A Systematic Analysis of Large Language Models as Soft Reasoners: The Case of Syllogistic Inferences

The reasoning abilities of Large Language Models (LLMs) are becoming a central focus of study in NLP. In this paper, we consider the case of syllogistic reasoning, an area of deductive reasoning studied extensively in logic and cognitive psychology. Previous research has shown that pre-trained LLMs exhibit reasoning biases, such as content effects, avoid answering that no conclusion follows, display human-like difficulties, and struggle with multi-step reasoning. We contribute to this research line by systematically investigating the effects of chainof-thought reasoning, in-context learning (ICL), and supervised fine-tuning (SFT) on syllogistic reasoning, considering syllogisms with conclusions that support or violate world knowledge, as well as ones with multiple premises. Crucially, we go beyond the standard focus on accuracy, with an in-depth analysis of the conclusions generated by the models. Our results suggest that the behavior of pre-trained LLMs can be explained by heuristics studied in cognitive science and that both ICL and SFT improve model performance on valid inferences, although only the latter mitigates most reasoning biases without harming model consistency. 1 Recent research (Lampinen et al., 2023; Eisape et al., 2024) shows that SOTA LLMs prompted with Chain-of-Thought (CoT) display humanlike reasoning biases in syllogistic reasoning; they have difficulties with the examples in Figure 1 , they (i) suffer from a content effect bias, favoring a conclusion compatible with world knowledge (that is, 'believable'), independently of whether it follows from the premises; (ii) struggle with syllogisms that humans also find hard; (iii) are not able to rec-