Mislocated fixations during reading and the inverted optimal viewing position effect

Refixation probability during reading is lowest near the word center, suggestive of an optimal viewing position (OVP). Counterintuitively, fixation durations are largest at the OVP, a result called the Inverted Optimal Viewing Position (IOVP) effect (Vitu, McConkie, Kerr, and O'Regan, 2001). Current models of eye-movement control in reading fail to reproduce the IOVP effect. We propose a simple mechanism for generating this effect based on error-correction of mislocated fixations due to saccadic errors (Nuthmann, Engbert, and Kliegl, in press). First, we propose an algorithm for estimating proportions of mislocated fixations from experimental data yielding a higher probability for mislocated fixations near word boundaries. Second, we assume that mislocated fixations trigger an immediate start of a new saccade program causing a decrease of associated durations. Thus, the IOVP effect could emerge as a result of a coupling between cognitive and oculomotor processes.