Expert Advice during decision making

Sometimes we have to make decision and we have no clue what we are doing. In such a case we would like to rely on the advice of an expert. Sometimes that can lead us to a wrong path if the alleged expert talks out of their ass. But often it is a good strategy. The question I addressed in two studies was: How does expert advice have an impact on the brain signals related to decision making.

In the first study we investigated the differences in brain signals when expert advice either confirms or opposes your own thoughts. In the second study we used tDCS to change how much trust a person puts into expert advice. Here, we also looked at the brain signals and how they are impacted by both advice and tDCS.

Methods experiment 1

In this experiment participant had to decide which of two vases was the most expensive. After they made their choice the opinion of an expert, which was correct in 80% of trials, was shown.

During this task we recorded EEG (read more about EEG in the Neuroscience Methods section). We looked at the brain signals, so-called potentials, that were observed at two time points. First, when it was shown whether the expert cue agrees or disagrees with your own opinion. Second, when the eventual outcome was shown if your choice was correct or wrong.

Previous research has shown that during such decision making tasks three brain potentials are observed: the P200, the feedback-related negativity (FRN), and the P300. We investigated which of those three specifically relates to advice information.

Decision neuroscience

Results experiment 1

We found that both during the presentation of the expert cue and the eventual feedback the P200 was related to expert information. That is, signal amplitude was different for when the expert agrees or disagrees. 

The FRN was related to whether the outcome was better or worse than expected. This is indirectly related to expert advice, since an outcome where you were correct and the expert was wrong, is better than expected. But, no direct effect was obseved.

The P300 signals were larger for rewards compared to punishments, but there was no link to expert information

event-related potentials

Methods experiment 2

Again participant decided about vases, but this time the expert advice was given before your own decision. We also added advice from an amateur and a novice. Participants could use that information to guide their behavior.

We also applied tDCS (read more about tDCS in the Neuroscience Methods section) to the frontal part of the brain. We expected that participant would change the amount of following experts due to the stimulation.

Additionally, we looked at how tDCS affects the EEG signals.

tDCS experiment

Results experiment 2

As expected we found that people follow experts more than amateurs and novices. However, we were not successfull in changing this behavior with tDCS. There was no difference in following expert, amateurs and novices between when real tDCS or placebo was applied.

The absence of a behavioral effect does not mean that tDCS did nothing at all. We found that tDCS significantly affected EEG signals. Specifically, the FRN potential. But apparently that was not enough, or the wrong potential, to cause a change in trusting advice.

event-related potentials

Discussion

People rely heavily on expert advice, however, neuroscientific research on this topic is scarce. Here, we tried to find the brain signals related to expert advice. From the first experiment we found that the P200 is related to expert advice. In the second experiment we attempted to stimulate the frontal part of the brain with tDCS to change reliance on advice. Unfortunately we were not successful. This is what we call a null-result. However, the tDCS did affect the FRN brain signal. Since the first experiment shows that this potential is not directly related to expert advice, it is not strange that we found no behavioral effects. Now we have to think about how to change the location and other parameters of tDCS to hit the correct potential.

As a quick aside. Null-results are very frustrating and hard to publish. However, our null-results here show that they can have value. The absence of an effect in experiment 2 is in line with findings in experiment 1. Also, now that we know what not works, we can figure out a new way that does work.

Sources

 Wischnewski & Schutter (2019). Electrophysiological correlates of prediction formation in anticipation of reward‐ and punishment‐related feedback signals. Pyschopysiol, 56(8), e13379.  https://doi.org/10.1111/psyp.13379 

 Wischnewski, Bekkering & Schutter (2018). Frontal cortex electrophysiology in reward- and punishment-related feedback processing during advice-guided decision making: An interleaved EEG-DC stimulation study. Cogn Affect Behav Neurosci, 18(2), 249-262.  https://doi.org/10.3758/s13415-018-0566-8