They are GNNs with attention as the message passing function and additional concatenated positional embeddings. As for reasoning, these are not quite 'problems well-trained for', in the sense that they're not in the training data. But they are likely problems that have some abstract algorithmic similarity, which is the point.
I'm not quite sure what you mean that they cannot do causal graph analysis, since that was one of many different tasks provided in the various different types of reasoning studies in the paper I mentioned. In fact it may have been the best performing task. Perhaps try checking the paper again - it's quite a lot of experiments and text, so it's understandable to not ingest all of it quickly.
In addition, if you're interested in seeing further evidence of algorithmic reasoning capabilities occurring in transformers, Hattie Zhou has a good paper on that as well. https://arxiv.org/pdf/2211.09066.pdf
The story is really not shaping up to be 'stochastic parrots' if any real deep analysis is performed. The only way that I see someone could have such a conclusion is if they are not an expert in the field, and simply glance at the mechanics for a few seconds and try to ham handedly describe the system (hence the phrase: "it just predicts next token"). Of course, this is a bit harsh, and I don't mean to suggest that these systems are somehow performing similar brain-like reasoning mechanisms (whatever that may mean) etc, but stating that they cannot reason (when there is literature on the subject) because 'its just statistics' is definitely not accurate.
