Cue integration in spatial search for jointly learned landmarks but not for separately learned landmarks.

Abstract

The authors investigated how humans use multiple landmarks to locate a goal. Participants searched for a hidden goal location along a line between 2 distinct landmarks on a computer screen. On baseline trials, the location of the landmarks and goal varied, but the distance between each of the landmarks and the goal was held constant, with 1 landmark always closer to the goal. In Experiment 1, some baseline trials provided both landmarks, and some provided only 1 landmark. On probe trials, both landmarks were shifted apart relative to the previously learned goal location. Participants searched between the locations specified by the 2 landmarks and their search locations were shifted more toward the nearer landmark, suggesting a weighted integration of the conflicting landmarks. Moreover, the observed variance in search responses when both cues were presented in their normal locations was reduced compared to the variance on tests with single landmarks. However, the variance reduction and the weightings of the landmarks did not always show Bayesian optimality. In Experiment 2, some participants were trained only with each of the single landmarks. On subsequent tests with the 2 cues in conflict, searching did not shift toward the nearer landmark and the variance of search responses of these single-cue trained participants was larger than their variance on single-landmark tests, and even larger than the variance predicted by using the 2 landmarks alternatively on different trials. Taken together, these results indicate that cue combination occurs only when the landmarks are presented together during the initial learning experience. (PsycINFO Database Record

Spatial integration during performance in pigeons

“…Our results involving compounds of spatial cues do not show evidence of spatial averaging, but instead show discrete responses at individual target locations. Similar results have been found for pigeons using landmarks during foraging (e.g., (Spetch et al, 1996), and humans in spatial search tasks (Baguley et al, 2006; Du et al, 2017). Why compounds of spatial cues would not produce averaging, but compounds of temporal cues would remains a mystery and is a topic in need of future research.…”

“…Such within-compound associations have been shown to be a critical determinant of comparator effects. Similar research has found that humans fail to integrate separately-learned spatial cues when presented in compound at test (Baguley et al, 2006; Du et al, 2017). Instead, like the pigeons in Experiment 2, spatial search in humans on compound tests appears to be guided by each cue independently, with factors like proximity to the target as strong determinants for weighting cue choice.…”

Integration of visual landmark cues in spatial memory

“…The discrepancy in findings between Baguley et al ( 2006 ) and Du et al ( 2017 ) might be attributable to some key methodological differences. For example, Baguley et al ( 2006 ) did not vary the absolute location of the landmarks and horizontal line on the computer screen, which may have allowed for participants to encode the target relative to the edges of the screen, while Du et al ( 2017 ) varied the absolute location of the landmarks and horizontal (or vertical) line while keeping their relative distances constant. Baguley et al ( 2006 ) also had participants learn many stimulus–target pairs during learning, requiring participants to encode more information than might have been possible.…”

“…This pattern maintained even when participants learned the location of the target in the presence of both landmarks (Experiment 3). However, recent findings by Du et al ( 2017 ) using a similar task found that participants optimally combined two landmark cues when estimating the location of a target on both horizontal and vertical axes.…”

Relative reinforcer rates determine pigeons’ attention allocation when separately trained stimuli are presented together

“…In contrast, when stimuli are trained separately, no division of attention is required because only one stimulus is presented at a time. Thus, when stimuli are later presented simultaneously, subjects may not actually divide their attention between the stimuli-instead, they may attend selectively to one stimulus (Du, McMillan, Madan, Spetch, & Mou, 2017;Yokoyama, Dailey, & Chase, 2006; see also Heinemann, Chase, & Mandell, 1968;Leith & Maki, 1975). If such selective attention occurs, whether it is controlled by the relative reinforcer rates previously associated with each stimulus is unknown.…”