The term direction of fit is used in the philosophy of intentionality to distinguish between types of representations. It is commonly applied in two related senses: first, to distinguish the mental states of belief and desire; and second, to distinguish between types of linguistic utterances, such as indicative and imperative sentences. First, philosophers of mind distinguish between mind-to-world (i.e., mind-to-fit-world) and world-to-mind (i.e., world-to-fit-mind) directions of fit. In the former, mental states such as beliefs are subject to updates in order to fit evidence provided by the world (the mind changes to fit the world, thus beliefs have a mind-to-world direction of fit). In the latter, mental states such as desires motivate the agent to change the world in order to fit the desired state in the mind (the world changes to fit the mind, thus desires have a world-to-mind direction of fit). Similarly, philosophers of language, in particular advocates of speech act theory such as John Searle, distinguish between word-to-world and world-to-word directions of fit. In the former, utterances such as indicative sentences attempt to describe the world; for a statement, the state of affairs is considered appropriate if the content expressed by the words fit the way the world really is (a word-to-world direction of fit). In the latter, utterances such as imperative sentences attempt to cause a change in the world; for an order, the state of affairs is considered appropriate when the world changes to fit the words (a world-to-word direction of fit). In both cases, the issue is how representations are considered satisfactory. Beliefs and descriptive statements are considered satisfactory when the states of affairs they represent match the world; i.e. when they are true. Desires and orders are considered satisfactory when world matches the state of affairs they represent; i.e. when they are fulfilled.

In general

In philosophy of mind, a belief has a mind-to-world direction of fit. A belief (that p, say) depicts the world as being in a state of affairs such that p is true. Beliefs, some philosophers have argued, aim at the truth and so aim to fit the world. A belief is satisfied when it fits the world. A desire, on the other hand, normally expresses a yet to be realized state of affairs and so has a world-to-mind direction of fit. A desire that p, unlike a belief, doesn't depict the world as being in the state that p; rather it expresses a desire that the world be such that p is true. Desire is a state that is satisfied when the world fits it. A way to account for the difference is that a (rational) person that holds the belief that p when confronted with evidence that not-p, will revise his belief, whereas a person that desires that p can retain his desire that p in the face of evidence that not-p. To a philosopher of language a word-to-world fit occurs when, say, a sports journalist correctly names Jones as a goal scorer; while if the journalist mistakenly names Smith as the goal scorer, the printed account does not display a word-to-world fit, and must be altered such that it matches the real world. Conversely, a world-to-word fit occurs when a fan of Smith's team opines that they deserved to win the match, even though they lost. In this case, the world would have to change to make the sports fan's wish become true. However, in the case of, say, a judge delivering a death sentence to a criminal declared guilty by a jury, the utterances of the judge alter the world, through the fact of that utterance; and, in this case, the judge is generating a world-to-word-to-world fit (see below). So, if the judge's opinion is upheld, the world must be altered to match the content of the judge's utterance (i.e., the criminal must be executed).

In medieval philosophy

According to Thomas Aquinas (Summa Theologica, Part I, Question 21, Article 2), there are two kinds of "truth" (veritas), both understood as correspondence between mind (intellectus) or words (oratio) and world ("things", res):

In philosophy of language

Perhaps the first to speak of a "direction of fit" was the philosopher J. L. Austin. Austin did not use the distinction between different directions of fit to contrast commands or expressions of intention to assertions, or desires to beliefs. He rather distinguishes different ways of asserting that an item is of a certain type. In a detailed analysis of the distinctions between various scenarios, such as (a) mislabeling a triangle as a square (which Austin regarded as an act of linguistic violence) and (b) inaccurately describing a triangular object as a square (which Austin considered an act of factual violence), Austin introduced a conceptual differentiation. He labeled these distinctions as follows: The concept of direction of fit can also apply to speech acts: e.g., statements, guesses and conjectures have word-to-world direction of fit, while commands and promises have a world-to-word direction of fit. John Searle and Daniel Vanderveken assert that there are only four possible "directions of fit" in language: Searle used this notion of "direction of fit" to create a taxonomy of illocutionary acts. Although Elizabeth Anscombe never employed the term "the direction of fit", Searle has strongly argued that the following passage from her work Intention was, by far, "the best illustration" of the distinction between the tasks of "[getting] the words (more strictly their propositional content) to match the world... [and that of getting] the world to match the words":

In philosophy of mind

According to Velleman, when used in the domain of the philosophy of mind, the concept direction of fit represents the distinguishing feature between two types of intentional mental states: In some forms of mind-body dualism, a matching factum and faciendum must be present in a person's mind in order for him to act intentionally. If a person has the belief that action (A) will lead to state (S), and has the desire that state (S) obtain, then he will perform action (A). The action is directly caused by simultaneous presence of the two mental states; no further explanation is needed. According to Velleman: Ruth Millikan has also written influentially about representations, noting that many primitive representations used by animals are characterized by a dual direction of fit; she terms such representations "pushmi-pullyu representations." As an example, she cites the role of bee dances in both informing other bees about the location of resources (indicative, or dance-to-world) and directing their action (imperative, or world-to-dance): Millikan suggests that many perceptual representations (including those in humans) have such a dual function, both providing an agent information about the state of affairs in the world and suggesting action possibilities to change that state of affairs. She further remarked on the similarity between such representations and the role of affordances in the ecological perception theory of psychologist J. J. Gibson. However, Millikan's openness to Gibson's framework has not been typical of philosophers of mind, and later philosophers such as Bence Nanay have explored the possibility of action-oriented perception without endorsing affordance theory. The predictive coding framework of neural representations developed by neuroscientist Karl Friston and philosopher Andy Clark has similarly been observed to be an attempt to construe representations with dual direction of fit, uniting predictions (mind-to-world) and actions (world-to-mind).