A group of scholars at Harvard University is spearheading a campaign to make sure the early-childhood programs policymakers put in place to disrupt intergenerational poverty are backed by the latest science.

The idea sounds entirely reasonable, but it’s all too rare in practice, says Jack P. Shonkoff, the director of the university’s Center on the Developing Child and the chair of the National Scientific Council on the Developing Child. That’s because program grants and policies are generally structured in ways that incentivize “positive” results. Agreements along the lines of, “We’ll give you funding to test this specific policy intervention and if you can prove it worked in three years, we’ll give you more,” are standard. Shonkoff and his colleagues think that model needs a major update.

On Wednesday, the center will publish a report that calls for an online and in-person network that uses recent advances in scientists’ understanding of the way young brains grow to create and test early-childhood interventions. “The absence of a science-based R&D platform in the early childhood field threatens the future of all children, families, and communities whose challenges are not being addressed adequately by existing policies and programs,” write the authors.

Brain science is evolving quickly, Shonkoff said. Scientists now understand in great detail, for instance, that exposure to stress can speed up the opening and closing of certain periods of brain development, suggesting that countering the lifelong consequences of toxic stress is something best tackled in the first several years of life. Scientists also know now how children’s brain architecture changes when they have relationships with strong, caring adults. The science indicates that helping parents improve the way they interact with their babies has real impact on physical and mental health as those children grow into adulthood.

But many existing programs focus on either providing services for babies or, separately, for parents. Fewer focus on helping vulnerable adults turn their lives around with an eye toward improving their parenting skills, in part, Shonkoff thinks, because people with ideas for those kinds of interventions haven’t had the space to test them and, crucially, to fail and try again. Right now, “there’s no place for people to be talking about things that aren’t working,” he said. “Where is the culture of innovation here?”

Consider Early Head Start, a program aimed at helping disadvantaged children ages 0 to 3 across the United States. While Early Head Start has the infrastructure in place to test and scale a variety of programs, the report argues that it has so far provided little clarity on which of its programs work best for which people, and how they might be replicated.

Shonkoff says the current focus on determining so-called “best practices” by looking at average impact when it comes to interventions for children living in poverty is tantamount to searching for a single treatment for “cancer,” even though cancer comes in many forms and is caused by many factors. Take acute lymphoblastic leukemia (ALL). In 1965, the year the first Head Start center was opened, the survival rate for ALL was below 5 percent. Now, it’s above 90 percent because doctors choose different treatment plans for different patients. They know how to do that because researchers have developed and tested those treatment plans. That’s not happening, Shonkoff said, for early childhood.

In short, the idea is to invest in different people and programs who understand the science behind child development and give them the ability to test different interventions. Shonkoff isn’t delusional; he knows that trying to convince cash-strapped government officials to fund speculation isn’t going to work. “It really requires a new breed of philanthropy,” he said—investors who made their fortunes in the high-risk, high-yield world of venture capital and are comfortable in that space, for example.

The Center on the Developing Child has already begun working with several programs to test the early-childhood R&D concept in what it’s calling an “innovation cluster” in Washington state. The Children’s Home Society piloted a video coaching program there aimed at improving executive function and self-regulation skills (things like self-control and the ability to retain and use new information) in both children and their parents. Parents were filmed interacting with their babies. Instead of focusing on what they did wrong, coaches clipped a few seconds of good interaction and used that to positively reinforce the ways caregivers could help their babies develop. The researchers originally tested the idea primarily with mothers but found, after several adaptations, that it was fathers who had previously been disengaged who benefited the most, which allowed them to tailor the program to dads. A separate intervention to bring mental health services to moms flopped, but when adapted, proved effective when it was targeted at teen moms specifically.

There’s “growing interest” in this model of implementing and testing new ideas, said Jason Gortney, the director of the Office of Policy and Innovation at the Children’s Home Society of Washington. But he acknowledges that it takes a willingness to fail and rebound publicly. It’s no longer good enough, as Shonkoff said, to implement wide-ranging practices that only work for some people when science can identify specifically who they work for and when. That doesn’t mean that good ideas shouldn’t be scaled; it means they should be scaled strategically.

To help with that, the center wants to create a robust data hub that uses common measurements so that researchers and policymakers can see which programs are working and where. He acknowledges the challenge of maintaining and policing such a database, but warns that the consequence of pushing forward with programs and policies that aren’t backed by the latest science stands to hurt the most vulnerable children. Putting effective programs in place, he said, can help reduce the widening gaps between children who are born with a lot and those who are born with little. It’s not that all current programs are bad, it’s that many “have failed to keep pace with the benefits of growing up in a high-income family in a rapidly changing world,” the report says. Ultimately, Shonkoff and his colleagues call for “a research and development (R&D) approach that transcends partisan disagreement because it is built on a rigorously peer-reviewed, science- based understanding of how the foundations of learning, behavior, and health are built or weakened over time.”

The idea might sound overly optimistic, but, as Shonkoff said bluntly, “You want somebody in the back room to be figuring this out … It’s a message of real, genuine optimism.”