1016/0006-3223(92)90120-O.

Segrin, C. (2000). Social skills deficits associated with depression. Clinical Psychology Review, 20, 379–403. doi:10.1016/S0272-7358(98)00104-4.

Sheline, Y. I., Barch, D. M., Donnelly, J. M., Ollinger, J. M., Snyder, A. Z., & Mintun, M. A. (2001). Increased amygdala response to masked emotional faces in depressed subjects resolves with antidepressant treatment: An fMRI study. Biological Psychiatry, 50, 651–658. Silk, J., Dahl, R., Neal, R., Forber, E., Axelson, D., Birmaher, B., & Siegle, G. (2008). Pupillary

Reactivity to Emotional Information in Child and Adolescent Depression: Links to Clinical and Ecological Measures. American Journal of Psychiatry, 164, 1873–1880.

doi:10.1176/appi.ajp.2007.06111816.

Simms, M. S. (1967). Pupillary response of male and female subjects to pupillary difference in male and female picture stimuli. Perception & Psychophysics, 2, 553-555.

Snellen, H. (1862). Letterproeven tot Bepaling der Gezigtsscherpte. Utrecht: PW van der Weijer. Tamietto, M., Castelli, L., Vighetti, S., Perozzo, P., Geminiani, G., Weiskrantz, L. & de Gelder,

B. (2009). Unseen facial and bodily expressions trig- ger fast emotional reactions. [case reports research support, non-U.S. gov’t]. Proc. Natl. Acad. Sci. U.S.A. 106, 17661– 17666.

(33)

Tomasello, M., Hare, B., Lehmann, H., & Call, J. (2007). Reliance on head versus eyes in the gaze following of great apes and human infants: the cooperative eye hypothesis. Journal of Human Evolution, 52, 314–320. doi:10.1016/j.jhevol.2006.10.001

Van Baaren, R. B., Holland, R. W., Kawakami, K., & van Knippenberg, A. (2004). Mimicry and prosocial behavior. Psychological Science  : A Journal of the American Psychological Society, 15, 71–74. doi:10.1111/j.0963-7214.2004.01501012.x

West, B. T., Welch, K. B., Galecki, A.T. (2006) Linear Mixed Models: A Practical Guide Using Statistical Software: Chapman and Hall/CRC.

Wiseman, R. & Watt, C. (2010). Judging a book by its cover: The unconscious influence of pupil size on consumer choice. Perception, 39, 1417-1419.

Wolkenstein, L., Schönenberg, M., Schirm, E., & Hautzinger, M. (2011). I can see what you feel, but I can’t deal with it: Impaired theory of mind in depression. Journal of Affective

(34)

Appendix A Bivariate Correlations between all continuous variables.

1. 2. 3. 4. 5. 6. 7. 8. 9.

1. Baseline corrected pupil size

2. Baseline Pupil Size -.028

3. Reaction times .004 .025 4. HAMD scores -.194 .102 .033 5. EQ scores .118 -.024 .096 -.048 6. ToM Intentionality .096 -.079 -.308** -.005 -.208 7. ToM Appropriateness .078 -.099 -0.120 -.143 -.246* .590** 8. GD Intentionality .172 .059 -.106 -.118 -.114 .385** .357** 9. GD Appropriateness .003 -.108 -.042 -.053 -0181 0.311** .263* .432** * p < .05 ** p < .01

(35)

Appendix B

Graph B1. Mean trust investments for constricting, static and dilating pupil partners for each group separately. Error bars indicate ±1 SE. (*** p < .001)

Graph B2. Comparison of the two groups on mean reaction time for

constricting, static and dilating pupil partners. Error bars indicate ±1 SE. (* p < .05)

(36)

Appendix C

Pupil size in depressed and control subjects in reaction to partners with constricting and static pupils. Both groups do not show significant pupil constriction mimicry.

Time after stimulus presentation

M ea n B as el in e C o rr ec te d Pu p il Si ze

(37)

Appendix D

This appendix includes the multilevel models of all analyses. Additional covariates of level of arousal (baseline pupil size), reaction time, as well as ToM and GD intentionality and

appropriateness scores were included. Yet, none of these covariates could explain the significant Group X Partner’s Pupil interaction. Furthermore, HAMD and EQ scores were included to explore whether they are linked to trust investments on their own. Again, no significant effects were found.

Table D1. Multilevel models for trust interference.

