Bayesian brain theories propose a fundamental shift in how we think about the mind, with goals and desires replaced by predictions and errors. In @NatureComms @CeliaHeyes, @ClarePress & I argue why we should resist this ‘desert landscape’ view of cognition https://www.nature.com/articles/s41467-020-18332-9

Why do we do anything at all? Classic accounts of cognition suggest that to explain behaviour we need to distinguish what an animal believes from what it desires.

However, many have been enticed by the idea that a whole host of natural phenomena - from brain to bacterium - can be thought of as prediction-making error-minimising systems. These accounts stress there is no essential difference between beliefs and desires.

While this way of thinking could unify the mind and life sciences, abandoning beliefs & desires may limit our ability to explain cognition & behaviour in health and disease. For example, precision-weighting can't currently solve how one predictive model acts in both 'modes'.

We think the mind is not as featureless as the desert landscape view suggests. But there may be ways for predictive processing models to explicitly incorporate the distinction between beliefs and desires in the future.