Priming of Abstract Letter Identities in the Letter Match Task

Abstract

Previous studies of letter recognition have not found priming for abstract letter identities. We used a task that required participants to decide whether a target is the same or different from a reference letter presented in opposite case, which avoids the shortcomings of tasks used in previous studies. We found robust priming effects in this task, which were the same size for letter pairs that have similar visual features across case (e.g., c/C, x/X) and dissimilar features (e.g., a/A, b/B). Also, the pattern of priming was the same whether the prime was in the same or different case as the reference. We take these findings as evidence that abstract letter identities support priming in this task. We suggest that the same-different match task is a useful tool for studying representations used to support masked priming in letter recognition and with other stimuli with limited set size.

Letter Visibility and the Optimal Viewing Position Effect of Isolated Connected and Un-Connected Letters in Arabic

“…Th is fi nding would suggest that participants rely on abstract lett er representations, regardless of case (lowercase/uppercase for the Roman alphabet) or lett er position (isolated/initial/middle/fi nal for the Arabic alphabet). Th e present data are consistent with the fi ndings obtained by Kinoshita and Kaplan (2008) with the Roman alphabet in a cross-case lett er match task (i.e., similar priming for a-A and c-C). Likewise, the present data are also consistent with previous fi ndings obtained by Kinoshita and Norris (2009) using the masked priming same-diff erent task with English words.…”

“…Several possible forms of mental representations (categories) have been suggested in the literature as underlying the processing of isolated lett ers (Miller & Vaknin, 2012): (1) abstract visual representations that depict the critical visual features of lett ers (e.g., Mycroft et al, 2002;Posner & Mitchell, 1967); (2) abstract phonemic representations that represent the critical phonological characteristics of how lett ers are pronounced (e.g., Carrasco et al, 1988); (3) abstract nominal Unauthenticated Download Date | 5/10/18 4:59 PM representations that is, lett er names (e.g., Posner & Mitchell, 1967); and (4) abstract non-code-specifi c representations that refl ect the convention that two diff erent lett er shapes (e.g., A a) are the same lett er (e.g., Kinoshita & Kaplan, 2008;Miozzo & Caramazza, 1998). It is worth mentioning that in the Arabic alphabet, the words for the diff erent lett ers have similar voice onsets.…”

Facilitation versus Inhibition in the Masked Priming Same–Different Matching Task

“…Recently, Perea and Acha (2009) and Kinoshita and Norris (2010) avoided the above-cited confound by manipulating not the relationship between prime and target, but the relationship between probe and prime (i.e., a zero-contingency experiment; see also Kinoshita & Kaplan, 2008, for a similar experiment with single letters). For "same" trials, the comparison was exactly the same as that in previous experiments (e.g., house -house-HOUSE vs. house -water-HOUSE), whereas for "different" trials, the comparison was between trials like field -field -HOUSE versus field -water -HOUSE.…”

“…For "different" responses, response times were higher (and error rates higher) in the related condition than in the unrelated condition (see Kinoshita & Kaplan, 2008;Perea & Acha, 2009), although the effect only reached the classical criterion for significance in the single letter experiment of Kinoshita and Kaplan (2008). This inhibitory trend can also be easily explained by the Bayesian Reader model: Under a zero-contingency scenario, in the sequence field-field-HOUSE, the prime field is the same as the probe; this means that the prime field generates evidence supporting a "same" response.…”

Stimulus/Response Learning in Masked Congruency Priming of Faces: Evidence for Covert Mental Classifications?

“…Importantly, this priming would not require unconscious semantic processing of the prime (Abrams & Greenwald, 2000). Nonetheless, later studies of masked priming controlled for such S-R learning-for example, by using "novel" primes (i.e., primes that were never seen as probes; Greenwald et al, 2003;Klauer et al, 2007;Naccache & Dehaene, 2001b;Quinn & Kinoshita, 2007), or by using tasks in which S-R learning is unlikely to contribute to priming (Kinoshita & Kaplan, 2008;Kinoshita & Norris, 2009)-and still found reliable priming. Indeed, a recent meta-analysis supported at least two contributions to masked semantic priming: S-R learning and unconscious semantic access (Van den Bussche, Van Den Noortgate, & Reynvoet, 2009).…”