Category transfer in sequential causal learning: the unbroken mechanism hypothesis.

Area of Science:

• Cognitive Psychology
• Causal Inference
• Machine Learning

Background:

• Humans can learn categories from causal input and apply them to new situations.
• Category transfer is not always consistent, with learners sometimes abandoning learned categories.

Purpose of the Study:

• To investigate the conditions under which categories are transferred in causal learning.
• To propose and test the "unbroken mechanism hypothesis" as an alternative to essentialist explanations for category transfer.

Main Methods:

• Two experiments were conducted to test the unbroken mechanism hypothesis.
• Participants engaged in causal learning tasks involving category induction and transfer.

Main Results:

• Findings support the unbroken mechanism hypothesis.
• Category transfer is linked to the assumption of a continuous causal mechanism between related tasks.

Conclusions:

• The unbroken mechanism hypothesis provides a novel explanation for category transfer in causal learning.
• Learners' assumptions about the continuity of causal mechanisms significantly influence category transfer.