IntroductionObservational learning has been shown to be involved in both pain induction and modulation 11Consequences of caring: Pain in the human context.,12Vicarious influences on pain-threshold determinations.,17Goubert L Vlaeyen JWS Crombez G Craig KD Learning about pain from others: an observational learning account., which has important consequences for the development, prevention and management of various pain syndromes 17Goubert L Vlaeyen JWS Crombez G Craig KD Learning about pain from others: an observational learning account.. Growing evidence shows that observational learning is also effective in producing placebo effects. In previous studies, placebo analgesia 10Placebo analgesia induced by social observational learning.,13Egorova N Park J Orr SP Kirsch I Gollub RL Kong J Not seeing or feeling is still believing: conscious and non-conscious pain modulation after direct and observational learning.,19Hunter T Siess F Colloca L Socially induced placebo analgesia: a comparison of a pre-recorded versus live face-to-face observation.,29The neural processes of acquiring placebo effects through observation.,32The Effect of the Type and Colour of Placebo Stimuli on Placebo Effects Induced by Observational Learning. and nocebo hyperalgesia 31The effect of the sex of a model on nocebo hyperalgesia induced by social observational learning.,34Kröner-Herwig B: Nocebo hyperalgesia induced by social observational learning.,35Vögtle E Kröner-Herwig B Barke A Nocebo hyperalgesia: contributions of social observation and body-related cognitive styles. were powerfully induced by observing another person experiencing analgesia or hyperalgesia, respectively. The high efficiency of observational learning in modeling placebo effects has meaningful implications for clinical trials and medical practice 7Responding to nocebos through observation: social contagion of negative emotions..Modeling of placebo effects is shown to be effective not only when the model's behavior is observed directly but also when it is displayed in a video recording 13Egorova N Park J Orr SP Kirsch I Gollub RL Kong J Not seeing or feeling is still believing: conscious and non-conscious pain modulation after direct and observational learning.,29The neural processes of acquiring placebo effects through observation.,34Kröner-Herwig B: Nocebo hyperalgesia induced by social observational learning.,35Vögtle E Kröner-Herwig B Barke A Nocebo hyperalgesia: contributions of social observation and body-related cognitive styles.. Bajcar and Bąbel's integrative model of placebo effects induced by observational learning 3How Does Observational Learning Produce Placebo Effects? A Model Integrating Research Findings. emphasizes the effectiveness of these sources of social learning and the importance of the observer's individual characteristics when modeling placebo effects. In some research, the empathic concern of the observer correlated positively with the magnitude of observationally induced placebo effects 10Placebo analgesia induced by social observational learning.,19Hunter T Siess F Colloca L Socially induced placebo analgesia: a comparison of a pre-recorded versus live face-to-face observation.,22Beyond conformity: Social influences on pain reports and physiology.,26Piedimonte A Benedetti F Guerra G Zamfira DA Vighetti S Carlino E I expect what you expect: An electrophysiological study on social expectation of pain.,31The effect of the sex of a model on nocebo hyperalgesia induced by social observational learning.; however, other studies did not show such a relationship 32The Effect of the Type and Colour of Placebo Stimuli on Placebo Effects Induced by Observational Learning.,34Kröner-Herwig B: Nocebo hyperalgesia induced by social observational learning.,35Vögtle E Kröner-Herwig B Barke A Nocebo hyperalgesia: contributions of social observation and body-related cognitive styles.. Previous research also indicates that the observer's catastrophizing 34Kröner-Herwig B: Nocebo hyperalgesia induced by social observational learning. and optimism 22Beyond conformity: Social influences on pain reports and physiology. are possible factors that may affect the magnitude of observationally learned placebo effects.In addition to the observer's characteristics, social learning theory emphasizes the importance of the model's features. Previous research has shown that the sex of the model is a significant factor that influences the magnitude of nocebo hyperalgesia induced by observational learning: a male model was found to induce a greater effect than a female model, regardless of the observers’ sex 31The effect of the sex of a model on nocebo hyperalgesia induced by social observational learning.. To the best of our knowledge, no other characteristics of models have yet been examined as factors that potentially affect observationally induced placebo effects.Research findings from other fields of study have shown that confidence expressed in finding a solution to a problem can affect the effectiveness of observational learning. The model's confidence statements significantly increased the observer's degree of persistence on different tasks 39Effects of model persistence and statements of confidence on children's self-efficacy and problem solving.. Moreover, self-confidence determines whether or not one is a good and effective leader 21Leadership: do traits matter?. and is associated with better ability to motivate others 37Self-Confidence: A Concept Analysis.. Therefore, self-confidence seems to attract other people's attention and keep it on the individual, what, according to the social learning theory , is an important factor in acquiring new behaviors through the observation. Thus, the main aim of the study was to investigate the effect of a model's self-confidence on observationally acquired placebo analgesia. We hypothesized that a self-confident model would elicit a more robust placebo effect than an unself-confident model. Our secondary aim was to verify whether the observer's self-esteem and self-efficacy influence observational learning. Self-esteem is a positive or negative attitude towards the self, a kind of global self-assessment 28Society and the Adolescent Self-Image.. Study by Weiss 36Social learning of work values in organizations. found that subordinates who had low levels of self-esteem displayed greater compliance of values ​​with their supervisor. In turn, self-efficacy was defined by Bandura 4Self-efficacy: Toward a unifying theory of behavioral change. as people's beliefs about their capabilities to produce designated levels of performance. People's behavior is driven by expectations about situation, result of an action and self-efficacy. What is more, both these traits may affect task-based functioning and are related to the type and frequency of activities undertaken 8The effects of self-esteem, success–failure, and self-consciousness on task performance.,25Locke EA Frederick E Lee C Bobko P Effect of self-efficacy, goals, and task strategies on task performance.,27Rank J Nelson NE Allen TD Xu X Leadership predictors of innovation and task performance: Subordinates’ self-esteem and self-presentation as moderators.,33Examining the relationships between self-efficacy, task-relevant attentional control, and task performance: Evidence from event-related brain potentials.. We hypothesized that high self-efficacy and high self-esteem of participants would predict weaker placebo analgesia. Furthermore, previous research on placebo analgesia induced by observational learning indicates that the placebo effect did not extinguish over time 10Placebo analgesia induced by social observational learning.,13Egorova N Park J Orr SP Kirsch I Gollub RL Kong J Not seeing or feeling is still believing: conscious and non-conscious pain modulation after direct and observational learning.. Since only a few studies have tested the extinction of observationally induced placebo analgesia, our additional aim was to verify whether these results would also replicate in our study.Materials and methods Study designThe study consisted of three groups: self-confident model (SCM), unself-confident model (UCM) and the control group. During the experiment, participants received a series of pain stimuli preceded by a blue or orange color displayed on the computer screen and rated their intensity. Beforehand, a video recording of a male model undergoing the same experimental procedure was displayed in the SCM and UCM groups. The model shown on the recording in the SCM group behaved in a self-confident manner, while the same model in the UCM group behaved in an unself-confident manner. In both groups, the model rated pain stimuli in the placebo condition as less painful, and in the non-placebo condition as more painful. One of these conditions was signalized by blue and another by orange color (counterbalanced), which were displayed on the computer screen right before the application of pain stimulus. No videos were presented to participants in the control group. The overall design is presented in Figure 1.

