Expectation Effects on Emotional Processing

Background:

Research demonstrated that positive expectations are related to a perception bias towards positive information and changes in neural networks of affective processing. However, it's unclear how underlying expectancy effects modulate response biases to emotional inputs. A recent study showed the importance of cognitive control and top-down regulation in expectation effects on emotional processing, mechanisms that may be less effective in cognitively vulnerable individuals, such as those with depression. Research on placebo analgesia indicates that not only such higher level-processes but also lower-level processes play a significant role in placebo effects. How such lower level processes are involved in affective placebo effects is relatively unknown but would be highly relevant given that long-term modulation of the emotional system also relies on computationally less costly bottom-up processes typically shaped by learning and experiences. Therefore, this study aims to investigate the neurobehavioral mechanisms of how expectations and prior experiences modulate emotional processing.

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 study days. On the screening day, in addition to assessing eligibility, participants will receive a saline nasal spray that is introduced as oxytocin. Treatment expectations will be induced via an established expectation induction protocol that systematically manipulates expectations in the context of face perception. Positive expectations about "oxytocin" will be induced via a video documentary detailing the mechanisms of oxytocin, highlighting its mood-enhancing effects and role in emotional processing. In detail, participants will complete a short training of an emotion classification task, view the video documentary, self-administer the nasal spray, rate their expectations using a visual analogue scale (VAS, from "no expected mood enhancement" to "strong expected mood enhancement"), and then complete the same training again. However, the second training is covertly manipulated and is designed to facilitate the detection of happy facial expressions. At the end of the day they will rate their experience with a VAS (from "experienced no positive mood change" to "experienced a large positive mood change"). This expectation induction combined with the covertly manipulated training is already implemented on the screening day to establish positive prior experiences associated with "oxytocin" and therefore enhance placebo effects on the study days. If a participant does not believe in the treatment on the screening day, they will not be included for the main study.

In a cross-over experiment, on two study days that are one week apart, participants will undergo fMRI scanning. Before scanning, participants will receive a saline nasal spray that is either introduced as oxytocin (placebo condition) or saline (control condition) in a counter-balanced design. Expectations will be reinforced via the video documentary. After the video, participants will receive verbal instructions about their assigned condition for that day (either "oxytocin" or control), self-administer the nasal spray, and rate their expectations. Depending on the assigned condition, participants will then either complete a standard training (control condition) or a covertly manipulated training ("oxytocin" condition). Participants will assess their current mood (using a VAS) at three time points: before nasal spray administration, after the manipulated training and after the fMRI scan. At the end of each study day, participants will rate their experience. While in the scanner, an emotion classification task is performed, in which participants label images of subtle emotional face expressions, varying in emotional intensity, as happy, fearful, or neutral. As a control measure, horizontal gaze trajectories (fixation and gaze shifts patterns) will be recorded using eye-tracking.

Hypotheses:

The investigators will examine the effects of positive expectations on mood state and task accuracy and investigate how positive expectations modulate uni- and multivariate fMRI signal patterns associated with the processing and categorization of emotional stimuli. More specifically, they hypothesize the following:

• Positive expectations enhance mood and improve the accuracy of detecting subtle happy facial expressions.
• fMRI patterns in sensory, affective, and cognitive areas in response to subtle happy faces are more similar to patterns evoked by unambiguous happy face localizers in the positive treatment expectation condition.
• Pattern dissimilarity across happiness intensities decreases in the placebo condition.
• Expectation effects on mood, happy face accuracy, and neural patterns are related to expectation ratings and measures of prior experience induced by the manipulated training.

Analysis plan:

For the behavioral analysis, statistical protocols developed in our lab will be applied, with a main focus on response threshold and sensitivity derived from psychometric functions. The effects of expectation on mood state and task accuracy and post-treatment experience will be assessed using statistical tests from the general linear model framework, including t-tests, repeated measures ANOVA and Pearson correlations. Statistical significance will be assumed based on an alpha value of 0.05. Emotion-specific psychometric response functions will be derived from regressing individual accuracy rates from the placebo condition on the control condition. The psychometric response functions will allow us to disentangle expectation effects with respect to response tendencies (intercept) and perceptual sensitivity (slope).

For the univariate fMRI analysis, a two-level random effects general linear model (GLM) approach will be employed, modeling task-related activity using the hemodynamic response function (HRF). Whole-brain analysis (cluster-level corrected) and a small-volume regions of interest (ROI) approach will be applied (p < .05 FWE corrected). A multivariate representational similarity analysis (RSA) approach will be used to compare the similarity of fMRI patterns during subtle emotional face presentation with patterns derived from unambiguous emotional face localizers. RSA protocols have previously been developed in our lab.

The here presented study is part of the collaborative research center (CRC) SFB/TRR289 and is funded by the Deutsche Forschungsgemeinschaft (DFG, ID: 422744262).