They may, for example, take a likely hypothesis, and then “mutate’ that hypothesis to generate a number of new, slightly different hypotheses with different probabilities.
While the researcher in this article are emphasizing the relevance to development psychology, this "mutating of hypotheses" reminds me of Hegelian philosophy. Are children engaging in some form of natural dialectics?
If children’s playful explorations are so unconstrained how could they actually lead to rational causal learning?Recent research by Schulz (Bonawitz et al. 2011; Cook et al. 2011; Schulz et al., 2007, 2008; see also Legare, 2012) has begun to address this issue. Schulz and her colleagues have shown that children’s exploratory play involves a kind of intuitive experimentation.
Fascinating question! The idea that learning occurs naturally as children play goes all the way back to Piaget, but investigating how different forms of play, such as guided vs unguided, impact the learning experience seems valuable in constructing (or deconstructing) the early childhood experience for the learner. I reviewed the Legare article as it was the most recent in this area of research, and it indicated that even when student engage in "inconsistent" exploration, they are still able to identify causality in their experiences. In layman's terms, student's do not need to be posed a hypothesis first to be able to later think about a hypothesis in relation to a previous experience.
How do they decide which hypotheses to test, and which evidence to use to test them?
The article cited below indicates children prefer "simple"hypotheses. It defines simple as hypotheses with a "good explanation." Even among 4, 5, and 6-year olds, different age groups favored different kinds of "good explanations." In response to the question highlighted in the article, there is evidence to support that deciding on hypotheses and evidence is highly dependent on the child's development stage.
Walker, C.M., Bonawitz, E. & Lombrozo, T. Effects of explaining on children's preference for simpler hypotheses. Psychon Bull Rev 24, 1538–1547 (2017). https://doi.org/10.3758/s13423-016-1144-0