A Princeton-led collaboration with Rutgers has been awarded a $16 million federal grant to enhance the understanding of mental health disorders through the lens of computational psychiatry. Spearheaded by Princeton neuroscience and psychology professor Yael Niv, Ph.D., the five-year Conte center grant funded by the National Institute of Mental Health aims to explore how different disorders like depression, anxiety, and substance use disorder may be affected by differences in latent cause inference.

Latent cause inference, first described by Niv’s lab in 2010, is part of the fundamental detective work the brain does every waking second to make sense of the world. Hearing a loud sound in the middle of the night, for instance, prompts the mind to search for clues to the hidden (latent) causes of the noise. Is it July? How old are my kids? Based on the evidence, the brain might infer it is fireworks, or that a child fell out of their bed.

But if someone’s brain is wired another way, they might come to a different conclusion.

A person with schizophrenia might infer a more ominous cause, like a bomb. Or someone with post-traumatic stress disorder (PTSD) might misinterpret the sound as a sign of imminent danger due to past trauma associated with loud noises.

“We’re really trying to understand this cognitive process of latent cause inference, how it works in the brain, and how it affects different mental health conditions,” Niv said. “Our findings so far indicate this is a process that interacts with almost everything—learning, attention, memory, and perception. It's really basic. And since it’s so basic, if it goes wrong even a little bit, it can lead to changes that may contribute to mental health concerns.”

Given the crucial role latent cause inference plays in mental wellness, even small misfires in how the brain interprets hidden causes could ripple across how someone perceives reality, learns from experiences, and processes emotions.

To better understand how latent cause inference may undergird fundamental brain processes that may go awry in mental illness, Niv brought together a group of scientists at Princeton and Rutgers University to tackle such a big question across multiple, and more manageable, projects.

“We took a group of people who were really interested in collaborating at Princeton and Rutgers and found, through conversations over a year and a half, an area of common interest,” Niv said. “It turned out to be latent cause inference.”

In addition to Niv, the Princeton team includes neuroscience professor Ilana Witten, Ph.D., and dually-appointed neuroscience and psychology professors Jonathan Cohen, M.D., Ph.D., Nathaniel Daw, Ph.D., and Kenneth Norman, Ph.D., and from Rutgers, Gary Aston-Jones, Ph.D., Anna Konova, Ph.D., Avram Holmes, Ph.D., Andrew Westbrook, Ph.D., and David Zald, Ph.D.

The grant will fund four main projects: 

• Project 1: Latent Cause Inference as a Fundamental Cognitive Process
  Led by Niv, Daw, and Westbrook, this project will investigate the relationship between individual differences in latent cause inference and mental health symptoms, as well as the brain circuitry involved.
• Project 2: Latent-Cause Inference in Compulsion
  Led by Zald and Konova, this project will explore a new interpretation of compulsive disorders – such as obsessive-compulsive disorder and substance use disorder – as resulting from over-splitting of latent causes.
• Project 3: Latent Cause Inference in Anxiety
  Led by Niv, Norman, and Holmes, this project will examine how the interaction between latent cause inference and memory processes may contribute to anxiety disorders.
• Project 4: Neural Mechanisms Underlying Latent Cause Inference
  Led by Witten and Aston-Jones, this project will study the neural mechanisms of latent cause inference in rats by measuring brain activity and manipulating the brain hormone orexin within the amygdala, a critical region for emotion and social behavior.

These projects will be supported by three research cores focused on behavioral testing and clinical assessment, computational modeling, and neuroimaging, led by Konova and Niv, Daw and Cohen, and Zald and Norman, respectively.