Source df1 df2 F p

A) Trust Inference from Partner's Pupil Size1

Corrected model 5 4392 13.420 0.000

Group 1 4392 3.907 0.051

Partner’s Pupil 2 4392 27.067 0.000

Group * Partner’s Pupil 2 4392 4.139 0.016

B) Trust Inference from Partner’s Pupil Size if Level of Arousal included1

Corrected Model 6 4390 11.694 0.000

Group 1 4390 1.919 0.166

Partner’s Pupil 2 4390 26.896 0.000

Level of arousal 1 4390 2.935 0.087

C) Trust Inference from Partner’s Pupil Size if RT included1

Group 1 4421 3.773 0.052

Partner’s Pupil 2 4421 28.088 0.000

RT 1 4421 0.024 0.877

D) Trust Inference from Partner’s Pupil Size HAMD instead of group included1

HAMD 1 4260 0.873 0.350

Partners’s Pupil 2 4260 20.023 0.000

HAMD * Partner’s Pupil 1 4260 1.747 0.174

E) Trust Inference from Partner’s Pupil Size if EQ scores included1

(38)

Partner’s Pupil 2 4314 19.556 0.000 Partner’s Pupil * EQ 2 4314 2.677 0.069 F) Reaction Time to Partner's Pupil Size2

Group 1 4392 1.857 0.173

Partner’s Pupil 2 4392 2.430 0.088

Group * Partner’s Pupil 2 4392 2.862 0.057 G) Trust Inference from Partner's Pupil Size ToM Intentionality included1

Group 1 4367 3.055 0.081

Partner’s Pupil 2 4367 27.627 0.000

Group * Partner’s Pupil 2 4367 4.179 0.015 ToM Intentionality 1 4367 0.393 0.531 E) Trust Inference from Partner's Pupil Size ToM Appropriateness included1

Group 1 4367 2.197 0.138

Partner’s Pupil 2 4367 27.627 0.000

Group * Partner’s Pupil 2 4367 4.174 0.015 ToM Appropriateness 1 4367 1.411 0.235 F) Trust Inference from Partner's Pupil Size GD Intentionality included1

Group 1 4367 3.136 0.077

Partner’s Pupil 2 4367 27.627 0.000

GD Intentionality 1 4367 0.160 0.689

G) Trust Inference from Partner's Pupil Size GD Appropriateness included1

Corrected model 11 4367 11.655 0.000

Group 1 4367 3.083 0.079

Partner’s Pupil 2 4367 27.627 0.000

Group * Partner’s Pupil 2 4367 4.179 0.015 GD Appropriateness 1 4367 2.073 0.150 1_{Dependent variable: Trust (operationalized as investment)}

2_{Dependent variable: Reaction Time}

Table D2. Multilevel Models of Pupil Mimicry.

Source df1 df2 F p

A) Pupil Mimicry

Intercept 1 77.061 139.610 0.000

Group 1 76.951 0.009 0.923

Partner’s Pupil 2 4705.171 6.086 0.002

Group * Partner’s Pupil 2 4705.121 0.973 0.378

(39)

quad 1 78.684 28.627 0.000 Partner’s Pupil * lin 2 50257.495 10.943 0.000 B) Pupil Mimicry if Level of Arousal (baseline pupil size) included

Intercept 9 14.186 70.567 0.000

Group 1 71.35 70.567 0.280

Partner’s Pupil 2 4694.489 70.567 0.002 Level of arousal 1 574.674 70.567 0.000 Group * Partner’s Pupil 2 4694.277 70.567 0.379

lin 1 78.157 70.567 0.000

quad 1 78.663 70.567 0.000

Partner’s Pupil * lin 2 49930.363 70.567 0.000 C) Pupil Mimicry HAMD Scores included

Intercept 1 76.714 93.250 0.000

HAMD 1 76.654 0.211 0.647

Partner’s Pupil 2 4614.853 6.366 0.002

HAMD * Partner’s Pupil 2 4612.996 1.167 0.311

lin 1 77.781 63.654 0.000

qaud 1 77.850 25.592 0.000

Partner’s Pupil * lin 2 51770.350 15.594 0.000 D) Pupil Mimicry EQ scores included

Intercept 1 75.980 57.078 0.000 EQ 1 75.916 0.133 0.717 Partner’s Pupil 1 4639.145 2.084 0.125 EQ * Partner’s Pupil 2 4642.762 0.044 0.957 lin 1 77.207 50.052 0.000 qaud 1 77.583 32.526 0.000

