Henry Van Peters Wilson demonstrated the effect by dissociating sponge tissue and watching the cells reaggregate into small clumps that grew and reorganised into functional sponges with proper canals and chambers. The result was foundational for cell biology, because it showed that cells carry surface recognition machinery that lets like find like without any central plan.
The adhesion is mediated by large proteoglycan complexes on the cell surface, and it is species-specific: cells from two different sponge species mixed together tend to sort into separate aggregates rather than forming a chimera. This specificity is what made the experiment so influential.
The usual overstatement is that any sponge put through a blender will always reform. Success depends on species, cell condition, temperature and how the dissociation was done, and many attempts fail. What remains an active question is how far this reaggregation reflects normal regeneration in the wild rather than an artefact of laboratory handling.