UC Santa Barbara

Studies on multiple patient groups suggest that reasoning has a hemispheric asymmetry component. Previously, we proposed that neural networks in the left hemisphere are driven toward increasing and maintaining certainty, while right frontal networks prioritize congruence between beliefs and evidence. We tested the predictions of this framework with two high definition transcranial direct current stimulation (HD-tDCS) experiments and one functional magnetic resonance imaging (fMRI) experiment. In both HD-tDCS studies, we aimed to induce (or amplify) hemispheric asymmetry in healthy participants as they completed novel reasoning tasks. Each participant completed three tDCS sessions: a LH-bias session, in which the anode was placed over the left inferior frontal gyrus (IFG; BA45) and the cathode over the right IFG; a RH-bias session, in which the anode was placed over the right IFG and the cathode over the left IFG; and a sham session, which served as a control.

In the first HD-tDCS experiment, participants (N=26) completed a probabilistic inference task that required the integration of evidence and one’s prior background knowledge. Consistent with predictions, we found that the intensity of RH-bias stimulation was associated with 1) collecting more evidence, 2) adopting a higher threshold for stopping evidence collection, and 3) making less certain guesses than an ideal Bayesian updater during the evidence presentation. Contrary to predictions, we found that greater LH-bias intensity was associated with more evidence collection, and LH-bias stimulation was associated with greater belief backtracks after encountering conflicting evidence than RH-bias or sham stimulation.

The second HD-tDCS experiment followed a similar stimulation protocol but used reasoning problems that were more deeply embedded in real-world contexts in order to create more salient belief-evidence conflicts. During each stimulation session, 24 participants 1) judged whether a criminal suspect was guilty or not guilty based on crime scene evidence, 2) judged whether or not to pass a law based on arguments in favor and in opposition to it, and 3) judged whether a news headline was real or fake. We found that RH-bias stimulation reduced belief polarization after conflict, which was consistent with our predictions. Similarly, when evidence conflicted participants’ strong beliefs, they backtracked on their beliefs more under RH-bias stimulation compared to sham stimulation and, albeit to a lesser extent, compared to LH-bias stimulation. Under RH-bias stimulation, participants were less likely to judge real news headlines as being real, which resulted in poorer discrimination of real vs. fake headlines compared to sham and LH-bias stimulation.

Finally, in the fMRI experiment, we examined lateralization in frontal anatomical regions for contrasts that we predicted to be more left-lateralized or more right-lateralized. Participants (N=36) completed a modified version of the state guessing task that was used in the first tDCS experiment. Consistent with predictions, contrasts involving uncertainty and belief advances were generally more left-lateralized and contrasts involving conflicting evidence and belief backtracks were more right-lateralized.

We show that HD-tDCS can alter belief updating in healthy individuals in a way that is consistent with the patient literature, but additional experiments are necessary to disentangle the causal relationships between different reasoning biases and neural activity in left and right frontal neural networks.