Which early childhood experiences most affect mental health?

Anna Vannucci has a twin. Her younger sisters are also twins. Despite the shared genes and the close proximity in age (all four siblings were born within 15 months of each other), Vannucci and her siblings all had deeply different personalities. At the age of 8, Vannucci sat her siblings down on a couch and asked them questions to try to get to the bottom of why they all seemed so different.

That early interest in clinical psychology led Vannucci to a college major in the field, and, eventually, to Columbia’s psychology department, where she is now a PhD candidate in psychology professor Nim Tottenham’s lab. With colleagues in that lab, Vannucci recently authored a paper that uses AI to better predict which experiences in early childhood predict later mental health challenges.

Columbia News spoke with Vannucci about the findings and her professional path.

What was the goal of this paper?

Adversity in childhood is a huge problem. It’s often linked to increased risk for mental health problems. What we don’t know is why some kids go on to develop mental health problems, while others do very well following early life adversity and demonstrate a lot of resilience. This paper applied machine learning to disentangle some of this.

What did you find?

One of the main findings we discovered is that emotional abuse, like insulting or humiliating a child, is the most telling indicator of risk for mental health problems. This was surprising because in the clinical world there’s a huge emphasis on the physical aspects of maltreatment, such as physical or sexual abuse. But this shows that it’s the emotional aspects of these adversities that have the most salient effects.

Something else that we found is that significant separation from caregivers – like shifting from a biological caregiver to a foster family or an adoptive family or separations due to parents being deported can destabilize a child’s sense of security and emotional foundation.

We also found that consistent parenting strategies like setting routines and establishing expectations and being responsive to your child can lower risk and foster resilience. The brain remains resilient through young adulthood. Adversity is not destiny. If you have early life adversity and supportive, consistent caregiving later on, it can really help you.

What brought you to this question?

We know in general that adversity increases risk for virtually every health issue. But we don’t currently know the risk that a child has for needing intervention or certain things like that. So what this allowed us to do is try to pick out experiences in a data driven way to pinpoint which kids are going to be in need of intervention. For this study, we gathered detailed information about early experiences in a large group of children with a wide variety of early caregiving experiences (including experiences with foster care, orphanages abroad, and adopted children, for example). Previous studies typically would only include children with one type of experience. There is no other data set out there like this.

What’s the machine learning aspect of this paper?

We use an existing data set to create what’s called a training sample, which will learn all the associations between certain early life adversities and mental health outcomes, and make predictions of what the outcomes will be. Once we have that, we can present the AI model with data from new kids, and it will give them a score that essentially, quite accurately, predicts their future mental health based on their early childhood experiences.

How do you assess whether someone was emotionally abused?

We gather lots of details from parents. For ethical reasons we only sample children who are currently in a stable situation, so our interviews are usually with adoptive parents who are telling us about abuse in a previous household. There are some other situations, too, like biological parents who have been separated from their kids for a while, for example, while a parent seeks treatment for a substance use issue, and then the kids return home. We also do a parent report questionnaire. So we try to ask questions in many types of ways to get a full picture of a child’s experiences.

What drew you to this work?

I was initially interested in clinical psychology, and wanted to work in treatment and prevention in the world of eating disorders. Often you find that the root cause in those instances is something traumatic. I found that I was often bringing the stories home with me in a difficult way, so I decided to pivot to this work, thinking about the mechanisms behind the emergence of some psychiatric issues in a broader way.