Local and Global Probability Affects on Maze Navigation

Sixuan Chen, a masters student working with me as been dealing with the challenges of being a student and conducting research at the University of Waterloo while the pandemic confines her to her home country of China. She has overcome these challenges to get a nice project up and running where participants navigate mazes in the light of local cues (arrows saying go right or go left) and global cues (the exit is likely to be here (or maybe there)). These cues are stochastic and variable and we make the mazes particularly hard by masking the walls of the maze so that the navigation is dependent on the local cues and global goal information only.

Like we see with Isabella's work on local and global probability cues in a visual prediction task, the local cue seems the more potent, but Sixuan' work (still preliminary at this stage) shows a definite impact of the global probability cue and the impact varies, as one would expect, with the reliability of the two information sources. What is interesting about this is that the scale of the cue, near-by, local, direct, and immediate or global, distant, and far into the future seem to be independently contributing to performance.

Sixuan has a poster of this work in progress. If you have ideas or suggestions she would love to hear from you.