Can the Evidence for Explanatory Reasoning Be Explained Away?

Recent evidence appears to show a close connection between explanation and belief revision, specifically, the revision of graded beliefs. Insofar as this is also evidence of violations of Bayesian norms of reasoning, the question arises whether we are facing a new bias here, on a par with previously discovered biases in probabilistic reasoning. We consider an apparently successful attempt by Costello and Watts to explain away a number of known such biases in terms of sampling error, which makes those biases look entirely innocuous and compatible with the descriptive adequacy of Bayesian psychology in any but the most uninteresting way. Specifically, we query whether this attempt can be extended to neutralize the aforementioned evidence allegedly showing that explanatory considerations influence our reasoning in ways inconsistent with Bayesian prescriptions.