In this talk I am going to present a linguistic account of human quantified reasoning. Given that experimental investigations of deduction are presented verbally and require most of the time a verbal conclusion, it is well-known that linguistic processes play an important role to encode the premises and to produce a response. I will go a step further in arguing that the operations of encoding and decoding are not just playing the role of transduction, but that they are at the very heart of some aspects of the reasoning process itself. Following Geurts (2003), among others, I have identified the semantic property of monotonicity as being the relevant concept for developing a linguistic account of reasoning. Monotonicity is the property of some linguistic expressions that consists in allowing for directional inferences. For instance, from the statement that “Every farmer who owns a donkey bits it”, it may be inferred—because of the monotonicity properties of the quantifier “every” and the information that being a male donkey is being a donkey—that “Every farmer who owns a male donkey bits it”. I will therefore argue that, to the extent that it relies on monotonicity, quantified reasoning is a matter of interchanging set-denoting representations, going from a set to one of its subsets, or vice versa. Plan: In the first part of the talk, I will introduce the audience with the most important questions facing the psychology of quantified reasoning: the competence question—how do subjects successfully solve reasoning problems? and the bias question—why do subjects also make many systematic errors? I will quickly review current theories, and argue that these theories provide incomplete or wrong accounts. In the second part, I will develop my monotonicity account. I will nevertheless have to supplement it with two other linguistic processes, namely symmetry and existential import. I will articulate these inferential processes into a proof-theoretic system. In the third part, I will discuss the cognitive plausibility of this theory as an answer to the competence account. In the fourth part, I will criticize Geurts’ (2003) answer to the performance question. His account is based on a system of weighted complexity values ascribed once and for all to the aforementioned inferences and to some grammatical structures. I will argue instead that the complexity values of these inferences are contextually determined, and will sketch an adaptive system meant at predicting these values. References: Geurts, B. (2003). Reasoning with quantifiers. Cognition 86. 223-251.