Search Papers | Poster Sessions | All Posters
Poster A73 in Poster Session A - Tuesday, August 6, 2024, 4:15 – 6:15 pm, Johnson Ice Rink
Unfolding mental spaces onto a plane.
Frédéric Gosselin1 (), Aya Azbane1, Nikolaus Kriegeskorte2, Ian Charest1; 1Département de psychologie, Université de Montréal, Montréal, Québec, Canada, 2Zuckerman Mind Brain Behavior Institute, Columbia University, New York, USA
Methods for collecting similarity judgments face significant challenges, including task efficiency and changes in strategy. Here we introduce the unfolding task, which enables efficient acquisition of similarity judgments protected against strategy change. On each trial, the participant is presented with a planar graph of the items whose similarities are to be judged. The task is to adjust the graph with mouse drag-and-drop operations such that the lengths of the edges accurately reflect the similarities among the items. Critically, only a subset of the item pairs are connected by edges on each trial, and the similarities among connected items can be conveyed without distortion. All pairwise judgments are acquired across multiple trials. Like the multiarrangement task (MAT), the unfolding task uses 2D arrangement of items. However, unfolding the graph avoids the distortion of distances in the 2D arrangement. We demonstrate the accuracy of our method with a simple experiment with known ground-truth similarities. We compared the ability of the unfolding task to recover the ground-truth similarities with the current state-of-the-art MAT. The next step will be to test the task with larger, more complex naturalistic stimulus spaces.
Keywords: similarity judgments psychophysics