Frontopolar theta oscillations link metacognition with prospective decision making

The study investigates whether frontopolar cortex (FPC) activity, specifically theta-band oscillations, causally contributes to metacognitive judgments in value-based decision making and how such metacognitive processes influence prospective decisions that guard against future temptations. Prior work established neural correlates of subjective value in prefrontal regions but provided limited causal evidence linking FPC to metacognition in economic choices. Conflicting findings from TMS studies in perceptual domains suggested that disrupting FPC might improve metacognition, leaving the functional role of FPC debated. The authors hypothesize that enhancing FPC theta oscillations improves metacognitive sensitivity (accurate confidence reports reflecting objective decision uncertainty) and increases prospective self-control via precommitment, particularly when there is a high risk of preference reversals. They further predict a positive relationship between explicit metacognitive ability (retrospective confidence accuracy) and the willingness to precommit in anticipation of future preference changes.

Neuroeconomic research has mapped subjective value signals to ventromedial and dorsolateral prefrontal cortices, among others. Metacognition—accurate introspective confidence reports—has been associated correlationally with FPC activity in value-based and perceptual tasks. However, its causal role remains debated, with some perceptual TMS studies reporting improved metacognition after FPC disruption. Theoretical accounts propose that metacognition facilitates precommitment by enabling awareness of susceptibility to preference reversals. FPC has also been implicated in precommitment behavior. Theta oscillations in prefrontal networks relate to decision difficulty and confidence, and cross-frequency coupling suggests a role for theta in integrating distributed information, potentially enabling FPC to read out confidence signals from VMPFC/DLPFC. This work builds on these findings to test whether entraining FPC theta rhythms causally enhances metacognition and affects prospective self-control choices.

Design: Within-subjects experiment with three stimulation conditions: theta tACS (5 Hz), gamma tACS (80 Hz, active control), and sham. Thirty-seven participants (after exclusion of one due to side effects; age 19–31, mean 22.9; 17 female) completed both tasks under all stimulation conditions.

Stimulation: Transcranial alternating current stimulation at 2 mA peak-to-peak. Electrode montage: active 5×7 cm over Fpz (frontopolar cortex) and reference 10×10 cm over CPz (10–20 system). Current modeling (SimNIBS 2.1) indicated maximal field in FPC and minimal under reference. Each mini-block included 15 s ramp-up, ~160 s stimulation, 5 s ramp-down (sham ramped down after ramp-up). Blocks were separated by 30 s breaks. Discomfort and flicker ratings were collected and included as covariates.

Tasks:

• Confidence accuracy task (retrospective metacognition): Intertemporal choices between smaller-sooner (SS; 0–10 CHF today, step 1 CHF) and larger-later (LL; fixed 10 CHF, delay 1–180 days at 1, 10, 20, 40, 80, 120, 180). After each choice, participants rated confidence (0–20). Total 180 trials, organized into 9 mini-blocks of 20 trials (three per tACS condition; order counterbalanced).
• Precommitment task (prospective self-control): Choice between binding precommitment to LL (10 CHF at 29–208 days) versus postponement option that preserves a later choice between SS (0–10 CHF at 28 days) and the same LL (e.g., 7 CHF in 28 days vs 10 CHF in 68 days). If postponing, participants were re-contacted after 28 days for a final decision with adjusted delays. Total 90 trials in 3 mini-blocks (one per tACS condition; order counterbalanced).

Modeling and analysis: Individual time preferences were estimated per tACS condition by fitting hyperbolic discounting to choices, yielding discount rate k. Choices were linked to subjective value differences via a softmax function (inverse temperature β). Mixed generalized linear models (MGLMs, lme4 in R) analyzed binary choices with fixed effects for tACS contrasts (theta–sham, theta–gamma), subjective value difference (DV = SV_LL − SV_SS), confidence (z-standardized within subject), and interactions. Metacognitive sensitivity was operationalized as the DV × Confidence interaction predicting choice: stronger interaction indicates better metacognitive readout of decision uncertainty. For the precommitment task, predictors included DV_initial (value difference at initial choice), Reversal risk (DV_final − DV_initial; higher indicates greater likelihood of future preference reversal), tACS contrasts, and interactions, especially Reversal risk × DV_initial and their modulation by tACS. Random effects included participant-specific intercepts and random slopes for fixed effects. Confidence intervals for fixed effects were obtained by parametric bootstrap. Covariates included task order, tACS condition order, block-wise aversiveness, and post-experiment ratings of discomfort/flicker impact. Payment was incentive-compatible with one randomly selected trial paid after its delay; postponed trials were resolved at 28 days.

