Advertisement

Attention, Perception, & Psychophysics

, Volume 80, Issue 2, pp 586–599 | Cite as

The role of top-down knowledge about environmental context in egocentric distance judgments

  • John W. Philbeck
  • Daniel A. Gajewski
  • Sandra Mihelič Jaidzeka
  • Courtney P. Wallin
Article
  • 170 Downloads

Abstract

Judgments of egocentric distances in well-lit natural environments can differ substantially in indoor versus outdoor contexts. Visual cues (e.g., linear perspective, texture gradients) no doubt play a strong role in context-dependent judgments when cues are abundant. Here we investigated a possible top-down influence on distance judgments that might play a unique role under conditions of perceptual uncertainty: assumptions or knowledge that one is indoors or outdoors. We presented targets in a large outdoor field and in an indoor classroom. To control visual distance and depth cues between the environments, we restricted the field of view by using a 14-deg aperture. Evidence of context effects depended on the response mode: Blindfolded-walking responses were systematically shorter indoors than outdoors, whereas verbal and size gesture judgments showed no context effects. These results suggest that top-down knowledge about the environmental context does not strongly influence visually perceived egocentric distance. However, this knowledge can operate as an output-level bias, such that blindfolded-walking responses are shorter when observers’ top-down knowledge indicates that they are indoors and when the size of the room is uncertain.

Keywords

3-D perception Space perception Perception and action Scene perception 

The ability to determine how far away objects are from ourselves is crucial for interacting effectively with the world. Much progress has been made in understanding how vision is used to determine self-to-object distances (i.e., egocentric distances). Past work has focused predominantly on identifying discrete sources of visual information (e.g., binocular parallax or absolute disparity, accommodation, angular declination, etc.) and on characterizing their relative effectiveness (e.g., Foley, 1991; Philbeck & Loomis, 1997; Sedgwick, 1986). In addition to this online visual information, however, some stored knowledge about the environment is almost always available. Even without vision, one typically knows whether one is indoors or outdoors, on the basis of one’s recent experience seeing the environment and navigating to the current location. In principle, this stored knowledge could influence one’s judgments of egocentric distance. For example, if one assumes that the surrounding environment is relatively small, very large distances would be quite unlikely, and this could tend to bias the range of overt judgments of distance (and possibly even the underlying perceived distances) toward shorter distances (Yang & Purves, 2003). Presumably, the influence of this kind of information would be largely outweighed when online visual information about object distances is abundant. Under conditions of perceptual uncertainty, however, expectations and assumptions (about the size of the environment, in this case) may well play a more prominent role (Geisler & Kersten, 2002). A variety of factors can create perceptual uncertainty by limiting the visual system’s ability to extract and process visual spatial cues (e.g., low light level, high attentional load, restricted field of view, and brief viewing durations; Barlow, 1958; Gajewski, Philbeck, Pothier & Chichka, 2010; Lunsman et al., 2008; Watson, 1986). These conditions occur often in real-world settings, and thus there are many everyday opportunities for this top-down knowledge to potentially influence how objects are localized. Testing for this kind of influence presents challenges, but this article describes a series of experiments designed to evaluate this idea, specifically with respect to knowledge that one is in an indoor versus an outdoor environment.

Indoor and outdoor environments differ along a number of dimensions: For instance, indoor environments are typically more enclosed, smaller, less “open,” and less “natural,” and also present more boundaries and constraints on possible actions than do outdoor environments. Although real environments vary in the extent to which they exhibit these properties, here we focused on more prototypical indoor versus outdoor environmental contexts—that is, those that can be assumed to exhibit most of the features commonly associated with these contexts. Some past work has predicted, and shown, differences in spatial judgments between indoor and outdoor environments. Specifically, several authors have noted that well-lit indoor settings are often more cluttered than outdoor settings, and thus offer a more “cue-rich” stimulus environment for localizing objects in distance—denser gradients of texture, binocular disparity, linear perspective, and so forth (e.g., Cutting & Vishton, 1995; Teghtsoonian & Teghtsoonian, 1970; Toye, 1986; Wagner, 1985; Wu, He, & Ooi, 2007). One consequence of relatively impoverished cue conditions is a tendency for perceived distances to be biased toward a single, specific distance, which varies between observers but is typically between 1 and 3 m (the so-called “specific-distance tendency”; Gogel, 1984). If outdoor environments on average offer a more impoverished constellation of distance and depth cues than do indoor environments, one might thus expect that outdoor distance judgments beyond this range should be smaller than would indoor judgments of the same physical object distances—that is, they should be more biased toward the specific-distance tendency. For example, a target placed beyond 3 m would be more prone to underestimation outdoors because the relatively impoverished distance information in that environment is insufficient to fully counteract the bias toward the 1- to 3-m range of the specific-distance tendency. This is indeed the general pattern (Andre & Rogers, 2006; Da Silva, 1985; Teghtsoonian & Teghtsoonian, 1970).

