Introduction: Learning requires individuals to make arbitrary associations between environmental stimuli and behaviors leading to reward. However, it remains unclear which changes in neuronal activity reflect an animal’s internal representation of a new memory rather than changes in the animal’s external environment. Using a novel learning task in primates and multiple-neuronal recordings, we identify a neuronal signal in the prefrontal cortex (PFC) that is specific to learning.
Methods: Multiple-neuronal recordings were obtained from the dorsolateral PFC of monkeys performing a novel associative learning task. The monkeys were rewarded with drops of juice if they correctly identified one of two images associated with a reward. The rewarded image and reward quantity were changed at random intervals such that a change in reward was either associated or not associated with learning. The firing rate, entropy, variance and spike cross-correlations of the recordings were analyzed before and after a switch in reward. Comparisons were then made between switches in which the animals were required to learn a new association and those in which they experienced a change in reward without learning required.
Results: 50 total neurons were recorded in two primates. During episodes of learning, few neurons (3/50) demonstrated a change in firing rate. In comparison, more neurons (17/50) demonstrated a change in firing variance, suggesting that neuronal responses became unstable when the monkeys needed to change their responses from a well-learned association to a new response. When no learning was required, a similar small number of neurons (3 and 5 out of 13) demonstrated a change in firing rate and variance, respectively. Minimal changes were noted in entropy or spike cross-correlations.
Conclusions: These data suggest that destabilization of PFC neural activity is a principle neural signature of learning new associations. More importantly, changes in neuronal stability appear to reflect the animal’s internal representation of learned associations.
Patient Care: Understanding of the normal neuronal activity underlying learning provides a foundation for further investigating pathologic states in which learning and memory formation is impaired. Focusing on the population neuron activity of the prefrontal cortex also provides a basis for exploring the more complex multi-modal communication between different parts of the cortex and deep brain structures that is likely integral to learning and memory formation.
Learning Objectives: By the conclusion of this session, participants should be able to: (1) understand the explore-exploit model of learning, (2) explain how the data presented demonstrates a network destabilization in the prefrontal cortex that is specific to learning, (3) discuss the "internal" and "external" states in reward-based learning in the context of the study design.