Partner’s Pupil * lin 2 49602.316 11.350 0.000 E) Pupil Mimicry RT included

Intercept 1 82.295 131.145 0.000

Group 1 76.923 0.008 0.930

Partner’s Pupil 1 4703.768 6.148 0.002

RT 1 4752. 271 1.163 0.281

lin 1 77.888 51.660 0.000

qaud 1 78.685 28.667 0.000

Partner’s Pupil * lin 2 50245.561 10.944 0.000 F) Pupil Mimicry ToM Intentionality included

Intercept 1 76.689 24.546 0.070

Group 1 76.170 0.021 0.885

Partner’s Pupil 2 4705.453 6.094 0.002

ToM Intentionality 1 76.578 3.114 0.082 Group * Partner’s Pupil 2 4705.409 0.971 0.379

(40)

quad 1 78.671 28.622 0.000 Partner’s Pupil * lin 2 50258.839 10.943 0.000 G) Pupil Mimicry ToM Appropriateness included

Intercept 1 76.020 11.453 0.001

Group 1 75.935 0.014 0.906

Partner’s Pupil 2 4705.207 6.088 0.002

ToM Appropriateness 1 76.948 0.009 0.924 Group * Partner’s Pupil 2 4705.157 0.972 0.378

lin 1 77.888 51.664 0.000

quad 1 78.679 28.659 0.000

Partner’s Pupil * lin 2 50257.552 10.944 0.000 H) Pupil Mimicry GD Intentionality included

Intercept 1 75.882 3.621 0.061

Group 1 76.000 0.018 0.895

Partner’s Pupil 2 4705.216 6.088 0.002

GD Intentionality 1 75.878 0.061 0.805

lin 1 77.888 51.664 0.000

quad 1 78.679 28.660 0.000

Partner’s Pupil * lin 2 50257.586 10.944 0.000 I) Pupil Mimicry GD Appropriateness included

Intercept 1 75.882 3.960 0.050

Group 1 76.005 0.025 0.874

Partner’s Pupil 2 4705.224 6.087 0.002

GD Appropriateness 1 75.875 0.637 0.427 Group * Partner’s Pupil 2 4705.174 0.972 0.378

lin 1 77.888 51.664 0.000

quad 1 78.681 28.647 0.000

Partner’s Pupil * lin 2 50257.580 10.944 0.000 Dependent Variable: Baseline-corrected pupil size.

Table D3. Multilevel Models of Pupil Dilation Mimicry.

Source df1 df2 F p

A) Pupil Dilation Mimicry

Corrected Model 6 70.570 19.526 0.000

Group 1 70.570 0.012 0.912

Partner’s Pupil 1 70.570 10.395 0.001

lin 1 70.570 62.875 0.000

quad 1 70.570 24.195 0.000

Partner’s Pupil * lin 1 70.570 17.872 0.000 B) Pupil Dilation Mimicry if Level of Arousal (baseline pupil size) included

(41)

Corrected Model 7 70.569 17.336 0.000

Group 1 70.569 0.438 0.508

Partner’s Pupil 1 70.569 10.594 0.001

Level of arousal 1 70.569 4.197 0.040

lin 1 70.569 50.502 0.000

quad 1 70.569 28.496 0.000

Partner’s Pupil * lin 1 70.569 10.964 0.000 C) Pupil Dilation Mimicry EQ scores included

Corrected Model 6 70.570 19.212 0.000 EQ 1 70.570 0.222 0.638 Partner’s Pupil 1 70.570 3.598 0.058 EQ * Partner’s Pupil 1 70.570 0.068 0.794 lin 1 70.570 62.875 0.000 quad 1 70.570 24.197 0.000

Partner’s Pupil * lin 1 70.570 17.867 0.000 D) Pupil Dilation Mimicry RT included

Corrected Model 7 70.569 16.775 0.000

Group 1 70.569 0.011 0.917

Partner’s Pupil 1 70.569 10.475 0.001

RT 1 70.569 0.269 0.604

lin 1 70.569 62.876 0.000

quad 1 70.569 24.194 0.000

Partner’s Pupil * lin 1 70.569 17.871 0.000 E) Pupil Dilation Mimicry ToM Intentionality included

