Anna K. Lawrance¹ (), Mateusz Janiszewski², Hilda Deborah³, Andy J. Fraass¹, Duncan Johannessen¹, Lucinda J. Leonard¹, Dipendra J. Mandal³, Brett D. Roads⁴, James W. Tanaka¹; ¹University of Victoria, ²Aalto University, ³Norwegian University of Science and Technology, ⁴University College London

To the untrained eye, rocks offer limited perceptual information to aid in accurate categorizations—making geology an ideal domain to study the development of semantic and visuoperceptual knowledge. This study examined the formation and restructuring of perceived rock-type categories in undergraduates (N=48) enrolled in an introductory-level geology course. Through this work, we addressed three questions: 1) How do categories for rock knowledge develop? 2) How does the acquisition of real-world expertise reshape semantic and visuoperceptual categories? 3) Is the nature of one’s category restructuring indicative of academic performance? We quantified category restructuring trajectories using PsiZ, a machine learning package that generates a multi-dimensional category representation (i.e., psychological embedding) based on the participant’s similarity judgments (Roads & Love, 2020). On each trial, participants were presented with a visual array of nine rocks—depicted by images on visuoperceptual trials and labels (e.g., “basalt”) on semantic trials—and were asked to select the two most similar peripherally presented rocks to the reference rock. Visuoperceptual and semantic category structures were assessed at the start and end of the course. How does category structure relate to academic success? A comparison of the top and bottom 25% of students, based on lab-test performance, revealed diverging trajectories. Post-instruction, image and label judgments of high-performers were highly correlated demonstrating strong integration of visuoperceptual and semantic knowledge. Contrastingly, low-performers displayed faulty visuoperceptual and semantic knowledge as demonstrated by poorly differentiated and conceptually inaccurate clusters of rock images and labels. Strikingly, despite no differences in prior exposure to the field, the two groups showed distinct pre-instruction visuoperceptual structures. High-performers exhibited significantly greater exemplar differentiation; this differentiation (among high-performers) was then maintained, but reconfigured (to assemble conceptually-accurate rock-type clusters) post-instruction. These findings suggest that the groups approached the task, at both timepoints, with markedly different levels of perceptual sensitivity.

Acknowledgements: This research was funded by an NSERC Discovery Grant held by Dr. James W. Tanaka.