Noradrenaline, Acetylcholine and Dynamic Learning in Healthy Humans

Background: Previous research has shown that human learners are able to encode the probabilities of positive and negative outcomes in parallel, and adjust their learning behaviour to the information content of positive and negative outcomes (Pulcu and Browning 2017). In reinforcement learning tasks, the volatility (also known as unexpected uncertainty arising when the environment changes suddenly) of outcomes directly influences how informative the outcomes are perceived to be (Behrens, Woolrich et al. 2007). Previous studies consistently shown that human participants use a higher learning rate while learning from outcomes with high volatility (Behrens, Woolrich et al. 2007, Browning, Behrens et al. 2015, Pulcu and Browning 2017). Another source of uncertainty is expected uncertainty, which arises from naturally varying outcomes (e.g. day to day changes in temperature). Different sources of uncertainty in the environment are argued to be associated with the activity of different neurotransmitter systems in the human brain (Yu and Dayan, 2005)(Angela and Dayan 2005). For example, outcome volatility (unexpected uncertainty) is thought to be associated with the activity of the central norepinephrine (NE) system, whereas outcome variation (expected uncertainty) has been linked to the cholinergic (ACY) system.

Studies using pupillometry (a non-invasive way of inferring central norepinephrine activity in the human brain) showed that pupil size is associated with the volatility of outcomes during reinforcement learning tasks (Browning, Behrens et al. 2015, Pulcu and Browning 2017). However, pupil size can also be influenced by the activity of other neurotransmitters including acetylcholine limiting the degree to which these results can be attributed to the NE system. Further, previous human studies have been associational in nature and thus the causal role of the NE and ACY systems in human learning under uncertainty have yet to be tested. The current study will address this knowledge gap by investigating how administering a single dose of the norepinephrine reuptake inhibitor reboxetine or cholinesterase inhibitor rivastigmine influence choice behaviour and pupillary correlates of human reinforcement learning.

Study Structure: All participants will be recruited via the recruitment pipeline described below. The randomisation procedure used will treat participants as if 2 parallel studies were running: a reboxetine vs. placebo and a rivastigmine vs. placebo study. Researchers will assign participants to either the reboxetine or rivastigmine arm (NB researchers will not be blind to this allocation although participants will be). Both studies will randomise participants in a 1:1 ratio, stratified by gender, to either the active drug or placebo (NB both researchers and participants will remain blind to this allocation). This will result in an overall group size for the reboxetine:rivastigmine:placebo groups of 30:30:60 (with participants randomised to placebo being specifically associated with the reboxetine or rivastigmine group). The rationale for this approach is that, while study procedures are identical for all participants, the investigators are addressing 2 separate questions in this study (the effect of NE manipulation and the effect of ACY manipulation on learning) which are more straightforwardly assessed using two, separate, placebo controlled comparisons rather than a larger three group study.