Corrected Model 7 70.569 17.073 0.000

Group 1 70.569 0.010 0.920

Partner’s Pupil 1 70.569 10.417 0.001

lin 1 70.569 62.874 0.000

quad 1 70.569 24.178 0.000

Partner’s Pupil * lin 1 70.569 17.873 0.000 F) Pupil Dilation Mimicry ToM Appropriateness included

Corrected Model 7 70.569 16.736 0.000

Group 1 70.569 0.013 0.910

Partner’s Pupil 1 70.569 10.396 0.001

lin 1 70.569 62.875 0.000

quad 1 70.569 24.184 0.000

(42)

G) Pupil Dilation Mimicry GD Intentionality included

Corrected Model 7 70.569 16.738 0.000

Group 1 70.569 0.017 0.897

Partner’s Pupil 1 70.569 10.396 0.001

GD Intentionality 1 70.569 0.019 0.889

lin 1 70.569 62.875 0.000

quad 1 70.569 24.187 0.000

Partner’s Pupil * lin 1 70.569 17.872 0.000 H) Pupil Dilation Mimicry GD Appropriateness included

Corrected Model 7 70.569 16.782 0.000

Group 1 70.569 0.024 0.877

Partner’s Pupil 1 70.569 10.395 0.001

GD Appropriateness 1 70.569 0.334 0.564 Group * Partner’s Pupil 1 70.569 2.155 0.142

lin 1 70.569 62.875 0.000

quad 1 70.569 24.175 0.000

Partner’s Pupil * lin 1 70.569 17.872 0.000 Dependent Variable: Baseline-corrected pupil size.

Table D4. Multilevel Models of Pupil Constriction Mimicry.

Source df1 df2 F p

A) Pupil Constriction Mimicry

Corrected Model 6 70.524 11.602 0.000

Group 1 70.524 0.041 0.840

Partner’s Pupil 1 70.524 0.282 0.595

lin 1 70.524 35.694 0.000

quad 1 70.524 24.698 0.000

cub 1 70.524 6.116 0.013

B) Pupil Constriction Mimicry if Level of Arousal (baseline pupil size) included

Corrected Model 7 70.523 11.751 0.000

Group 1 70.523 1.302 0.254

Partner’s Pupil 1 70.523 0.356 0.551

Level of arousal 1 70.523 12.804 0.000

lin 1 70.523 35.694 0.000

quad 1 70.523 24.526 0.000

cub 1 70.523 6.115 0.013

C) Pupil Constriction Mimicry HAMD Scores included

(43)

HAMD 1 70.111 0.111 0.739

Partner’s Pupil 1 70.111 0.645 0.422

HAMD * Partner’s Pupil 1 70.111 1.070 0.301

lin 1 70.111 46.502 0.000

quad 1 70.111 24.528 0.000

cub 1 70.111 5.889 0.015

D) Pupil Constriction Mimicry EQ scores included

Corrected Model 6 70.524 11.475 0.000 EQ 1 70.524 0.085 0.771 Partner’s Pupil 1 70.524 0.121 0.728 EQ * Partner’s Pupil 1 70.524 0.000 0.985 lin 1 70.524 35.694 0.000 quad 1 70.524 24.707 0.000 cub 1 70.524 6.117 0.000

E) Pupil Constriction Mimicry RT included

Corrected Model 7 70.523 10.073 0.000

Group 1 70.523 0.036 0.849

Partner’s Pupil 1 70.523 0.309 0.578

RT 1 70.523 0.902 0.342

lin 1 70.523 35.688 0.000

quad 1 70.523 24.701 0.000

cub 1 70.523 6.120 0.013

F) Pupil Constriction Mimicry ToM Intentionality included

Corrected Model 7 70.523 10.428 0.000

Group 1 70.523 0.003 0.959

Partner’s Pupil 1 70.523 0.282 0.596

lin 1 70.523 35.693 0.000

quad 1 70.523 24.679 0.000

cub 1 70.523 6.117 0.013

G) Pupil Constriction Mimicry ToM Appropriateness included

Corrected Model 7 70.523 9.944 0.000

Group 1 70.523 0.049 0.824

Partner’s Pupil 1 70.523 0.281 0.596

lin 1 70.523 35.694 0.000

quad 1 70.523 24.687 0.000

cub 1 70.523 6.116 0.013

H) Pupil Constriction Mimicry GD Intentionality included