Positive Expectation Effects on Early Emotional Processing

Overview

Insights into the mechanisms of expectation effects in the emotional domain can be invaluable for the development of potential therapeutic interventions for mood disorders. Recent findings demonstrate that positive expectations alone can induce a positivity bias on the behavioral and the neural level (Baker et al.,2022, Mostauli et al., 2025). This is intriguing given that antidepressant treatments, which have high placebo rates are reported to reduce a negativity bias commonly observed in depression. Research on placebo analgesia has shown that both higher-order cognitive expectations and lower-level learning mechanisms, such as conditioning, play a key role in placebo effects. The role of such lower-level processes is particularly underexplored in affective placebo effects. Closing this gap is crucial, as the long-term modulation of the emotional system likely depends on cognitively less demanding, bottom-up processes shaped by learning and conditioning. This is particularly relevant in clinical populations, where cognitive resources may be limited, but there is an abundance of prior experiences (i.e., learning). The present study therefore investigates how treatment expectations influence early psychophysiological processing of emotional stimuli. Beyond behavioral measures, we will directly assess early neural and attentional mechanisms using sensory event-related potentials (ERPs) and reflexive gaze shifts. By combining EEG and eye-tracking, we aim to identify early markers of expectation effects during emotional processing in healthy participants (N=44 plus 15% potential dropout, 50% women), who perform an emotion classification task. Emotional faces will be presented at varied levels of stimulus visibility (alpha transparency), allowing us to model perceptual sensitivity and response bias without inducing ceiling effects. Treatment expectation will be induced by verbal instructions using an established protocol (e.g. Baker et al., 2022, Mostauli et al., 2025). We hypothesize that positive expectations enhance mood and decrease reaction times paralleled by better accuracy. Further, we hypothesize that positive treatment expectation enhances gaze shifts toward the mouth region which is the primary diagnostic feature for happy faces. On the neural level we expect that positive expectations modulate EEG signal patterns associated with early emotional valence processing. Additionally, whole-brain and time-frequency analyses, as well as representational similarity analyses, are planned to further explore into neural expectation effects and potential changes in the hierarchical representation of emotional processing.