• Time preferences and confidence bias: Discount rates k did not differ between stimulation conditions (Wilcoxon rank-sum tests, all W < 651, p > 0.72). Mean confidence ratings were unaffected by tACS (paired t-tests, all t < 1, p > 0.67).
• Retrospective metacognition (confidence accuracy task): Participants showed significant metacognitive sensitivity overall (DV × Confidence beta = 2.22, 95% CI = [1.19, 3.43]). Theta tACS increased metacognitive sensitivity relative to sham (tACStheta-sham × DV × Confidence beta = 2.57, CI = [1.32, 4.34]) and relative to gamma (tACStheta-gamma × DV × Confidence beta = 1.39, CI = [0.37, 2.80]). Results held after controlling for mean confidence per condition. Gamma vs sham also showed an increase (gamma-sham × DV × Confidence beta = 1.14, CI = [0.31, 2.24]), but effects were significantly weaker than theta.
• Prospective decision making (precommitment): Theta tACS increased sensitivity of precommitment choices to Reversal risk relative to sham (tACStheta-sham × Reversal risk beta = 2.05, CI = [0.04, 6.24]), especially when participants initially preferred LL (tACStheta-sham × DV_initial × Reversal risk beta = 2.02, CI = [0.45, 4.20]). Relative to gamma, theta increased sensitivity in the three-way interaction (tACStheta-gamma × DV_initial × Reversal risk beta = 2.79, CI = [0.15, 6.48]); the lower-order theta–gamma × Reversal risk interaction was not significant (beta = −0.08, CI = [−2.40, 2.82]). No significant gamma vs sham effects on precommitment.
• Individual differences: Under sham, higher metacognitive sensitivity (DV × Confidence) correlated with greater willingness to precommit as Reversal risk increased (r = 0.61, p < 0.001). Mean confidence showed a positive, not negative, relation to precommitment (r = 0.33, p = 0.05). The theta-induced increase in metacognitive sensitivity predicted the theta-induced increase in precommitment sensitivity to Reversal risk (r = 0.56, p < 0.001).

The findings provide causal evidence that frontopolar theta oscillations support metacognition in value-based decision making. Enhancing FPC theta strengthened the link between objective decision uncertainty (value difference) and subjective confidence without altering time preferences or confidence bias, consistent with FPC acting as a readout hub for confidence information from valuation and control regions (e.g., VMPFC, DLPFC). Although gamma stimulation showed some improvement relative to sham, theta effects were stronger, aligning with theories of theta–gamma coupling and the integrative role of theta in coordinating distributed networks. Critically, improved metacognition translated into prospective behavior: theta stimulation increased sensitivity to the benefits of precommitment, particularly when a future preference reversal was likely, and individual metacognitive ability predicted precommitment propensity. These results link the neural implementation of metacognition to self-control strategies that protect against future temptations, advancing understanding of FPC’s role in both retrospective and prospective judgments. The study also helps reconcile discrepancies with prior TMS work by using frequency-specific entrainment (tACS), active control conditions, and discomfort controls, suggesting that disrupting FPC nonspecifically may yield different outcomes than entraining functional rhythms. Clinically, the results suggest that deficits in frontal theta dynamics could underlie impaired metacognition and prospective control in disorders such as schizophrenia, addiction, and OCD, highlighting potential targets for intervention.

Enhancing frontopolar theta oscillations causally improves metacognitive sensitivity in intertemporal choices and increases willingness to precommit when preference reversals are likely. The work establishes a mechanistic link between FPC theta activity, metacognitive monitoring of preference stability, and prospective self-control. Future research should examine the specificity of these effects with active control sites, determine whether benefits depend on aligning stimulation to individual theta frequencies, disentangle local versus network coherence mechanisms, and test translational potential in populations with metacognitive and self-control deficits.

• Spatial specificity: tACS has low spatial resolution; adjacent regions (dorsomedial PFC, VMPFC, DLPFC) may have been influenced. No active control site was included, preventing strong claims about regional specificity.
• Mechanistic ambiguity: The study cannot distinguish whether effects reflect local FPC activity changes or altered coherence with other regions.
• Sensory side effects: Although controlled and unlikely to impact value-based decisions, phosphenes/discomfort cannot be definitively excluded as contributors.
• Behavioral scope: Precommitment to LL is not universally optimal (e.g., financial constraints), potentially adding variance to behavior.
• Inter-individual variability: Variability in alignment between individual endogenous theta rhythms and fixed 5 Hz stimulation, and general susceptibility to brain stimulation, may modulate effects; this was not directly manipulated.