Cholinergic and dopaminergic effects on prediction error and uncertainty responses during sensory associative learning

Navigating the physical world requires learning probabilistic associations between sensory events and their change in time (volatility). Bayesian accounts of this process rest on hierarchical prediction errors (PEs) that are weighted by estimates uncertainty (or its inverse, precision). In a previous fMRI study we found low-level precision-weighted PEs about visual outcomes (that update beliefs associations) activated putative dopaminergic midbrain; contrast, cue-outcome volatility) cholinergic basal forebrain. These findings suggested selective influences at different levels. Here, tested hypothesis, repeating our under pharmacological manipulations healthy participants. Specifically, performed two studies with between-subject double-blind placebo-controlled design: 1 used antagonists (amisulpride) muscarinic (biperiden) receptors, 2 enhancing drugs (levodopa) (galantamine) modulation. Pooled across all conditions 2, respectively, midbrain high-level forebrain as study. However, effects brain activity associated these computational quantities only when splitting into PE precision components: brainstem region putatively containing (pedunculopontine laterodorsal tegmental) nuclei, biperiden (compared to placebo) enhanced responses attenuated activity, while amisulpride reduced responses. Additionally, midbrain, galantamine compared placebo (in body-weight dependent manner) activity. Task behaviour was not affected any drugs. results do support hypothesis clear-cut dichotomy inference levels neurotransmitter systems, but suggest more complex interaction neuromodulatory systems quantities. present may have been confounds inherent fMRI. We discuss outline improved experimental tests for future.