More recently, Lappin, Shelton, and Rieser (2006) manipulated environmental context (indoor hallway, indoor atrium, and outdoor lawn) and found evidence of an expansion of perceived egocentric distances indoors relative to outdoor environments. Witt, Stefanucci, Riener, and Proffitt (2007) found that egocentric distance judgments were larger when targets were seen in the context of the shorter end of a long indoor hallway than the longer end of the same hallway; although the environment was the same in both cases, the visual surroundings and spatial cues specified a smaller versus a larger context. However, neither of these studies was designed to isolate the cue bases underlying these context effects, and therefore, important details about the mechanism of these findings remain unknown. Another complication is that the context effects in Witt et al. differed by response mode. No context effects were found when blindfolded walking was directed to the target location itself. However, context effects were found when participants used blindfolded walking, in which they viewed a target, then closed the eyes and attempted to reproduce the target distance by walking in a direction opposite to the target. Context effects were also found when participants used a visual matching response, in which participants viewed a target and then directed an experimenter to position a second target to match the egocentric distance of the first. Fully accounting for response-mode differences is a sizeable challenge, and the Witt et al. study was not designed with that intention. Nevertheless, for the present purposes, these studies are intriguing because they suggest that distance judgments are larger in smaller environments, at least under some conditions.

One difficulty for any attempt to assess the possible role of assumptions about environmental context on perceived distance is that these assumptions are almost invariably confounded with the presence of strictly visual determinants of setting size. As we noted above, it is reasonable to assume that the weighting of top-down influences will be minimized when visual cues about object distance are abundant, so in an ideal test, some restriction of the visual cues would create a situation that would be favorable for such an influence to manifest. Our approach here was to use field-of-view (FOV) restriction to control the visual cues between indoor and outdoor environments. In this case, the occluder created a “reduced cue” context, one that occludes most or all of the visual features that would ordinarily signal that one is in an indoor or outdoor environment. Although it is difficult to achieve a complete match between the FOV-restricted indoor and outdoor stimulus views (e.g., some variations in lighting and surface texture might still occur), we argue that this manipulation is largely successful in controlling the purely visual cues. To manipulate participants’ assumptions or knowledge about the surrounding environmental context, we took advantage of their ability to remember how they had navigated to reach the testing area. We tested them in an indoor or an outdoor environment, on the assumption that they would be aware that they were indoors or outdoors. With the visual stimulus more or less the same between environments, any indoor–outdoor effect should be based on the participants’ top-down awareness of the environmental context.

These experiments also involved several different response modes: blindfolded walking, verbal reports, and “size gestures.” In the size gesture response, participants indicate the target’s size by gesturing with their unseen hands. Leveraging from evidence that size and distance judgments are linked, estimates of target size have been used as an indirect measure of perceived distance (for a review, see Epstein, Park, & Casey, 1961). Because participants do not report directly on the target distance, this measure provides a means of minimizing a particular form of response bias, or postperceptual processing—namely, using explicit reasoning to deduce the target’s distance, rather than reporting on its perceived distance (Philbeck & Witt, 2015). In addition to response bias, response mode differences in spatial judgments can stem from a wide variety of other factors, such as the locus of cortical processing and the specific aspect of perception thought to underlie the response (e.g., perceived distances vs. perceived locations or relative vs. absolute perception; Andre & Rogers, 2006; Foley, 1977; Loomis, da Silva, Fujita, & Fukusima, 1992; Philbeck & Loomis, 1997; Vishton, Rea, Cutting, & Nuñez, 1999). Our manipulation of response modes here was not intended to comprehensively discriminate between these distinctions. Instead, we used blindfolded walking and verbal reports because both are commonly used in indoor and outdoor environments (e.g., Andre & Rogers, 2006; see also Loomis et al., 1992, for an early review). Most crucially for our purposes, blindfolded-walking responses are constrained by the size of the environment, and this could be one source of the environmental-context effects. Verbal reports and size gestures are not similarly constrained, and thus provide a means of evaluating this issue.

Experiment 1

If assumptions about the size of the unseen environmental context influence visually perceived egocentric distance, this influence should manifest when vision is restricted to the target and the ground plane that immediately surrounds it. Under such circumstances, the visual determinants of environmental context are occluded and the visual cues to egocentric target distance are largely equated across environments. Presumably, however, observers retain the knowledge that they are indoors or outdoors, owing to their recent memory of traveling to the testing environment, and this knowledge is not impacted by FOV restriction.

