Pupil dilation reflects decision uncertainty and alters choice patterns

Three years after starting the project, two years after completing data collection and almost exactly one year after submitting the manuscript, the first paper of my PhD is now out! In this project, we investigated the relationship between pupil dilation, decision uncertainty and across-trial patterns of decisions.

When making decisions about the world, each choice is associated with a sense of uncertainty: the probability that a choice is correct or wrong, given all the available evidence. This quantity of decision uncertainty can be derived from a simple mathematical model. This model gives rise to a set of specific predictions (see top row of the figure). Based on previous work, we hypothesised that ascending brainstem systems, which release modulatory neurotransmitters such as dopamine and noradrenaline throughout the brain, may broadcast decision uncertainty signal across the brain.

We asked 27 participants to perform a difficult decision-making task while measuring the size of their pupil, a proxy for the activity of brainstem nuclei. After participants had made their choice, but before receiving feedback about the correctness of that choice, their pupil size indicated decision uncertainty (bottom row of the figure). The pupil dilation during decision formation on all trials. But, as we hypothesised, pupil size increased most strongly when people were guessing.


Three predictions for decision uncertainty (top), and the corresponding data patterns in pupil responses (bottom). Most informatively, any signal representing decision uncertainty will decrease with stronger sensory evidence on correct trials, but increase on error trials.

Interestingly, similar behavioural and neural uncertainty signals have been described in different species (rats, monkeys and humans), brain areas (orbitofrontal cortex anterior cingulate cortex and thalamus) and behavioural tasks (from odour discrimination to general knowledge questions). It is often difficult to connect such different studies. By using the same theoretical framework across studies, however, we can begin to integrate findings from different species and experimental setups. The simple model described above is our theoretical coat rack: it keeps experimental findings organised and connected, and it helps us avoid ending up with a set of experimental findings like a loose pile of clothes on the floor.

We further reasoned that the pupil-linked neuromodulatory systems driven by decision uncertainty may affect choices later in time. We found that uncertainty-related pupil responses indeed affected a specific aspect of behaviour: serial choice bias, or the tendency to systematically repeat or alternate decisions. Specifically, after trials with a large pupil dilation, observers reduced their dominant tendency to repeat that previous choice. Ongoing work in the lab investigates the factors that drive these serial biases and their neural bases. Our current behavioural and pupillometry findings set the stage for further research into the neural basis of uncertainty signals and the way they modulate serial choice bias.

Last but not least, all the data we collected for this study are available at FigShare and the task and analysis code (in Matlab and Python) is on GitHub.

Urai AE, Braun A and Donner TH. (2017) Pupil-linked arousal is driven by decision uncertainty and alters serial choice bias. Nature Communications 8:14637. DOI: 10.1038/ncomms14637.

A slightly edited version of this post also appeared on our lab website. See here for news items covering the study.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s