We are discussing the difference between living with language and without language. — Athena

Yes. That's part of it. There's also the transition between. There are also different kinds of languages consisting of different kinds of meaningful behaviours, marks, utterances, etc.

Indeed, what counts as language matters in more than one way.

For example, when a dog checks out a bowl, because it expects there to be food in it, and is disappointed, I don't suppose it says to itself "Oh, my belief that there was food there is wrong" or anything similar. It simply walks away. But that action counts as a recognition that its belief was false. — Ludwig V

Recognizing that the bowl is empty is not the same as recognizing that one's own belief about food being in the bowl is false. The former is about the food and the bowl. The latter is about one's own thought/belief. The dog can directly perceive the food, the bowl, and its own hunger. Thought and belief are not directly perceptible things. Nor are truth/falsity. Nor is meaning. Nor are social/institutional facts. Nor are any number of abstractions.

I cannot find good ground for claiming that any creature incapable of naming and descriptive practices is capable of abstraction. Recognizing that one's own belief is false requires comparison/contrast between the belief and what the belief is about. That seems to require a skillset unobtainable to dogs.

But it's not enough for him to generalize and understand that (1) whenever he kills a chicken, he will be in trouble. He also needs to understand that (2) if he does not kill chickens, Janus wll not be displeased with him.

There's more to Jimi than just recognizing causal correlations. — Ludwig V

Of course there is more to any thinking creature than just the recognition/attribution of causality, but it seems to me that that process, regardless of the creature, is more than adequate for being a case of thinking(thought/belief).

I'm not convinced that Jimi knows he's in trouble, so I question the account above on its presuppositional ground.

It is more than enough that Jimi inferred that his own behaviour caused Janus'. Here, all Jimi needs to avoid killing chickens is to believe that if he does Janus will do whatever Janus did the first time. He does not need to understand that if he does not kill chickens Janus will not be displeased. He just needs to believe that if he does, Janus will do what he did the first time. His belief that his own behaviour caused Janus' comes replete with the further inference/belief/expectation that if he does not, Janus will not do that either. That's how the recognition/attribution of causality works.

I'm not sure that dogs have a concept of causality as such. — Ludwig V

I agree but...

Where does the need for having a concept of causality come from? Again, I do not find the notion of concept to be of help. Generally speaking, it seems to be a step backward instead of forward. One can recognize/attribute causal relationships, which is what is meant by "recognize/attribute causality" without having a concept of causality(thinking about causality as a subject matter in and of itself). A creature can believe that X causes Y without having a concept of causality. Recognizing/attributing causality requires only inferring that.