Neural underpinnings of open-label placebo effects in emotional distress

The study investigates whether open-label placebos (OLPs)—placebos administered without deception—produce genuine psychobiological effects on emotional distress, and identifies their neural underpinnings. Although OLPs have shown benefits across symptoms such as irritable bowel syndrome, depression, pain, and anxiety, it remains unclear if these effects reflect objective biological changes or response bias, since OLP paradigms cannot be double-blinded. Prior neuroimaging work has mapped conventional (deceptive) placebo effects to networks including DLPFC, ACC, insula, amygdala, and periaqueductal gray, but it is unknown whether OLPs engage similar mechanisms. The present work uses two experiments—one behavioral and one fMRI—to test whether OLPs reduce self-reported emotional distress to negative images and to identify associated brain activity. The authors hypothesized reduced distress with OLPs (study 1) and engagement of placebo-related networks (study 2), particularly regions implicated in affect modulation (e.g., orbitofrontal cortex, ACC, PAG).

The introduction reviews foundational and recent literature on placebo effects, including ethical limitations of deceptive placebos and emerging evidence for OLP efficacy across conditions (pain, depression, anxiety, IBS). It highlights that few studies have examined objective physiological outcomes of OLPs, with mixed results (no effects on wound healing; nuanced effects on allergic responses moderated by belief; EEG evidence for reduced emotional distress). Mechanistic accounts for conventional placebos emphasize conditioning, expectations, and social interaction; neuroimaging implicates a descending modulatory network (DLPFC, ACC, insula, somatomotor regions, amygdala, PAG), especially in placebo analgesia and emotional placebo responses. Whether OLPs recruit similar neural mechanisms remains unclear, motivating the current fMRI study focused on emotional distress induced by negative images.

Study 1 (behavioral): 112 healthy participants (mean age 23.53 ± 7.32 years; 67 females) without neurological or psychiatric history provided informed consent. Participants were randomly assigned to OLP or control. OLP group: read a presentation describing placebo effects and evidence that non-deceptive placebos can work; then received a labeled placebo saline nasal spray and were told it could reduce negative reactions if they believed it would. Control group: read a presentation about neurological pain processes; received the same saline nasal spray described as needed for better physiological readings; were unaware of the placebo focus. Task: Participants viewed 30 negative and 10 neutral IAPS images (balanced for normative valence/arousal; identical set to Guevarra et al.) presented for 6000 ms each, followed by a fixation (4000 ms) and a 5000 ms rating period using a 1–9 Likert scale (1 = not at all negative, 9 = very negative). Post-task, participants rated expectation that the spray reduced their distress (VAS) and evaluation of the presentation (six quality items). Statistical analysis: mixed-factorial ANOVA with group (OLP vs control) as between-subjects factor and picture type (negative, neutral) as within-subjects factor; post-hoc t-tests as appropriate. Study adhered to the Declaration of Helsinki and local ethics approval.

Study 2 (fMRI): Independent sample of 44 healthy female participants (22.34 ± 2.62 years), randomized to OLP or control, with identical pre-scan manipulation (presentations and nasal spray). fMRI task: 45 negative and 45 neutral pictures (from Guevarra et al. plus additional IAPS images), each presented for 4 s followed by 12 s fixation, across three blocks. During inter-block breaks, participants received an additional nasal spray application (total 6 applications). To avoid introspection-related confounds, affect ratings were collected at the end via a 4-button Likert scale (1 = not at all negative, 4 = very negative), not trial-by-trial. Post-scan questionnaires assessed expectation (spray effectiveness), belief in OLPs, general belief in placebos, dispositional optimism (LOT-R), trait anxiety (STAI), social desirability (SES), presentation quality, and experimenter warmth/competence.

Imaging acquisition: 3T Siemens Tim Trio (six participants on 3T Prisma Fit, evenly distributed across groups). Structural T1 MP-RAGE (TR=1650 ms, TE=5 ms). Functional EPI: TR=2000 ms, TE=35 ms, flip angle 80°, FOV 224 mm, 32 slices, voxel size 3.125×3.125 mm. Stimuli presented on a screen; participants viewed via mirror on head coil.

Preprocessing and analysis: SPM12 pipeline including realignment, sinc interpolation, normalization to MNI space (3 mm isotropic voxels), and 8 mm FWHM Gaussian smoothing. First-level GLM modeled negative vs neutral images (hemodynamic response function). Second-level random-effects analyses compared contrasts between OLP and control. Whole-brain results reported at voxelwise p<0.001 uncorrected. A priori ROI small-volume corrections (SVC; p<0.05 FWE within ROIs) for bilateral amygdala, PAG (5 mm spheres), right orbitofrontal cortex, right insula, ACC, bilateral DLPFC, and bilateral hippocampi (10 mm spheres). For the general negative>neutral contrast (collapsed across groups), amygdala ROIs were also tested (5 mm sphere, FWE p<0.05). Anatomical labels via SPM Anatomy Toolbox.