Figure 1Study design. Participants were randomly assigned to one of the three groups: self-confident model, unself-confident model, or control. In the calibration phase, the tactile sensation (t) and pain (T) thresholds were determined for each participant. Then, participants from groups 1 and 2 watched a video recording with a model (self-confident or unself-confident, respectively) undergoing the experimental procedure. The model received a series of pain stimuli preceded by blue or orange stimuli and then rated pain intensity on the NRS. Half of the participants watched the model rating stimuli preceded by orange as more painful than stimuli preceded by blue; the other half watched the model rating stimuli preceded by orange as less painful than those preceded by blue. All groups went through the testing phase, which consisted of 16 pain stimuli at the same intensity of 2T: half of them preceded by orange, half by blue. Participants rated pain intensity on the NRS. Then, the participants filled in the Rosenberg Self-Esteem Scale (SES) and the General Self-Efficacy Scale (GSES).

 ParticipantsA total of 60 healthy volunteers, including 36 females (60%), participated in the study (mean age = 23.65 ± 2.54). Each of the three groups consisted of 20 randomly assigned participants, including 12 females. All participants were physically and mentally healthy, free of pain, and were not taking any type of pain medication. None of them had taken part in any previous pain-related studies. The exclusion criteria were age below 18 or over 35; pain during the last month; any regular medication, including non-prescription drugs; use of illegal drugs; history of any respiratory, circulatory, neurological, musculoskeletal and/or psychiatric disorders; current symptoms of depression and/or anxiety. The criteria were based on those proposed by Gierthmühlen and collaborators 16Gierthmühlen J Enax-Krumova EK Attal N Bouhassira D Cruccu G Finnerup NB Haanpää M Hansson P Jensen TS Freynhagen R Kennedy JD Mainka T Rice ASC Segerdahl M Sindrup SH Serra J Tölle T Treede R-D Baron R Maier C Who is healthy? Aspects to consider when including healthy volunteers in QST-based studies-a consensus statement by the EUROPAIN and NEUROPAIN consortia..

