Psychology has spent more than a decade wrestling with a painful question: how many famous findings are actually solid? In April 2026, the SCORE project — a seven-year effort involving 865 collaborators and 3,900 claims — reported that only about half of tested social and behavioral-science claims replicated, although reproducibility was better when data-sharing practices were stronger. That is the crisis behind the surprising rise of “big-team science,” in which many laboratories test the same idea together instead of trusting a single eye-catching study. (nature.com)
One of the boldest answers is ManyBabies, a global consortium launched in 2015 to improve replication and research practice in developmental psychology. Its recent 2026 reflection argues that the first decade of work has changed the field’s mindset: variation across infants, laboratories, and cultures is not just annoying “noise,” but valuable data. In its first major project, researchers across 67 labs confirmed that infants generally prefer infant-directed speech, yet the strength of that preference varied with factors such as age, method, and language background. (manybabies.org)
Just as importantly, the same collaborative machinery can produce uncomfortable results. ManyBabies 4 tested 1,018 infants from 37 labs across five continents in a preregistered replication of the famous “helper versus hinderer” experiment, which had suggested that babies naturally prefer kind characters. The large study did not find a reliable preference for helpers. Rather than being a failure, that null result is scientifically useful: it suggests the original effect may be weaker, less general, or more developmentally delayed than researchers once believed. (babylab.princeton.edu)
Dogs are now joining the rescue mission. ManyDogs is an international consortium built around the same open-science ideals of transparency, preregistration, and multi-site collaboration. Its first major study examined whether dogs treat human pointing as a meaningful communicative cue. Across 20 research sites and 455 dogs, the answer was more cautious than dog lovers might expect: dogs did follow points, but only weakly, and the differences between cue conditions were limited. With ManyDogs 2 recruiting sites in 2026, the larger lesson is clear. Babies and dogs will not “save” psychology by being cute; they might help save it by forcing researchers to be humbler, larger-scale, and more rigorous. (manydogsproject.github.io)










