Spatial Updating Of Single Targets

The first study (Loomis et al., 2002) dealt with the spatial updating of a single target (Figure 3-1 C). The target location on a given trial was conveyed either by a loudspeaker positioned there and emitting recorded speech ("Speaker 3") or by an utterance of the form "2 o'clock, 12 feet" spoken by the experimenter who was standing near the participant. In the former case, the participant would localize the sound in 3-D perceptual space and then form a spatial image of its perceived location (Figure 3-1A), and, in the latter case, form a spatial image based on the meaning of the utterance (Figure 3-1 B). For either input modality (3-D sound or spatial language), once the spatial image has been formed, behavior ought to be flexibly directed toward the spatial image while the participant moves through environment without further sensory information about the initially specified target location. To the extent that spatial images from perception and language are functionally equivalent, participants ought to show the same accuracy of updating performance, especially when care has been taken to match the locations of the spatial images of the two modalities. Before we began this research, we did not know whether spatial updating of targets specified by language was even possible.

Ten of the participants were sighted and 6 were blind from early in life. The 4 panels of Figure 3-3 depict the layouts for the experiment. The blindfolded participant stood at the origin facing in the direction corresponding to the top of the figure. The locations of the targets used in a grassy field are depicted by the X's. The nominal target distances ranged from 1.83 m to 4.27 m (6 to 16 ft) and the nominal target azimuths ranged form -90° (left of straight ahead) to 90° (right of straight ahead). On a given trial the participant heard speech from the target speaker in the field or the utterance by the experimenter. The participant then walked without vision and with sound-blocking hearing protectors to the estimated target locations. On some trials, the participant walked directly to the target and on other trials, was first led by the experimenter to the turn point 2.7 m in front and then attempted to walk unaided the rest of the way to the target. In the case of spatial language, the performance on direct walking trials is, by itself, not of great interest, because participants need not update relative to a spatial image in order to perform well. Given an utterance like "10'oclock, 8 feet", the participant need only turn to 10'oclock and walk what seems like 8 ft. However, this strategy does not work for the indirect walking trials. To perform well on these trials, the participant needs to convert the meaning of the utterance into a spatial image and then perform updating with respect to it (panels B and C of Figure 3-1).

If participants performed spatial updating perfectly, they should have walked to the same location on direct and indirect trials for the same nominal stimulus. The measure of updating performance used was the spatial separation between the terminal points of the walking trajectories on direct trials and the terminal points on indirect trials. The 4 panels of Figure 3-3 give the results in the form of centroids of terminal points, averaged over participants, for the direct and indirect paths to the different target locations. Sighted participants performed better as shown by the smaller separations between direct and indirect terminal points. However, because the separations are all quite small in comparison with the target distances and the distances walked, the data indicate spatial updating for both 3-D sound and for language, as well as for both groups of participants. The statistical analysis revealed just slightly poorer updating performance overall for language compared to auditory perception.

Figure 3-3. Stimulus layout and results of an experiment on spatial updating of single auditory targets ("3-D sound") and single targets specified by spatial language (Loomis et al, 2002). The experiment was conducted outdoors with both blind and blindfolded sighted participants. While standing at the origin, the participant heard a sound from a loudspeaker at one of the locations (X) or heard an utterance specifying one of the same locations (e.g., "10 o'clock, 12 feet"). The participant then attempted to walk to the location either directly or indirectly. In the latter case, the participant was guided forward 2.74 m to the turn point and then walked the rest of the way to the estimated target location. The open circles are the centroids of the indirect path stopping points, and the closed circles are the centroids of the direct path stopping points. Reprint of Figure 7 from Loomis et al. (2002). Spatial updating of locations specified by 3-D sound and spatial language. Journal of Experimental Psychology: Learning, Memory, & Cognition, 28, 335-345. Reprinted with permission.

Figure 3-3. Stimulus layout and results of an experiment on spatial updating of single auditory targets ("3-D sound") and single targets specified by spatial language (Loomis et al, 2002). The experiment was conducted outdoors with both blind and blindfolded sighted participants. While standing at the origin, the participant heard a sound from a loudspeaker at one of the locations (X) or heard an utterance specifying one of the same locations (e.g., "10 o'clock, 12 feet"). The participant then attempted to walk to the location either directly or indirectly. In the latter case, the participant was guided forward 2.74 m to the turn point and then walked the rest of the way to the estimated target location. The open circles are the centroids of the indirect path stopping points, and the closed circles are the centroids of the direct path stopping points. Reprint of Figure 7 from Loomis et al. (2002). Spatial updating of locations specified by 3-D sound and spatial language. Journal of Experimental Psychology: Learning, Memory, & Cognition, 28, 335-345. Reprinted with permission.

These results indicate that spatial images from perception and language exhibit near functional equivalence with respect to the processes involved in spatial updating of one target at a time.

Of secondary interest is the fact that the terminal points for the 3-D sound condition were generally somewhat displaced in distance from the target locations. This is consistent with other research showing that the perceived distance of sound sources is contracted relative to the range of physical source distances, with under-perception of far distances being the norm (e.g., Loomis et al., 1999; Zahorik et al., 2005).

