The brain’s primitive “fear center” may be much more than that, as new research on the amygdala suggests it plays a crucial role in learning, emotion, and decision-making.
Researchers from Dartmouth College revealed their findings in a recent paper published in Nature Communications, questioning the common assumption that the amygdala’s primary function is to drive us to avoid what we fear.
Instead, they argue that the brain’s amygdala helps us to decide between options in complex learning decisions, adjusting as old strategies fail.
The Amygdala is More Than Fear
“People have labeled the amygdala as an emotional fear system, but there is nothing really primitive in the brain, even when you talk about this area,” said senior author Alireza Soltani.
Soltani and his team found that damage to the amygdala was associated with disruptions in learning, suggesting a link between the structure and decision-making in learning.
According to the researchers, the type of learning decisions they are interested in is like using a new, unfamiliar coffee maker. In such a situation, one is faced with two strategies: either attempt to operate the machine as closely as possible to one previously used, or rely on sensory cues to direct one’s attention, such as pressing a button with a blinking light next to it.
“The key distinction is whether learning should be tied to a motor action or the identity of the stimulus,” Soltani said. “Action-based learning involves considering the specific motor movements that can lead to a reward, while stimulus-based learning can be more flexible because it allows you to evaluate and select a desired stimulus without immediately considering the actions needed to get there.”
Learning in the Brain
The team hypothesized that, since these learning modes operate concurrently, some portion of the brain must decide which to follow for the best expected outcome, but they were initially unsure which to follow.
“We approached this question in a largely agnostic way, except that we were aware of the many contradictory findings regarding the role of the amygdala in learning, as well as its diverse contributions to multiple cognitive processes,” Soltani told The Debrief in an email.
For their project, the team relied on monkeys to make a learning decision and receive rewards of varying quality. During blocks of trials, the monkeys’ task was to touch a screen displaying various stimuli.
The two types of blocks involved “what” and “where” tasks, in which the monkeys received a better reward depending on which stimulus they touched on the screen or which side of the screen they touched. Based on feedback, the monkeys eventually learned how to elicit the better response. Then, the researchers introduced further uncertainty by switching the factor that led to a better response.
“They had to learn from feedback both what to learn and how to use that information to guide their choices,” Soltani said. “This required arbitration between two learning systems.”
In an extreme version of the task in which all blocks were “what” blocks, the researchers reduced uncertainty, despite the monkeys still having to determine which block produced the best result.
“Results from this task showed that monkeys with ventral striatum (VS) lesions but an intact amygdala were able to overcome some of their deficits over time,” Soltani explained. “This finding again points to the role of the amygdala in arbitration—specifically, in suppressing irrelevant strategies and allowing more appropriate learning systems to guide behavior.”
Computing the Brain
The binary nature of the experiments, in which the animals either received a reward or did not, still left several questions. To better understand the processes taking place in the monkeys’ brains, the researchers developed computer models to expand on the findings.
“We can gain some insight into what the monkeys learned by examining their choices—for example, by measuring win–stay/lose–switch behavior and other simple model-free metrics,” Soltani said. “However, to truly understand what they learned, what beliefs they formed, and how they used those beliefs to guide their behavior, we had to infer these processes from their choice patterns.”
“This requires building computational models that can generate behavior similar to the monkeys’ behavior based on hidden variables—internal states or beliefs—that are not directly observable but are presumed to be represented in the brain,” Soltani said. “By fitting these models to the monkeys’ choices as precisely as possible, we were able to infer the underlying learning processes and, in turn, identify how the amygdala contributes to the brain computations guiding decision-making.”
The Primitive Brain
The team says the work is exciting because it casts an area of the brain previously considered primitive and capable only of simple tasks, functioning as a driver of complex learning arbitration as the animals tried new strategies after old ones failed.
“The amygdala may help enable flexible, adaptive behavior in changing environments,” Soltani said. “I find that particularly exciting, because it reframes the amygdala not just as a fear-related structure, but as a key contributor to flexible cognition and behavior.”
The team says the biggest unanswered question is how decision-making occurs at the cellular level.
“Specifically, what do neurons in key brain regions (most likely in the prefrontal cortex) do to allow one learning system or strategy to exert a stronger influence on behavior than competing systems?” Soltani said. “How is one representation amplified while another is suppressed?”
“Answering these questions would bring us closer to understanding the neural mechanisms that enable flexible behavior,” Soltani concluded. “And understanding flexibility is central to understanding intelligence itself—which makes this line of research particularly exciting.”
The paper, “Contribution of Amygdala to Dynamic Model Arbitration Under Uncertainty,” appeared in Nature Communications on November 28, 2025.
Ryan Whalen covers science and technology for The Debrief. He holds an MA in History and a Master of Library and Information Science with a certificate in Data Science. He can be contacted at ryan@thedebrief.org, and follow him on Twitter @mdntwvlf.