Many people believe that images—photographs in particular—are truth bearers; that they provide meaningful testimony and have what philosophers sometimes call "factive", as opposed to “fictive”, status. We also commonly hear of how images are untrustworthy because they can be made to falsify the facts. I aim to explain why these ways of talking about images, in terms of their truth-value, have the misleading effect of reducing images to linguistic tokens. Furthermore, doing so overlooks, misunderstands or worse still ignores, the essentially mute but nonetheless powerful effectiveness of images as substitutes for the things they represent.

Almost all theories of representation refer to images as "signs" or "signifiers", as "readable" objects or "messages" that require "decoding", "deciphering" or "interpreting". In everyday language, we talk of how images "convey meaning", "have content" and are "about" the things to which they "refer". We also talk of what images "tell" us, what they "describe", "articulate", "suggest", "explain" and "imply". Characterising images as semantic entities in these ways has the great advantage of rendering them as truth-evaluable. The purpose of this paper is to explain that whilst images are indeed truth-evaluable, they are not fundamentally truth-dependent.