Do Large Language Models know who did what to whom?

Large Language Models (LLMs) are commonly criticized for not “understanding” language. However, many critiques target cognitive abilities that, in humans, are distinct from language processing. Here, we instead study a kind of understanding tightly linked to language: inferring “who did what to whom” (thematic roles) in a sentence. Does the common training objective of LLMs—word prediction—result in sentence representations that capture thematic roles? In two experiments, we characterized sentence representations in four LLMs. In contrast to human similarity judgments, in LLMs the overall representational similarity of sentence pairs reflected syntactic similarity but not whether their agent and patient assignments were identical vs. reversed. Furthermore, we found little evidence that thematic role information was available in any subset of hidden units. However, some attention heads robustly captured thematic roles, independently of syntax. Therefore, LLMs can extract thematic roles but, unlike humans, this information influences their representations only weakly.