This paper provides an epistemological and methodological analysis of the practice of using neural language models to simulate brain language processing. Firstly, neural language models are introduced; a study case showing how neural language models are being applied in cognitive neuroscience for simulative purposes is then presented; after recalling the main epistemological features of the simulative method in artificial intelligence, it is finally examined how the simulative method is modified when using neural language models. In particular, it is argued that the epistemic opacity of neural language models requires that the brain itself be used to simulate the model and to test hypotheses about the model, in what is called here a co-simulation.