As far as I understand, like Bayesian frameworks, the rational choice theory has a probabilistic aspect to it, since the focus is on predicting the actions of individuals assuming that their goal in any context is to maximize utility. However, though arguably both Bayesian models and rational choice theory can be extended to apply in different levels of cognition and behaviour, it seems that Bayesian models are more commonly used in studying 'lower -level' -processing, whereas one is more likely to run into RCT in literature focused on social phenomena.
I believe that the model of the rational agent can be useful in predicting or explaining human behaviour (and cognition) in certain contexts. Considering my own research interests (memory and environmental predictability), I imagine that framing the function of the brain from simply reducing uncertainty to maximizing utility useful when moving from basics to, say, memory biases. That being said, based on the reading it seems that the model of rational agent has been overused to explain a variety of phenomena and support claims that do not necessarily have that much evidence to back them. Simply because one can come up with a explanation for a phenomena that somewhat fits into the idea of a rational agent in terms of evolution does not mean that the explanation is correct. Though I admit my issue might lie with the claims of the authors in the specific papers and not the theory itself, I reckon I might feel more comfortable sticking with the Bayesian framework in the future.