Behavioral/statistical measures: Group comparisons (t-tests) for demographics, belief measures, expectations, and questionnaire scores; correlations between self-reported distress and expectancy/belief; correlations between placebo-related brain activations and expectancy, belief in OLPs, general belief in placebos, and personality variables.

Study 1: Self-reported emotional distress showed a robust main effect of picture type (F(1,110)=742.41, p<0.001, partial η²=0.87). There was a significant group × picture type interaction (F(1,110)=5.54, p=0.020, partial η²=0.05). Post-hoc: For negative pictures, OLP reduced distress relative to control (control: 5.51 ± 1.56; OLP: 4.85 ± 1.49; t(110)=2.30, p=0.011 one-sided, Cohen’s d=0.43). No group difference for neutral pictures (t(110)=-0.03, p>0.10). OLP group endorsed higher expectancy that the spray reduced distress (t(110)=-2.05, p=0.021), but expectancy strength did not correlate with distress (p>0.10).

Study 2: Groups did not differ in demographics or questionnaire measures (optimism, anxiety, social desirability, general belief in placebos), nor in presentation quality ratings or perceived warmth/competence of the experimenter. Manipulation check: OLP group reported higher belief in OLPs than controls (t(42)=2.41, p=0.011). No group difference in post-experiment expectancy that the spray reduced emotional responses (t(42)=0.10, p>0.10).

Behavior: OLP reduced self-reported emotional distress (controls: 2.04 ± 0.62; OLP: 1.60 ± 0.76 on 1–4 scale; t(42)=2.09, p=0.021, Cohen’s d=0.63). In the OLP group, distress did not correlate with expectancy (r=-0.16), belief in OLPs (r=0.20), general belief in placebos, or optimism (all p>0.10). Controls likewise showed no expectancy–distress correlation.

fMRI: Negative>neutral (collapsed across groups) activated right amygdala (ROI, FWE p<0.05), and regions including superior temporal gyrus (whole-brain p<0.001 uncorrected). OLP>control revealed increased activation in PAG and bilateral hippocampi (ROI-SVC, FWE p<0.05). Lowering threshold to p<0.005 uncorrected additionally revealed ACC activation; no prefrontal cortex activation was observed even at p<0.01 uncorrected. Control>OLP showed no significant clusters at p<0.001 uncorrected; at p<0.005 uncorrected, left supramarginal gyrus emerged. Placebo-related activations (PAG, hippocampi) were negatively associated with felt distress in the OLP group (Supplementary), and were not related to expectancy or belief in OLPs (all p>0.10). They were positively associated with general belief in placebos (PAG: r=0.61, p=0.002; right hippocampus: r=0.37, p=0.085; left hippocampus: r=0.40, p=0.056). No relationships with optimism, anxiety, or social desirability.

The findings demonstrate that open-label placebos reduce emotional distress elicited by highly negative images and that this effect is accompanied by engagement of brain regions implicated in affect modulation, including PAG, hippocampi, and ACC. These neural markers argue against a mere response bias explanation and support genuine psychophysiological modulation by OLPs. The involvement of PAG aligns with known placebo-related descending modulatory systems and potential opioid-mediated mechanisms; its role in fear and distress circuitry suggests top-down modulation of limbic responses. Hippocampal engagement, particularly anterior regions, points to modulation of anxiety-related states and interactions with amygdala and ACC within emotion-regulatory circuitry. Notably absent was prefrontal activation (e.g., DLPFC), a hallmark of conventional (deceptive) placebo effects tied to expectancy-driven cognitive control. The lack of prefrontal engagement and absence of correlations with expectancy or belief in OLPs suggest that OLP mechanisms may rely less on explicit expectation and more on lower-level or non-prefrontal modulatory processes. The observed relationship between placebo-related neural activation and general belief in placebos indicates that broad placebo beliefs, rather than acute expectancy, may facilitate OLP efficacy. Overall, the results indicate both overlap and divergence between OLP and conventional placebo neurobiology, supporting distinct yet partially shared mechanisms.

This work provides behavioral and neuroimaging evidence that open-label placebos can reduce emotional distress and engage affect-modulatory brain regions (PAG, hippocampi, ACC) without recruiting prefrontal cortex. The findings suggest that OLP effects are not solely response bias and may operate through mechanisms less dependent on explicit expectations. Contributions include the first fMRI characterization of OLP neural substrates in emotional distress and identification of links to general placebo belief. Future research should broaden samples (sex, age, clinical populations), incorporate richer affective rating instruments, test additional stress paradigms (e.g., Montreal Imaging Stress Task), examine neurochemical mechanisms (e.g., opioidergic involvement), and directly compare OLPs to conventional placebos to delineate mechanistic differences.

• Study 2 included only young female students, limiting generalizability by sex, age, and educational background. - Emotional distress ratings in the fMRI study used a coarse 4-point scale collected post-run rather than trial-wise; richer measures (e.g., SAM) could provide finer granularity. - The stress-induction paradigm relied on highly negative images; results should be replicated using other stress tasks (e.g., Montreal Imaging Stress Task). - OLP expectancy in study 2 was assessed after a delay outside the scanner, which may have reduced sensitivity to immediate expectancy effects.