Functions of primate amygdala neurons in economic decisions and social decision simulation

Abstract

Long implicated in aversive processing, the amygdala is now recognized as a key component of the brain systems that process rewards. Beyond reward valuation, recent findings from single-neuron recordings in monkeys indicate that primate amygdala neurons also play an important role in decision-making. The reward value signals encoded by amygdala neurons constitute suitable inputs to economic decision processes by being sensitive to reward contingency, relative reward quantity and temporal reward structure. During reward-based decisions, individual amygdala neurons encode both the value inputs and corresponding choice outputs of economic decision processes. The presence of such value-to-choice transitions in single amygdala neurons, together with other well-defined signatures of decision computation, indicate that a decision mechanism may be implemented locally within the primate amygdala. During social observation, specific amygdala neurons spontaneously encode these decision signatures to predict the choices of social partners, suggesting neural simulation of the partner's decision-making. The activity of these 'simulation neurons' could arise naturally from convergence between value neurons and social, self-other discriminating neurons. These findings identify single-neuron building blocks and computational architectures for decision-making and social behavior in the primate amygdala. An emerging understanding of the decision function of primate amygdala neurons can help identify potential vulnerabilities for amygdala dysfunction in human conditions afflicting social cognition and mental health.

Keywords: Choice; Learning; Prediction; Reward; Social cognition.

Fig. 1

Signaling of fundamental reward parameters in amygdala: contingency, amount and timing. (A, B) Contingency: animal learning theory states that the important variable for acquiring reward prediction is contingency, not stimulus-reward pairing. (A) No response in single amygdala neuron to fractal stimulus when the same reward occurs with the same frequency also in the absence of the stimulus (‘background’). Despite being paired with the reward, this stimulus is not differentially informative about reward. (B) Response in same neuron to same stimulus with less background reward; only in this situation is the reward contingent on the stimulus; the stimulus predicts the reward. (C) Reward value coding in single amygdala neuron. Reward response increases monotonically with liquid amount. (D) Reward timing in population of 86 amygdala neurons. Different temporal profiles of reward expectation related activity reflect different instantaneous reward probabilities (indicated by three different stimuli; top, singular reward at stimulus end; middle, flat reward rate during stimulus, rewarded trials excluded from analysis; bottom, no reward).

Fig. 2

Amygdala responses during economic decisions. (A). Single amygdala neuron coding value input (green) and choice output (blue) of an economic decision process. Monkeys chose to save liquid reward for later or spend (consume) on current trial; they could plan save-spend choices (but not left-right actions) before choice cues. The neuron’s activity transitioned from signaling value to predicting the forthcoming save-spend choice (partial R², sliding-window regression with value and choice regressors; arrowhead: onset of significant choice signal). (B). Amygdala neurons with planning activity for distant rewards. Neurons signaled the length of the planned choice sequence (dashed magenta curve, population activity) or its subjective value (solid magenta curve. Value activity was highest during sequences lasting six trials, which had the highest subjective value (black bars), i.e. were preferred by the animals, because they offered large reward (green curve) for moderate delay. (C). Single amygdala neuron tracking progress during sequential choices. Activity increased with consecutive save choices over 60-90 s until the monkey decided to spend the reward. (D). Decoding progress from amygdala neurons. Cross-validated progress-decoding accuracy of nearest-neighbor classifier. Inset: better decoding from basolateral than centromedial amygdala neurons.

Fig. 3

Amygdala responses during social decisions. (A). Two monkeys faced each other over a touch screen and took turns making choices between visual objects to learn object-reward probabilities (object values). (B) Amygdala simulation neuron. Neuron predicting object choice for the social partner monkey (left) but not for the recorded monkey (right). Neuronal responses were measured while the recorded monkey fixated the second object of two sequentially presented choice objects and before the partner monkey could move to choose the object on the touch screen. (C) Types of amygdala neurons recorded during the observational learning task. Object value neurons signal the value of specific choice objects, irrespective of whether value derives from own learning or social observation. Social neurons discriminate between self and other by showing differential activity on recorded monkey’s and partner’s trials. Different choice neurons signal either the recorded monkey’s own choices or the partner’s predicted choices (simulation neurons). (D) Biophysically plausible model of amygdala circuits for decision-making and social decision simulation, based on the recorded neuron types. Object-specific value neurons (Value layer) and self-other discriminating neurons (Social layer) provide convergent excitatory inputs to two separate decision systems (Choice layer) for computing own choices (Decision module) and for simulating social partner’s choices (Simulation module). Within each choice-layer module, groups of object-specific neurons endowed with recurrent excitatory connections compete with each other by mutual-inhibitory winner-take-all competition (mediated by inhibitory interneurons). Depending on activity in the social layer, value inputs selectively initiate competition in one of the two choice-layer modules.