Evaluating Experience-Based Geologic Field Instruction: Lessons Learned from A Large-Scale Eye-Tracking Experiment

Abstract:

A course centered on experience-based learning in field geology has been offered ten times at the University of Rochester. The centerpiece of the course is a 10-day field excursion to California featuring a broad cross-section of the geology of the state, from the San Andreas Fault to Death Valley. Here we describe results from a large-scale eye-tracking experiment aimed at understanding how experts and novices acquire visual geologic information. One ultimate goal of the project is to determine whether expert gaze patterns can be quantified to improve the instruction of beginning geology students. Another goal is to determine if aspects of the field experience can be transferred to the classroom/laboratory. Accordingly, ultra-high resolution segmented panoramic images have been collected at key sites visited during the field excursion. We have found that strict controls are needed in the field to obtain meaningful data; this often involves behavior atypical of geologists (e.g. limiting the field of view prior to data collection and placing time limits on scene viewing). Nevertheless some general conclusions can be made from a select data set. After an initial quick search, experts tend to exhibit scanning behavior that appears to support hypothesis testing. Novice fixations appear to define a scattered search pattern and/or one distracted by geologic noise in a scene. Noise sources include modern erosion features and vegetation. One way to quantify noise is through the use of saliency maps. With the caveat that our expert data set is small, our preliminary analysis suggests that experts tend to exhibit top-down behavior (indicating hypothesis driven responses) whereas novices show bottom-up gaze patterns, influenced by more salient features in a scene. We will present examples and discuss how these observations might be used to improve instruction.