State anxiety alters the neural oscillatory correlates of predictions and prediction errors during reward-based learning

Hein, Thomas; Ruiz, María Herrojo

doi:10.1016/j.neuroimage.2022.118895

Cited by 27 publications

(55 citation statements)

References 109 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, we hypothesised that trait anxiety modulates alpha/beta oscillatory activity during the representation of precision weights. In Hein and Ruiz (2022), the Group effect of pwPE on neural oscillatory responses occurred approximately between 1000–1600 ms post-outcome, similarly to the latency of precision-weighting and surprise signals.…”

Section: Resultsmentioning

confidence: 84%

“…In a separate exploratory analysis we evaluated the neural oscillatory responses associated with maintaining predictions about the tendency of the stimulus-outcome contingency, . As argued in our previous work (Hein and Ruiz, 2022), the neural representation of maintenance of predictions is likely to be diffuse in time and could be potentially observed somewhere between the stimuli presentation and the outcome resolution, co-occurring with response commission. Because the convolution approach can isolate and model overlapping events (Litvak et al, 2013), it is particularly suitable for analysing the expression of predictions in EEG/MEG data.…”

Section: Resultsmentioning

confidence: 99%

“…We did not consider spatial relations between anatomical labels but focused on spectrotemporal clusters. Based on the latency of the effects in our previous work (Hein and Ruiz, 2022), we chose as the temporal intervals of interest for the statistical analysis 200-1800 ms for the outcome-locked convolution models, and 100-700 ms for the stimulus-locked GLM. This analysis controlled the family-wise error rate (FWER) at level 0.025.…”

Section: Discussionmentioning

confidence: 99%

“…Although the oscillatory correlates of PC have been primarily investigated in the sensory domain, recent evidence demonstrates that a similar mechanism in the medial prefrontal cortex (mPFC) can explain decision-making processes during exploration-exploitation (Domenech et al, 2020). In reward-based learning tasks, we recently found that beta oscillations were atypically increased in state anxiety during the encoding of revelant pwPE signals (Sporn et al 2020; Hein and Ruiz 2022). In Hein and Ruiz (2022) there was also preliminary evidence for amplified beta activity maintaining (biased) predictions about the tendency of a stimulus-reward mapping in state anxiety.…”

Section: Introductionmentioning

confidence: 99%

“…In reward-based learning tasks, we recently found that beta oscillations were atypically increased in state anxiety during the encoding of revelant pwPE signals (Sporn et al 2020; Hein and Ruiz 2022). In Hein and Ruiz (2022) there was also preliminary evidence for amplified beta activity maintaining (biased) predictions about the tendency of a stimulus-reward mapping in state anxiety. The role of gamma oscillations in mediating altered learning in anxiety through prediction error signalling remains, however, speculative.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Changes in oscillations in anterior cingulate and medial prefrontal cortex are associated with altered signatures of Bayesian predictive coding in trait anxiety

Hein

Gong

Ivanova

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Recent advances in the computational understanding of decision-making processes have led to proposals that anxiety biases how individuals form beliefs and estimate uncertainty. The anxiety and decision-making circuitry broadly overlap in regions such as the medial prefrontal cortex (mPFC), anterior cingulate cortex (ACC), and orbitofrontal cortex (OFC). Changes in activity across these brain areas could help explain how misestimation of uncertainty and altered belief updating can lead to impaired learning in anxiety. To test this prediction, this study built on recent progress in rhythm-based formulations of Bayesian predictive coding to identify sources of oscillatory modulations across the ACC, mPFC, and OFC that are associated with altered learning in subclinical trait anxiety. In a magnetoencephalography (MEG) experiment, two groups of human participants pre-screened with high and low trait anxiety (HTA, LTA: 39) performed a volatile probabilistic reward-based learning task. We modelled behaviour using a hierarchical Bayesian learning model. Furthermore, we quantified the parametric effects of trial-wise estimates of unsigned precision-weighted prediction errors (pwPEs) and, separately, precision weights and surprise on source-reconstructed MEG time-frequency responses using convolution modelling. We showed that HTA interferes with overall reward-based learning performance associated with more stochastic decisions and more pronounced lose-shift tendencies. These behavioural effects were explained by an overestimation of volatility and faster belief updating in HTA when compared to LTA. On a neural level, we observed enhanced gamma responses and decreased alpha/beta activity in HTA during the encoding of unsigned pwPEs about about stimulus outcomes relative to LTA. These effects emerged primarily in the ACC and dorsomedial PFC (dmPFC), and they were accompanied by additional ACC alpha/beta modulations representing differential encoding of precision weights in each anxiety group. Our study supports the association between subclinical trait anxiety and faster updating of beliefs in a volatile environment through gamma and alpha/beta activity changes in the ACC and dmPFC.

show abstract

Section: Resultsmentioning

confidence: 84%