Prediction advantage as retrieval interference: an ACT-R model of processing possessive pronouns

Abstract

We propose a retrieval interference-based explanation of a prediction advantage effect observed in Stone et al. (2021). They reported two dual-task eye-tracking experiments in which participants listened to instructions involving German possessive pronouns, e.g. ‘Click on his blue button’, and were asked to select the correct object from a set of objects displayed on screen. Participants’ eye movements showed predictive processing, such that the target object was fixated before its name was heard. Moreover, when the target and the antecedent of the pronoun matched in gender, predictions arose earlier than when the two genders mismatched — a prediction advantage. We propose that the prediction advantage arises due to similarity-based interference during antecedent retrieval, such that the overlap of gender features between the antecedent and possessum boosts the activation level of the latter and helps predict it faster. We report an ACT-R model supporting this hypothesis. Our model also provides a computational implementation of the idea that prediction can be thought of as memory retrieval. In addition, we provide a preliminary ACT-R model of how linguistic processes could drive changes in visual attention.

Modeling prominence constraints for German pronouns as weighted retrieval cues

“…The cue-based retrieval theory (CBR, henceforth) proposed in Lewis and Vasishth (2005) and Lewis, Vasishth, and Van Dyke (2006) has been successfully applied to model the memory retrieval processes involved in forming dependencies between two linguistic units such as nounverb agreements (Wagers, Lau, & Phillips, 2009) and pronoun-antecedent dependencies (Dillon, Mishler, Sloggett, & Phillips, 2013;Parker & Phillips, 2017;Patil, Vasishth, & Lewis, 2016;Patil & Lago, 2021). The CBR theory, which is implemented in the general cognitive architecture ACT-R (Anderson, Byrne, Douglass, Lebiere, & Qin, 2004), describes sentence processing as a series of activation-based skilled memory retrievals.…”

“…We considered this to be the baseline model because the specification of retrieval cues was the same as the earlier CBR models of antecedent retrieval (e.g. Patil & Lago, 2021) and it did not have any extension to consider the manipulation of prominence factors in the design of the Schumacher et al (2016) experiment. The predictions of the model, in terms of the antecedent preferences, are shown in Fig.…”

“…In ACT-R all retrieval cues have the same weight, but in psycholinguistics it has been proposed that certain retrieval cues could be weighted higher than others (see for example: Parker, Shvartsman, & Van Dyke, 2017;Vasishth, Nicenboim, Engelmann, & Burchert, 2019;Patil & Lago, 2021). To incorporate the stronger effect of agenthood, we added cue weighting and weight the thematic role cue higher than the other cues and tested if that improved the model performance.…”

“…

We propose an ACT-R model of processing German personal and demonstrative pronouns. The model extends existing cuebased retrieval models of sentence processing (Lewis & Vasishth, 2005;Lewis et al 2006) and pronoun resolution (Parker & Phillips, 2017;Patil & Lago, 2021) by adding prominence constraints as weighted retrieval cues. We model data from an antecedent selection task reported in Schumacher et al (2016).

…”

Computational cognitive modeling of predictive sentence processing in a second language

“…We consider Patil and Lago's (2021) model of processing possessive pronouns as the starting point. They modeled the visual-world eye-tracking data from Stone et al (2021) in ACT-R (Anderson et al, 2004) and the cue-based retrieval framework (CBR, henceforth) (Lewis and Vasishth, 2005;Lewis et al, 2006).…”