Dynamic causal modelling reveals signal propagation of decision confidence in a perceptual decision-making task

Abdoreza Asadpour, KongFatt Wong-Lin

Research output: Contribution to conferencePosterpeer-review

Abstract

Perceptual decisions often entail a subjective level of confidence. Previous studies have investigated the neural representations of confidence based on behavioural measurements as well as their spatial representation in the human brain. However, large-scale neural circuit dynamical mechanisms underlying decision confidence and the effective connectivity of active brain regions for decision confidence remain unclear. In this study, we implemented dynamic causal modelling (DCM) on a previously acquired dataset (Gherman & Philiastides, 2020) to determine the dynamics of neural population activities and signal propagation during the early stage of a visual motion discrimination decision-making task. The task consisted of the discrimination of the coherent direction of random dot kinematograms and rating the confidence of each choice on a trial-by-trial basis in two blocks of 160 trials acquired by a simultaneous electroencephalogram and functional magnetic resonance imaging (EEG-fMRI) method. The stimulus-locked epochs with the correct responses were extracted from the EEG data, and the epochs of low confidence (less than 50%) and high confidence (greater than 60%) of 19 participants were selected as conditions. Based on Gherman & Philiastides (2018) and previous studies, we included eight brain regions whose activity is positively correlated with confidence. However, unlike Gherman & Philiastides (2018), we neurally modelled the EEG-based dynamics, informed by fMRI, using the DCM approach. Using SPM12, Bayesian model selection was used to determine the best model fit among four different event-related potential forward-backward candidate models (David et al., 2006) between the two conditions. Our findings showed the activity of left and right posterior parietal cortices (pPC) in addition to the extracted active brain regions from fMRI in Gherman & Philiastides (2018), while complementing the latter with higher temporal resolution. Importantly, our results revealed not only the temporal flow of information from the occipital lobe (OL) to the lateral orbitofrontal cortex (lOFC) through the pPC, but also the underlying temporal responses of excitatory and inhibitory neural populations. Using the parametric empirical Bayes approach, the relatively stronger connections with a 100% probability of occurrence among the participants were forward and backward connections from the OL to right pPC, lateral connection from right lOFC to left lOFC, forward connection from OL to left pPC, and forward connection from right striatum to anterior cingulate cortex, respectively.
Original languageEnglish
PagesProgram/Poster 402.15
Number of pages1
Publication statusPublished online - 14 Nov 2022
EventSociety for Neuroscience 2022 meeting: SfN 2022 - San Diego Convention Center, San Diego, United States
Duration: 11 Nov 202215 Nov 2022
https://www.sfn.org/meetings/neuroscience-2022

Conference

ConferenceSociety for Neuroscience 2022 meeting
Country/TerritoryUnited States
CitySan Diego
Period11/11/2215/11/22
Internet address

Keywords

  • perceptual decision making
  • dynamic causal modelling DCM
  • signal propagation
  • decision confidence
  • decision uncertainty
  • source localisation
  • effective connectivity analysis
  • EEG-fMRI
  • Bayesian model selection
  • large-scale neural network modelling

Fingerprint

Dive into the research topics of 'Dynamic causal modelling reveals signal propagation of decision confidence in a perceptual decision-making task'. Together they form a unique fingerprint.

Cite this