Repeated letters increase the ambiguity of strings: Evidence from identification, priming and same-different tasks

Letters are often repeated in words in many languages. The present work explored the mechanisms underlying processing of repeated and unique letters in strings across three experimental paradigms. In a 2AFC perceptual identification task, the insertion but not the deletion of a letter was harder to detect when it was repeated than when it was unique (Exp. 1). In a masked primed same-different task, deletion primes produced the same priming effect regardless of deletion type (repeated, unique; Exp. 2), but insertion primes were more effective when the additional inserted letter created a repetition than when it did not (Exp. 3). In a same-different perceptual identification task, foils created by modifying a repetition, by either repeating the wrong letter or substituting a repeated letter, were harder to reject than foils created by modifying unique letters (Exp. 4). Thus, repetition effects were task-dependent. Since considering representations alone would suggest repetition effects would always occur or never occur, this indicates the importance of modelling task-specific processes. The similarity calculations embedded in the Overlap Model (Gomez et al., 2008) appeared to always predict a repetition effect, but its decision rule for the task of Experiment 1 allowed it to predict the asymmetry between insertions and deletions. In the Letters in Time and Retinotopic Space (LTRS; Adelman, 2011) model, repetition effects arise only from briefly presented stimuli as their perception is incomplete. It was therefore consistent with Experiments 2-4 but required a task-specific response bias to account for the insertion-deletion asymmetry of Experiment 1.