Anxiety and the ability to predict an outcome

Making decisions is a complex process that is made easier when the outcomes of actions are predictable. Researchers know that people with high anxiety are more likely to interpret unexpected variability as a sign of catastrophe. For example, a good student with low or average anxiety may  receive a poor mark on an exam, and interpret this as a sign to study more or do better next time, while a good student with high anxiety might see this poor mark as a sign that they’ll fail the entire course. It is thought that anxious individuals have difficulty updating their expectations when outcomes become more variable, and that this difficulty may stem from an impaired ability to use environmental cues to help them learn to avoid a bad outcome. 

In a recent publication in the journal Nature Neuroscience, Browning et al. tested the idea that anxious individuals have a harder time updating the expected probability of actions. In this experiment, people were categorized by anxiety level and tested on an aversive learning task which required the participants to learn and then update their expected outcome of actions when contingencies changed. Specifically, participants had to choose between two visual cues to avoid receiving an electric shock. During a stable trial block, one cue predicted the receipt of electric shock with 75% probability and the other with 25% probability. On a volatile trial block, the probability of one shape predicting a shock was reversed five times. In all of the blocks, participants had to learn which cue predicted shock through trial and error. In addition to participant behavioral data, pupillary diameter was recorded as a measure of response to unexpected or arousing outcomes.

Individuals with a higher anxiety score showed a decreased ability to update their choices appropriately when the task switched from a stable, predictable block to a more volatile block, compared to participants with lower anxiety levels. In addition, people with lower anxiety scores showed a learning rate that is expected by Bayesian modeling. Interestingly, both high and low anxiety participants had the same pupillary dilation response to choice outcomes during blocks shifts when they  were ‘surprised’ by an unexpected outcome (a shock). However, participants with low anxiety traits showed significant increase in pupillary dilation to volatile trials after the outcome than did participants with higher anxiety.

Taken together, these data suggest that anxiety levels may impact how people learn new action-outcome contingencies, especially if the outcomes are based on probability. The pupillary response in these experiments is especially intriguing, as pupil dilation is known to be under the control of norepinephrine. This suggests that alterations to norepinephrine signaling may relate to how people are able to sense and learn new probabilistic information from their environment.

Publication

1. Browning M, Behrens TE, Jocham G, O'Reilly JX, Bishop SJ (2015) Anxious individuals have difficulty learning the causal statistics of aversive environments. Nature Neuroscience 18:590-596. doi: 10.1038/nn.3961