Participants were informed that they were participating in a study on pain mechanisms and that they would receive painful electrocutaneous stimulation. Having read the description of the experimental procedures, participants gave their informed, written consent to participate in the study. They were also informed that they could withdraw their consent at any time without providing a reason. When the study was completed, all participants were fully debriefed and informed about the actual aim of the experiment. Participants were not financially compensated, although they received some sweets and the possibility to draw a voucher to the bookstore. The study protocol was approved by the Research Ethics Committee at the Institute of Psychology, Jagiellonian University, Kraków, Poland (decision no. KE/03/052019).

 Sample sizeSample size calculation was performed using G*Power statistical software 14Faul F Erdfelder E Buchner A Lang A-G Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses., based on data extracted from a previous experiment conducted by Świder and Bąbel 31The effect of the sex of a model on nocebo hyperalgesia induced by social observational learning.. The calculation was based on the effect size (f = 0.29) derived from the comparison of the green- against red-associated NRS ratings between the female and male model condition. It was estimated that the minimum sample size required in each group was 16 (alpha = 0.05, 80%). However, to account for potential dropouts and to increase the power to detect the effect, 20 participants were examined in each group. Stimuli Pain stimuli

Electrocutaneous pain stimuli were delivered to the inner side of the non-dominant forearm of the participant through 2 durable stainless-steel disk electrodes that were 8 mm in diameter with 30 mm spacing. The electrocutaneous stimuli were square pulses with a duration of 200 μs, delivered by a Constant Current High Voltage Stimulator (Digitimer, Welwyn Garden City, England, Model DS7AH). The intensity of the electrocutaneous stimuli was set individually for each participant according to the calibration procedure.

 Color stimuli

The pain stimuli were preceded by the presentation of color stimuli (orange or blue) displayed in full-screen mode on a computer screen (17 inches, resolution 1280 × 1024) placed in front of the participant at a distance of approximately 50 cm. Color slides were presented according to a predetermined pseudorandom sequence. The experimental procedure was fully automatized with PsychoPy software, which integrates stimuli application and data collection in real time.

 Video recordingsThe video recordings were presented in the observation phase of the study. All of them showed the same male model rating 16 pain stimuli aloud on the NRS scale. The pain stimuli were preceded by 8 orange and 8 blue slides presented in a pseudorandom order. Half of the participants were randomly assigned to the condition in which the model rated stimuli preceded by orange as more painful (7 to 9 on the NRS) than stimuli preceded by blue (2 to 4 on the NRS); the other half were in the opposite condition, in which stimuli preceded by orange were rated as less painful (2 to 4 on the NRS) than those preceded by blue (7 to 9 on the NRS). This way, participants had an opportunity to associate each of the colors (blue or orange) with either high or low pain ratings. The different level of self-confidence between groups was manifested in the body posture and facial expressions of the model, as well as in the specific behavior accompanying the assessment of pain. In the SCM group, the person presented in the video recording had upright body posture, a loud voice, and was making quick pain assessments without hesitation. In the UCM group, the person presented on the video recording showed closed body posture, uncertain voice intonation, and visible hesitation before pain ratings. Previous research has shown that the male model induces stronger effect than a female model, regardless of the participants’ sex 31The effect of the sex of a model on nocebo hyperalgesia induced by social observational learning., thus only a male model was employed in this study. Measures

Pain intensity ratings were obtained by means of an 11-point numerical rating scale (NRS), ranging from 0 = “no pain” to 10 = “the most intense pain that is tolerable”.

Another 11-point numerical rating scale (“0” = the observed person does not possess this feature, “10” = the observed person definitely possesses this feature) was used by the participants to assess the level of the observed person's following features: self-confidence, charisma, competence, and attractiveness. This measure of the model's self-confidence was used later in the regression analysis.

