An Orthographic Prediction Error as the basis for efficient Visual Word Recognition

Abstract

Most current models assume that the perceptual and cognitive processes of visual word recognition and reading operate upon neuronally coded domain-general low-level visual representations -typically oriented line representations. We here demonstrate, consistent with neurophysiological theories of Bayesian-like predictive neural computations, that prior visual knowledge of words is utilized to 'explain away' redundant and highly expected parts of the visual percept. Subsequent processing stages, accordingly, operate upon an optimized representation of the visual input, the orthographic prediction error, highlighting only information relevant for word identification. We show that this informationally optimized representation is related to orthographic word characteristics, accounts for word recognition behavior, and is processed early in the visual processing stream, i.e., in occipital cortex and before 200 ms after word-onset. Based on these findings, we propose that prior visual-orthographic knowledge is used to optimize the representation of visually presented words, which in turn allows for highly efficient reading processes.

Development of neural specialization for print: Evidence for predictive coding in visual word recognition

“…Previous studies have repeatedly shown that predictive coding may underlie object recognition [25], low-level visual processing [26], action planning [27], or psychiatric disorders [28]. The present study supports the presence of the predictive coding mechanisms in visual word recognition (also see [29]) and adds converging evidence that predictive coding may be a general mechanism of information processing in the brain [30,31].…”

Can Prediction Error Explain Predictability Effects on the N1 during Picture-Word Verification?

“…However, prediction error magnitude should also be expected to vary continuously, from unpredicted word forms that are less to more orthographically similar to the predicted word form. This is comparable to Gagl et al's (2020) use of a pixel distance metric to calculate the continuous distance between a presented word form and a context-neutral prior. Such an approach could be applied to biasing contexts by instead calculating the orthographic distance between a presented word form and a context-informed prior, where the probability of observing certain pixels (or orthographic features) could be up-weighted proportional to prediction certainty.…”

“…More specifically, a predictive coding account of orthographic processing would expect error signals to vary as a function of the size of the difference between a general orthographic prior (e.g., an average word form) and a presented word form. Some recent findings appear to support the notion that the N1 reflects a neutral-context error signal, with greater distance from an orthographic prior eliciting greater amplitude (Gagl et al, 2020), while the profile of the N1's sensitivity to word form regularity over experience matches that expected under a predictive coding account (Huang et al, 2022; Zhao et al, 2019.…”