The structure of cognitive strategies for wayfinding decisions.

Literature proposes five distinct cognitive strategies for wayfinding decisions at intersections. Our study investigates whether those strategies rely on a generalized decision-making process, on two frame-specific processes-one in an egocentric and the other in an allocentric spatial reference frame, and/or on five strategy-specific processes. Participants took six trips along a prescribed route through five virtual mazes, each designed for decision-making by a particular strategy. We found that wayfinding accuracy on trips through a given maze correlated significantly with the accuracy on trips through another maze that was designed for a different reference frame (rbetween-frames = 0.20). Correlations were not significantly higher if the other maze was designed for the same reference frame (rwithin-frames = 0.19). However, correlations between trips through the same maze were significantly higher than those between trips through different mazes that were designed for the same reference frame (rwithin-maze = 0.52). We conclude that wayfinding decisions were based on a generalized cognitive process, as well as on strategy-specific processes, while the role of frame-specific processes-if any-was relatively smaller. Thus, the well-established dichotomy of egocentric versus allocentric spatial representations did not translate into a similar, observable dichotomy of decision-making.

Choice between decision-making strategies in human route-following.

To follow a prescribed route, we must decide which way to turn at intersections. To do so, we can memorize either the serial order of directions or the associations between spatial cues and directions ("at the drug store, turn left"). Here, we investigate which of these two strategies is used if both are available. In Task S, all intersections looked exactly alike, and participants therefore had to use the serial order strategy to decide which way their route continued. In Task SA, each intersection displayed a unique spatial cue, and participants therefore could use either strategy. In Task A, each intersection displayed a unique cue, but the serial order of cues varied between trips, and participants therefore had to use the associative cue strategy. We found that route-following accuracy increased from trip to trip, was higher on routes with 12 rather than 18 intersections, and was higher on Task SA than on the other two tasks, both with 12 and with 18 intersections. Furthermore, participants on Task SA acquired substantial knowledge about the serial order of directions as well as about cue-direction associations, both with 12 and with 18 intersections. From this we conclude that, when both strategies were available, participants did not pick the better one but rather used both. This represents dual encoding, a phenomenon previously described for more elementary memory tasks. We further conclude that dual encoding may be implemented even if the memory load is not very high (i.e., even with only 12 intersections).