Visual word recognition in the Bayesian Reader framework

Visual word recognition is traditionally viewed as a process of activating a lexical representation stored in long-term memory. Although this activation framework has been valuable in guiding research on visual word recognition and remains the dominant force, an alternative framework has emerged in the last decade. The Bayesian Reader framework, proposed by Norris (2006, 2009; Norris & Kinoshita, 2012a), regards the decision processes involved in a task as integral to explaining visual word recognition, and its central tenet is that human readers approximate optimal Bayesian decision-makers operating on noisy perceptual input. This chapter focuses on two issues fundamental to visual word recognition— the role of word frequency and the representation of letter order— and describes how the Bayesian Reader framework provides a principled account of the recent findings related to these issues that are challenging to the activation framework.