Periodic Reporting for period 4 - LearnAnx_CircAmyg (Learning and Anxiety in Amygdala-based Neural Circuits)
Reporting period: 2022-03-01 to 2022-07-31
We have a limited understanding of the mechanisms that underlie complex learning in emotional situations, and how these processes can go awry and result in maladaptive behaviors and in extreme cases, underlie psychopathologies. In this project, we developed models of complex behaviors motivated by either positive or negative reinforcers and examined how they are governed and coded in neural networks. We focus on the pathway between the Amygdala and the prefrontal-cortex, as it is the main network that underlies learning and memory formation in emotional contexts. It is also a network that is implicated in almost all psychiatric conditions, mainly in anxiety and post-trauma disorders. We record neural activity using deep multi-contact arrays to unveil the coding schemes embedded in these networks, and develop models of learning that are highly evolved in primates, such as rule learning, generalization (the ability to abstract) and social interactions. The development of more complex models of behavior is crucial and is the next logical and required step in achieving the translation of models for mood disorders, as well as in understanding the basic mechanisms behind the rich repertoire of primates’ emotional behaviors. We found that neural activity in the Amygdala responds differently to punishments and rewards in a way that can explain the differential behavior, and showed that wide networks in the primate brain use precise coding mechanisms to store emotional memories, as well as underlie changes in learning during uncertainty. We also extended these findings to social situations; and showed that the same brain circuit codes for the valence of social interaction and the processing of primary sensory stimuli, indicating a shared origin that allows rapid responses in the face of danger. This was strengthened by a cross-species comparison demonstrating a neural tradeoff between primate species and brain networks. Finally, we developed a rule-learning scheme and described the dynamics in these valence-based neural networks. Our future studies continue to elaborate on this path and investigate the neural codes and interactions between brain regions that underlie complex learning and its abstraction in aversive and positive environments. This work might lead to a better understanding of maladaptive behaviors and psychopathologies.
We developed several behavioral models and were able to record neural activity during the performance of these behaviors. We unveiled coding mechanisms in the primate amygdala that use temporal sequences of neural activations to enable the rehearsal of a recently acquired emotional memory, either positive or negative. This result demonstrates that a recently learned emotional event is ‘rehearsed’ in the brain to allow its consolidation into memory. We looked for differences between the human amygdala and the non-human-primate amygdala, and demonstrated that the ability of neurons to code information is richer in humans, yet more robust in non-humans. This finding might suggest why humans have a better ability to learn and generalize but are also more vulnerable to “bugs” in the code; ones that might even lead to psychopathologies. In addition, we developed a new framework of complex rule-learning, and show how networks in the prefrontal cortex and subcortical regions contribute in complementing manners to learning new rules and allow the generalization of reinforcement-based learning. We also discovered that the same networks represent social situations similar to how they represent primary reinforcers, suggesting that social responses (and disorders) might originate from the same mechanisms as primary responses to threat or reward. Finally, we showed 10 that uncertainty underlies changes in learning and contributes to maladaptive behaviors that might be linked to anxiety, and the related neural circuits that underlie this. The findings were described in ten peer-reviewed papers and opened a new line of research in the lab.
The development of new models for complex maladaptive learning is a much-required progress beyond the current models for anxiety/trauma, mostly based on classical conditioning. Because human psychopathology is more complex than simple associative learning, understanding how the brain learns and codes for these elaborated behaviors is a must. We made clear progress in these aspects and unveiled several behavioral mechanisms that differentiate between aversive to positive environments and might contribute to psychopathologies. We also performed the first cross-species comparison of neural codes, and developed a real-life social paradigm. In the last stage of the project we developed a closed-loop approach that allows us to test coding in the primate amygdala and related circuits during emotional learning, and even manipulate and modulate it to achieve better control of the process and hopefully suggest translational impact.