When people “fail to deny” unflattering claims, it is commonly taken to imply they are true. Yet, the ‘argument from ignorance’ – arguing in favour of something due to a lack of evidence against it – is often deemed a fallacy. Why is abduction from missing evidence permissible in some cases, but not others? We present a framework of factors which disambiguate these cases, using a Bayesian Network model. We suggest that a source’s silence often reflects a latent conflict between their motives for what they want their audience to believe, and complying with external constraints on their speech, like the need to be accurate. In these cases, silence implies that the source does not believe that what they would like to say is true, which licenses a probabilistic inference that it is false. We present data from two studies suggesting people infer from silence like this.

Belief polarization represents a puzzling and important dynamic in belief updating. There is growing awareness that belief polarization can be Bayesian. We provide pre-registered experimental evidence that beliefs can polarize when people receive conflicting testimony from two groups of sources if they have different beliefs about which group's members have greater independence in terms of the factors which affect their testimony. We show this is predicted by a Bayesian Network model of belief updating.