5. SPATIAL UPDATING OF MULTIPLE TARGETS

The second of the three studies dealt with spatial updating of multiple targets (Klatzky et al., 2003). Because the previous study involved only a single target, working memory was sufficient for retaining the target location. In this second study, participants were unable to retain up to 5 targets specified by language and auditory perception in working memory. Thus, in order to examine updating of multiple targets, we had participants spend up to 10 minutes learning the multiple target locations to criterion during a learning phase. They were subsequently tested on updating performance during a test phase. Thus, this study involved spatial images held in long term memory.

Unlike the previous study, this was done indoors where stronger reverberation cues likely resulted in the slightly more accurate auditory distance perception that was observed. There were two experiments in this study. In the first, participants used a pointer to indicate their estimates of the directions of the targets (before and after updating) and used verbal reports to indicate their estimates of the distances of the targets. In the second experiment, participants used the same procedure of direct and indirect walking to targets as in the previous study. Based on a recent analysis by Loomis and Philbeck (in press), we now believe that verbal reports of distance are systematically biased toward underestimation, producing what appears to be an updating error that is added to whatever true updating error there is. Accordingly, we focus here on the second experiment involving direct and indirect walking, which resulted in much better updating performance. Besides the use of multiple targets in an indoor setting, the experiment differed from the previous study in including a vision condition as well as including indirect walking in which participants sidestepped to the right while facing forward.

Figure 3-4 gives the spatial layouts for the experiment. The vision and language targets, indicated by the X's, were at the same nominal locations, ranging in distance from 0.91 m to 3.66 m (3 to 12 ft). Because of the tendency for under-perception of auditory distance, the auditory stimuli were delivered by loudspeakers at head level that were placed at slightly larger distances, 1.22 to 4.57 m (4 to 15 ft), in order to produce perceived distances close to those of the visual targets, which were perceived quite accurately. The turn points ("indirect waypoints") for indirect walking were either 2.5 m in front or to the right side of the origin. In the learning phase, participants in the 3-D sound condition heard spoken labels (e.g., "baby", "cat") from the positioned loudspeakers. In the language condition, participants heard the utterance of the target coordinates followed by two presentations of the label. For both 3-D sound and language conditions, synthetic speech of a male voice was used. In the vision condition, participants saw labels presented at head level. On a learning trial, the participant was exposed to each of the 5 targets and then, when prompted with a label, attempted to report the direction and distance of the target using a pointer and verbal report, respectively. The learning phase terminated when both pointing accuracy and accuracy of distance reports (assessed using rank order correlation) met strict criteria. In the updating phase, participants responded to each target label by walking either directly to the target or walking indirectly to the target after being passively guided to the turn point.

The results are given in Figure 3-4. As in the previous study, updating performance was measured by the separation between the terminal points of the direct and indirect paths. The generally small separations in all 6 panels of Figure 3-4 indicate that updating performance was good for all conditions.

Figure 3-4. Stimulus layout and results of an experiment on spatial updating of multiple auditory targets ("3-D sound"), visual targets, and targets specified by spatial language (Klatzky et al., 2003). The experiment was conducted indoors. While standing at the origin during the learning phase, the participant memorized the locations of multiple targets (Xs) presented using vision, spatial hearing, or spatial language. Then on each trial during the test phase, the participant attempted to walk to one of the locations either directly or indirectly. In the latter case, the participant was guided forward and sideways 2.5 m to the turn point and then walked the rest of the way to the target. The open circles are the centroids of the indirect path stopping points, and the closed circles are the centroids of the direct path stopping points. Modified version of Figure 4 from Klatzky et al. (2003). Encoding, learning, and spatial updating of multiple object locations specified by 3-D sound, spatial language, and vision. Experimental Brain Research, 149, 48-61. Reprinted with kind permission of Springer Science and Business Media.

Figure 3-4. Stimulus layout and results of an experiment on spatial updating of multiple auditory targets ("3-D sound"), visual targets, and targets specified by spatial language (Klatzky et al., 2003). The experiment was conducted indoors. While standing at the origin during the learning phase, the participant memorized the locations of multiple targets (Xs) presented using vision, spatial hearing, or spatial language. Then on each trial during the test phase, the participant attempted to walk to one of the locations either directly or indirectly. In the latter case, the participant was guided forward and sideways 2.5 m to the turn point and then walked the rest of the way to the target. The open circles are the centroids of the indirect path stopping points, and the closed circles are the centroids of the direct path stopping points. Modified version of Figure 4 from Klatzky et al. (2003). Encoding, learning, and spatial updating of multiple object locations specified by 3-D sound, spatial language, and vision. Experimental Brain Research, 149, 48-61. Reprinted with kind permission of Springer Science and Business Media.

The statistical analysis revealed no significant differences between vision and 3-D sound, but language was just slightly worse than vision. Overall, the results support near functional equivalence of spatial images from perception and language.

0 0

Post a comment