Dynamic coding and sequential integration of multiple reward attributes by primate amygdala neurons

The value of visual stimuli guides learning, decision-making and motivation. Although stimulus values often depend on multiple attributes, how neurons extract and integrate distinct value components from separate cues remains unclear. Here we recorded the activity of amygdala neurons while monkeys viewed sequential cues indicating the probability and magnitude of expected rewards. Amygdala neurons frequently signalled reward probability in an abstract, stimulus-independent code that generalized across cue formats. While some probability-coding neurons were insensitive to magnitude, signalling ‘pure’ probability rather than value, many neurons showed biphasic responses that signalled probability and magnitude in a dynamic (temporally-patterned) and flexible (reversible) value code. Specific neurons integrated these reward attributes into risk signals that quantified the uncertainty of expected rewards, distinct from value. Population codes were accurate, mutually transferable between value components and expressed differently across amygdala nuclei. Our findings identify amygdala neurons as a substrate for the sequential integration of multiple reward attributes into value and risk.