At the end of the experiment, the participants were asked to complete the following questionnaires:1)Adjective Check List (ACL) 18The adjective check list manual., which originally is a psychological assessment containing 300 adjectives used to identify common psychological traits. For this study, the list was adjusted and limited to 30 adjectives, 10 of which were related to self-confidence or its opposite (e.g. courageous, shy), and 20 of which were unrelated to self-confidence (e.g. intelligent, sad). In this study, the ACL served as a manipulation check measure. The participants' task was to choose adjectives that, in their opinion, best described the model seen on the video recording. Then, the frequency with which a particular adjective was chosen in the SCM and UCM groups was compared.2)Rosenberg Self-Esteem Scale (SES) 28Society and the Adolescent Self-Image., which is a unidimensional, 10-item scale that assesses global self-worth by measuring both positive and negative feelings about the self. All items are answered using a 4-point Likert scale ranging from “strongly agree” to “strongly disagree”. The total score ranges between 10 and 40 points; a higher score indicates higher self-esteem.3)General Self-Efficacy Scale (GSES) 30Generalized Self-Efficacy scale., which is a unidimensional, 10-item scale that measures optimistic beliefs about the self that clearly relate to personal agency. All items are answered using a 4-point Likert scale ranging from “not at all true” to “exactly true”. The total score ranges between 10 and 40 points, with a higher score indicating higher self-efficacy. ProcedureThe experiment consisted of three phases: calibration, observation, and testing. All groups underwent calibration and testing phases, while the observation phase was absent in the control group (Fig. 1). Calibration phase

During the calibration phase, the tactile sensation threshold (t) and pain threshold (T) were individually determined for each participant. The calibration procedure consisted of two ascending series of electrocutaneous stimuli, starting at 1 mA. The intensity of the stimuli was increased by 1 mA every 5 seconds until the participant detected the first tactile sensation (t) and then until the sensations became painful (T), which was clearly stated by the participant. The stimulus intensity for the rest of the experiment was calculated as a doubled mean of the pain threshold (2T mA).

 Observation phase

The observation phase took place only in the SCM and UCM groups. In this phase, the participants observed one of the video recordings described in the stimuli section above. Participants were informed that the purpose of watching the video was to learn how to use the pain intensity scale. In order to ensure that participants paid attention to both the pain ratings and the colors accompanying the pain stimuli, they were asked to write down the color of the slide and the model's rating on a special form.

 Testing phase

During this phase, each participant received and rated the pain intensity of 16 pain stimuli at an intensity of 2T mA, preceded by 8 orange and 8 blue stimuli presented in a pseudorandom order. Participants rated the pain intensity aloud, which was consistent with the procedure observed on the video recording. A single trial consisted of 1) a color slide displayed for 9 seconds, 2) a pain stimulus applied 7 seconds after the beginning of the trial with the color slide still visible, and 3) the NRS scale displayed on the same color slide.

 Manipulation check and control questions

After the testing phase was completed, participants were asked a series of questions to determine whether they had figured out the actual aim of the study. Participants were asked 1) whether the presented colors were linked to pain intensity, 2) whether the presented colors were linked to the pain ratings provided by the model, 3) whether observing the model facilitated their subsequent pain ratings, 4) whether the model's specific behavior or traits affected their pain ratings. To determine whether participants perceived the model as self-confident or unself-confident, they filled in the ACL list and then, rated the level of self-confidence, charisma, competence, and attractiveness of the model on the 0-10 NRS. Afterwards, they completed the psychological traits questionnaires (SES, GSES).

 Data reduction and statistical analysis

Participants rated 16 pain stimuli preceded by two colors (blue and orange). The color preceding the 8 pain stimuli that was rated by the model as less painful is subsequently referred to as a placebo, while the color preceding the 8 stimuli that was rated by the model as more painful is referred to as a non-placebo. The distinction between placebo and non-placebo reflects the two within-subject conditions to which participants were exposed during the experiment. Moreover, to explore the possible extinction of the placebo effect, the division into the first half of trials and the second half of trials was introduced. The first half of trials consisted of the first 4 stimuli from the placebo condition and the first 4 stimuli from the non-placebo condition. Respectively, the second half of trials consisted of the second 4 stimuli from the placebo condition and the second 4 stimuli from the non-placebo condition. The distinction between the first half and the second half of trials reflects one of the within-subject conditions in the further analyses.

Descriptive statistics (means and SDs) were calculated for the following variables: age, tactile and pain thresholds, self-esteem measured by SES, and general self-efficacy measured by GSES. To investigate if there were any between-group differences in the level of these variables, a one-way ANOVA design with ‘group’ (SCM, UCM, and control group) as an independent variable was conducted.

The chi-square test was performed to determine whether the SCM and UCM groups differed significantly in terms of the number of specific model characteristics chosen by participants on the ACL. The Student's t-test was performed to compare mean ratings of perceived model traits (self-confidence, charisma, competence, and attractiveness) between the SCM and UCM groups.

