Model building is typically based on the identification of a set of established facts in any given field of research, insofar as the model is then evaluated on how well it accounts for these facts. Psychology – and specifically visual word identification and reading – is no exception in this sense (e.g., Amenta & Crepaldi, 2012; Coltheart et al., 2001; Grainger & Jacobs, 1996). What counts as an established fact, however, was never discussed in great detail. It was typically considered, for example, that experimental effects need to replicate across, e.g., individuals, experimental settings, and languages if they are to be believed. The emphasis was on consistency, perhaps under a tacit assumption that the universal principles lying behind our cognitive structures determine our behaviour for the most part (or at least for that part that is relevant for model building). There are signs that a different approach is growing up in reading research. On a theoretical ground, Dennis Norris’ Bayesian reader (2006, 2009) has advanced the idea that models can dispense of static forms of representation (i.e., fixed architectures), and process information in a way that is dynamically constrained by context-specific requirements. Ram Frost (2012) has focused on language-specific constraints in the development of general theories of reading. On an empirical ground, the most notable recent advance in visual word identification concern the demonstration that some previously established (in the classic sense) effects depend heavily on language (Velan and Frost, 2011), task (e.g., Duñabeitia et al., 2011; Marelli et al., 2013; Kinoshita and Norris, 2009), or even individual differences (Andrews & Lo, 2012, 2013). Variability has become an intrinsic and informative aspect of cognitive processing, rather than a sign of experimental weakness. This Research Topic aims at moving forward in this new direction by providing an outlet for experimental and theoretical papers that: (i) explore more in depth the theoretical basis for considering variability as an intrinsic property of the human cognitive system; (ii) highlight new context-dependent experimental effects, in a way that is informative on the dynamics of the underlying cognitive processing; (iii) shed new light on known context-dependent experimental effects, again in a way that enhances their theoretical informativeness.