Neuromodulators are neurotransmitters that regulate the excitability and plasticity of other neural elements. We have long studied how they represent variables of computational importance such as prediction errors for rewards and punishments, average reward signals that determine the opportunity cost of slothful action, or forms of uncertainty.
We work on all the major ascending neuromodulators: dopamine (DA), serotonin (5-HT), norepinephrine (NE), and acetylcholine (ACh). There is a highly impressive set of new methods for measuring and manipulating neuromodulators, which means that controversies from the past are about to have sharp new light cast upon them. Along with collaborations with groups working on rodents, I collaborate with Read Montague in Virginia Tech who is measuring neuromodulators in humans using cyclic voltammetry, originally during implantation of deep brain stimulation electrodes for Parkinson’s disease and essential tremors, but soon also for epilepsy monitoring. Under suitable ethical constraints, the latter group will allow many advances in our understanding, given that patients are typically willing to engage in cognitive tasks over a relatively extended period.
We have completed projects on dopamine (Lloyd & Dayan, 2023, in which we consider how cognitive control might exploit this neuromodulator’s regulation of active approach to achieve appropriate ends; and Dayan, 2022, looking at interactions between liking and wanting in the context of flavour conditioning) and acetylcholine (Chebolu et al., 2022, looking at the role of ACh in arousal and vigilance through the medium of modelling a rodent sustained attention task). However, we are currently working on a rich paradigm for inducing different sorts of spatial uncertainty in mice in virtual reality, and are also analysing fibre photometry data on multiple neuromodulators in the IBL task. We also provided a detailed review of the involvement of the interaction between sympathetic (NE) and parasympathetic (ACh) somatic systems, and their neural partners, in post-encounter threat, a particularly important stage in the defensive control hierarchy (Roelofs & Dayan, 2022).