Background Only few experiments have used a learning procedure to reinforce expectations in the affective domain. Our research team recently established a protocol for expectation induction including a sham oxytocin treatment that reliably improved mood and positively influenced emotional perception in healthy individuals. Findings demonstrated that visual attention was positively biased by expectation. It can be assumed that expectations shape an attentional bias about the forthcoming sensory world in early processing states with relatively low cognitive costs. To investigate affective expectation effects on early attentional selection, we will therefore combine the recording of sensory ERPs and reflexive gaze shifts during emotional processing in healthy participants. Recruitment Plan Healthy participants will be recruited through an online advertisement on a voluntary basis. An initial short screening to assess basic eligibility (see inclusion and exclusion criteria) will take place over the phone, followed by an extensive screening on site, including the following questionnaires: * Demographics * Education * Generic rating scale for previous treatment experiences, treatment expectations, and treatment effects (G-EEE) * Beck Depression Inventory (BDI) Design The study consists of three days: one screening day and two days of performing the gaze-shift paradigm while recording EEG- and eye-tracking data. On the screening day, in addition to assessing eligibility, participants will perform one training of a reflexive gaze-shift paradigm. They will briefly view emotional faces (neutral, happy, fearful) that are unpredictably shifted upward or downward, manipulating the initial fixation to be on the eyes or mouth. Participants will classify the depicted emotion. The face stimuli are presented at pre-validated levels of stimulus visibility (alpha transparency). These alpha levels were determined in prior pilot studies to prevent ceiling effects in accuracy and to ensure comparable baseline performance across emotions (necessary for baseline training and the main task). During the baseline training and the main task, stimulus difficulty is adjusted accordingly so that emotional expressions are equally difficult to classify for the three emotions. The next two study days include a counterbalanced cross- over design and are scheduled one week apart. During the experimental sessions, EEG and eye-tracking data are recorded while participants perform the full classification task. Participants start the session by watching a video documentary detailing the mechanisms of oxytocin, highlighting its mood-enhancing effects and role in emotional processing. After watching the video documentary, participants will be informed about their assigned condition for that day (labelled as either oxytocin (Placebo) or saline (control)), self-administer the saline nasal spray, and rate their expectations. Afterwards they perform the main task (N = 204 trials). Happy, fearful and neutral facial stimuli are again presented at different visibility levels, followed by a block of fully visible stimuli (N=48 trials). Mood state is assessed throughout the session using a visual analogue scale (VAS, from "no expected mood enhancement" to "strong expected mood enhancement") at three time points: before nasal spray administration (baseline), before and after completing the task. At the end of each study day, participants will rate their experience and at the end of the third day they will rate their belief in the different treatments. Hypothesis The investigators will examine the effects of positive expectations on mood state, task accuracy, reaction times and eye movements and investigate how positive expectations modulate EEG signal patterns associated with early emotional valence processing. More specifically, they hypothesize the following: * Positive treatment expectation i) improves mood state, ii) decreases reaction time for happy expression identification, iii) enhances accuracy for happy expressions iv) enhances gaze shifts toward the mouth region for happy faces, and/or reduces gaze shifts toward the eye region for fearful faces. * Early ERP responses reflect modulated processing of expectation-congruent - and incongruent sensory input by positive treatment expectation compared to the control condition. * We also expect electrophysiological expectation effects on higher-order (e.g. prefrontal) and general processes (e.g. stimulus anticipation), explored by time-frequency and multivariate analyses. * Treatment effects on task performance and eye movement are related to electrophysiological changes. Analysis Plan Statistical analyses will be conducted using the general linear model framework (factorial designs and linear mixed-effects models) to assess treatment effects on state, task parameters (accuracy, reaction time), gaze shifts and ERPs (peak amplitudes and latencies). In addition, subjective expectation and experience ratings, as well as training performance will be included in separate analyses to address potential interaction effects. Behavioral task outcomes comprise emotion classification accuracy and reaction times. To investigate expectation effects on a response bias and/or discrimination ability, psychometric response functions across different visibility steps will be estimated (similar to Baker et al.,2022; Mostauli et al., 2025). To examine whether expectation has an effect on gaze behavior, we will analyze fixation changes (\>0.5°) that are triggered by the stimulus or occur within a period of 1000 ms following stimulus offset. Trials with fixation changes during the pre-stimulus period (-500 ms to 0 ms) or during stimulus presentation (0-150 ms) will be excluded. Blinks will be interpolated as long as the blink period does not exceed more that 15% of the whole trial. The primary dependent variable will be the proportion of fixation changes directed toward the alternative major facial feature (eyes vs. mouth). The EEG will be recorded from 64 electrodes (LiveAmp, Brain Products) while participants perform the gaze shift paradigm. Raw data will be preprocessed with EEGLAB (Delorme \& Makeig, 2004), epoched from -1000 to 2000 ms and corrected to a 200 ms pre-stimulus baseline. Independent component analysis using IClabel will be performed to detect eye and muscle-related activity in the data. Analyses will focus primarily on early, stimulus-locked neural activity preceding the first saccade. To examine treatment effects on early stimulus-locked neural activity, analyses will focus on established early ERP components within predefined regions of interest (ROIs), including: * P1-related activity at occipital electrodes * N170-related activity at occipito-temporal electrodes In addition to the main task, a functional localizer will be administered to assess neural patterns associated with emotion processing. Exploratory multivariate analyses may be conducted to identify emotion-sensitive spatiotemporal EEG patterns. These analyses may further inform representational similarity analyses (RSA) to examine potential differences in neural representational structure between experimental conditions. In addition to the predefined ROIs and multivariate localizer analyses, exploratory analyses will be conducted across all electrodes and time points to identify potential treatment-related effects in low and higher order networks. All analyses will be corrected for multiple comparisons using appropriate procedures (e.g. FDR or Bonferroni-correction). The calculated power should be sensitive to detect medium effect sizes (d=.40) of the expectation manipulation, including interactions with experimental conditions as well as differences between experimental groups. Assuming an alpha of 5% and a power of 90%, this results in a required sample size of N=44 (G\*Power 3.1). We expect a dropout rate of 15%, resulting in final sample sizes of 51 healthy participants. The here presented study is part of the collaborative research center (CRC) SFB/TRR289 and is funded by the Deutsche Forschungsgemeinschaft (DFG, ID: 422744262).