The main analysis was performed on participants' NRS pain ratings using the General Linear Model (GLM) repeated-measures procedure, with ‘group’ (SCM, UCM, and control) as a between-subject factor and ‘rating’ (placebo and non-placebo) and ‘trials’ (first half and second half) as within-subject factors. In order to verify the main hypothesis, the F-tests were followed by planned comparison tests. Differences in pain ratings between placebo versus non-placebo conditions in the SCM and UCM groups were compared with the difference between the two control conditions in the control group. Similarly, the differences between placebo and non-placebo pain ratings in the SCM and UCM groups were compared. Several additional planned comparisons on NRS ratings were performed to explore the extinction of placebo analgesia. Differences in placebo versus non-placebo pain ratings were compared within the first half and the second half trials in each group.

To verify whether the perceived level of self-confidence of the observed person or the individual traits of the participants affected the magnitude of placebo analgesia, a forward stepwise multiple regression analysis was performed in the experimental groups. Participants’ ratings of the self-confidence of a model as well as scores on the GSES and SES questionnaires were used as independent variables, while the difference between placebo and non-placebo pain ratings was a dependent variable.

The level of significance was set at p < .05. All the analyses were conducted using STATISTICA data analysis software, version 13 (StatSoft Inc., Tulsa, OK, USA).

Results Descriptive statistics and manipulation check analysesThe one-way ANOVA revealed that there were no differences between the groups in age, tactile threshold, pain threshold, self-esteem measured by SES, and general self-efficacy measured by GSES. Descriptive statistics for each group separately and for the whole sample are presented in Table 1.

Table 1Characteristics of participants and descriptive statistics for dependent variables in each group and in total (means and standard deviations)

Notes. SCM = Self-confident model; UCM = Unself-confident model; t = Tactile threshold; T = Pain threshold; SES = Rosenberg Self-Esteem Scale; GSES = General Self-Efficacy Scale; NRS = Numerical Rating Scale.

The chi-square tests showed that in the SCM group the observed person was significantly more often perceived as relaxed (χ2(2,N = 40) = 10.16, p < 0.01), decisive (χ2(2,N = 40) = 10.16, p < 0.01), and self-confident (χ2(2,N = 40) = 10.16, p < 0.01), compared with the UCM group. Moreover, the observed person in the UCM group was significantly more often perceived as shy (χ2(2,N = 40) = 8.48, p = 0.01) compared with the SCM group. The frequency of the remaining adjectives from the ACL did not differ significantly between experimental groups. The Student's t-test revealed a significant difference in the mean ratings of the perceived self-confidence of a model between the SCM and UCM groups (t(38)= 3.37, p < 0.01). Mean ratings of the other examined characteristics (charisma, attractiveness, and competence) did not differ significantly between the experimental groups. These results indicate that participants properly differentiated the model's level of self-confidence, thereby confirming the effectiveness of the experimental manipulation. Moreover, the other control questions confirmed that all participants noticed the link between colors and higher or lower pain ratings of the model. What is more, participants often claimed that observing the model facilitated their subsequent pain ratings (62.5%), but no one admitted that the model's specific behavior or traits affected their pain ratings.

 Primary analyses

The repeated-measures GLM on the pain ratings revealed a statistically significant main effect for “rating” (F(1, 57) = 10.89, p < 0.01, η2 = 0.16), “trials” (F(1, 57) = 4.08, p = 0.05, η2 = 0.07) and the “rating” x “group” interaction (F(2, 57) = 4.83, p = 0.01, η2 = 0.14). The “trials” x “group” (F(2, 57) = 1.34, p = 0.27, η2 = 0.04), “rating” x “trials” (F(1, 57) = 0.15, p = 0.71, η2 < 0.01) and “rating” x “trials” x “group” (F(2, 57) = 0.61, p = 0.55, η2 = 0.02) interactions were not statistically significant.

The between-group planned comparison of the difference in pain ratings between the placebo and the non-placebo condition in the SCM group compared with the difference in pain ratings between the two control conditions from the control group showed a significant effect (F(1, 57) = 6.56, p = 0.01, η2 = 0.10). A similar comparison between the UCM and the control group also revealed a significant difference in pain ratings (F(1, 57) = 7.87, pη2 = 0.12), indicating that participants in both experimental groups experienced less pain in the placebo condition than in the non-placebo condition, compared to the control group. The SCM and UCM groups did not differ significantly in the magnitude of the difference in pain ratings between the placebo and non-placebo conditions (F(1, 57) = 0.06, p = 0.81, η2 = 0.001) (Fig. 2).