Two Israeli researchers win $4 million in ERC grants to challenge assumptions about families and networks
Bar-Ilan University’s Prof. Reuma Gadassi-Polack is investigating whether children’s emotional difficulties can affect their parents’ mental health, reversing the direction of much traditional family research. Dr. Arnold Filtser is developing mathematical tools for algorithms operating in real-world systems where distance and cost do not follow conventional rules.
Two Bar-Ilan University researchers have won prestigious European Research Council Starting Grants totaling more than $4 million, funding projects that question some of the assumptions underpinning research into family mental health and computer science. Prof. Reuma Gadassi-Polack will investigate whether children’s emotional difficulties can affect their parents’ mental health, while Dr. Arnold Filtser is developing new mathematical tools for algorithms dealing with real-world distances and costs.
Gadassi-Polack of Bar-Ilan’s Faculty of Education and Gonda Brain Research Center received €1.97 million ($2.3 million), while Filtser of the university’s Department of Computer Science and Artificial Intelligence received €1.5 million ($1.74 million).
Gadassi-Polack, a practicing clinical psychologist, is questioning one of the most common ways researchers have studied childhood depression and anxiety: by looking primarily at how parents affect their children.
Her ERC-funded “Child2Parent” project asks the opposite question, whether children’s emotional difficulties affect their parents’ mental health.
Much of the existing research has focused on parental influence through genetics, parents’ own symptoms and parenting behavior. Gadassi-Polack’s project will instead examine the everyday processes through which parents help their children regulate emotions, including the intensity and duration of emotional experiences and their expression.
The project will seek to develop what the university describes as the first mechanistic model of child-to-parent symptom transmission.
Gadassi-Polack will study mother-adolescent pairs using a combination of hormonal measurements, brain imaging and information gathered from participants’ daily lives.
The underlying premise is that the relationship between a child’s and parent’s mental health may be more complicated than a one-way flow of influence from parent to child. By examining whether children’s emotional difficulties can in turn affect their parents, the research could offer a different way of understanding how mental-health problems develop and spread within families.
The project could also have implications for how interventions are designed, shifting some of the focus from treating an individual child to supporting the family as a whole.
Filtser’s research begins with a very different problem, but also challenges an assumption that has become embedded in a field.
His project, “(Non) Metric Embeddings,” focuses on mathematical tools used in algorithms for routing, data search, network design and grouping similar objects. Many of these algorithms rely on a technique known as metric embeddings, which has primarily been developed for metric spaces.
The problem is that many real-world measures of distance, cost or dissimilarity do not behave like a mathematical metric.
In a conventional metric space, the distance from A to B is the same as the distance from B to A. And traveling from A to B through an intermediate point cannot be shorter than traveling directly between the two points.
Real-world systems can violate both assumptions.
A flight from A to B, for example, can cost a different amount from a flight from B to A. An itinerary involving a stopover can also be cheaper than a direct flight.
Filtser’s research aims to extend the mathematical toolkit of metric embeddings to these non-metric settings.
Metric embeddings allow complicated distance data to be represented in a more structured form while approximately preserving distances. That structure can then be exploited by algorithms.
The goal of Filtser’s project is to develop analogous tools for situations where the underlying distances do not obey the standard rules. If successful, the work could enable more efficient algorithms for transportation and communication networks, improve network design and provide new approaches to problems that existing metric techniques cannot adequately address.
