Breaking the Grid: The Use of Verbal Strategies in a Working Memory Task

Abstract

Inner speech, the silent mental articulation of language, plays a critical role in cognition, from planning actions to regulating emotions. Current models [1] propose that inner speech exists on a continuum of complexity, ranging from abstract, wordless representations to fully articulated sentences.  The LISP (Levels of Inner Speech) project at the Babelfisch Psycholinguistic Lab in Vienna aims to systematically compare the neural dynamics of three distinct levels —semantic, auditory-phonological, and motor articulatory  [2]— by combining a working memory paradigm with electroencephalography (EEG) and using the same exact stimulus material for all three conditions and a visual control task. In each trial, four everyday objects are presented in a 6×6 grid. Participants are instructed to either remember the objects (to induce inner speech) or recall their positions (for the visual control task).

However, debriefing data from the first study suggests that a significant number of participants relied on verbal strategies during the visual condition (e.g., by mentally labelling grid coordinates). The present study tests whether removing visible grid lines from the stimulus material suppresses verbal strategies for the visual task.

Participants are shown the four objects either inside a grid (A) or without the grid (B) for four seconds followed by a seven-second maintenance phase. Then, four crosses are shown inside a grid (A) or without the grid (B) and participants judge via button press whether the positions are correct. Conditions A and B are presented within-subject in separate blocks of 40 trials each. After the experiment, participants are asked to report on their strategies for both versions of the task via an open text box as part of a short survey. The study will be conducted online, aiming to recruit 50 adult native speakers of German.

The open-ended responses will be analyzed using a topic modeling algorithm to identify dominant strategy clusters.

We hypothesize that, by removing the grid lines from the stimulus material, participants will not be able to use verbal strategies like counting coordinates to solve the task and instead rely on visual strategies.

The results will determine whether grid-free stimuli effectively suppress verbal encoding, thereby guiding the selection of paradigms for the subsequent EEG studies and possibly allowing the LISP project to move forward with an improved control condition. Additionally, the present study will serve as a trial run for employing quantitative content modeling on future debriefing data, thereby avoiding subjectivity introduced through the examiner.

References

[1] R. Grandchamp et al., “The CONDIALINT model: condensation, dialogality, and intentionality dimensions of inner speech within a hierarchical predictive control framework,” Frontiers in Psychology, vol. 10, Sep. 2019. doi: 10.3389/fpsyg.2019.02019.

[2] A. Rodriguez-Fornells, P. León-Cabrera, A. Gabarros, and J. Sierpowska, “Inner speech brain mapping. Is it possible to map what we cannot observe?,” in Springer eBooks, 2021, pp. 381–409. doi: 10.1007/978-3-030-75071-8_23.