This experiment tests three fundamental predictions. (1) Under FOV restriction, we predicted that egocentric distance judgments should be larger in outdoor than in indoor environments, in keeping with the top-down assumption that the distribution of possible distances is larger outdoors than indoors. (2) Under full-FOV conditions, we predicted that, if outdoor settings provide less abundant and/or less useful visual cues, underestimation should be greater outdoors than indoors.

Going further, comparing performance with and without FOV restriction would allow us to address the issue of whether purely visual cues that specify an environmental context may play a stronger role in some environments than others. As we mentioned earlier, there is general acknowledgement that the quality and quantity of visual cues might be higher indoors than outdoors, but little is currently known about the relative weightings of these cues in producing indoor–outdoor effects. Simply comparing distance judgments in indoor versus outdoor contexts under full-FOV viewing does not permit a clean assessment of the relative role of visual cues local to the target in comparison to visual cues about the environmental context, because a full FOV would typically confound local and contextual information in both settings. Comparing performance with and without FOV restriction across contexts would provide a measure of the extent to which purely visual cues to the environmental context play a role in influencing judgments of target distance, over and above the cues available in the region immediately local to the target. Accordingly, (3) our third prediction was that if visual cues specifying the environmental context are more abundant (or are weighted more strongly) in indoor than in outdoor environments, manipulating FOV should have a larger effect indoors than outdoors, because FOV restriction would occlude the more reliable or abundant contextual cues available in the indoor environment.

Method

Participants

Twenty-eight individuals from the George Washington University community participated in the experiment. All of the participants were 18 to 28 years of age, reported normal or corrected-to-normal vision, and were paid $10/h or participated in exchange for course credit.

Design

Environmental context (indoor vs. outdoor) and FOV (restricted vs. full view) were manipulated within subjects. All participants were tested in both contexts but with context blocked and context order counterbalanced. This counterbalancing was achieved by randomly assigning participants to one of two groups: the indoor-first group or the outdoor-first group. Within each context, FOV (restricted vs. full view) was also administered in blocks (i.e., blocks within blocks), though the restricted-view condition was always administered first within each context. Distance order was randomized within each block. Distance estimates were made via blindfolded walking (e.g., Gajewski et al., 2010; Gajewski, Philbeck, Wirtz, & Chichka, 2014; Philbeck & Loomis, 1997).

Stimuli and apparatus

The target was a thin (1 cm) wooden board (4.0 × 19.4 cm) painted bright yellow. It was placed flat on the ground at one of seven distances (3.0, 3.5, 4.0, 4.5, 5.0, 5.5, and 6.0 m) from the observer. Photographs of the indoor and outdoor environments are shown in Fig. 1a and b. The indoor environment was an empty laboratory that extended 9.5 m in depth from the participant’s viewing position (room dimensions were 5 × 11 m). The lighting was bright yet diffuse. The ceiling was covered with rectangular drop-ceiling tiles, and the walls were painted in a light cream color. The floor was covered with a light blue, low-pile carpet with no obvious repeating patterns (see Fig. 1a). The outdoor environment (see Fig. 1b) was a grassy campus square that extended approximately 23 m from the participant’s viewing position (the grassy dimensions were 17 × 38 m). The square was bounded by buildings and sidewalks, and at times pedestrians in the distance could be visible to the observer in the full-view condition. Although the outdoor environment was on campus and may have been seen by some participants at some point prior to their arrival to the session, they were nevertheless led onto the field without vision and without knowledge of their precise location on the field. Prior visual experience with the tested viewpoint was thus reasonably well-controlled.
Fig. 1

Photographs of the environments used in Experiments 14. White boxes show the approximate field of view in the restricted-view conditions. (a) Indoor environment with carpeted floor (Exps. 1 and 2 only) and with artificial turf (inset at lower right; Exps. 3 and 4 only). (b) Outdoor environment. Both full and restricted views were available in Experiments 1 and 2; only the restricted view was available in Experiments 3 and 4

In the restricted-view conditions, participants viewed the targets through an aperture that limited their FOV. To create the aperture, a set of clear laboratory goggles was covered with electrical tape, and a small opening was created in the tape that allowed a monocular view of the scene (right eye) approximately 14.5° high × 17.2° wide. Although not crucial for the purposes of this article, all conditions and all experiments here involved monocular viewing. This was achieved either with the goggle aperture (restricted-view conditions) or with an eye patch over the left eye (full-view conditions). Monocular viewing was chosen because it is simpler to implement monocular FOV restriction; binocular FOV restriction would require adjustment for each individual based on interpupillary distance. A blindfold was employed during walking, and hearing protectors were employed throughout the session in order to minimize the impact of auditory cues.

Procedure

In both contexts, participants were not allowed to see the testing space in advance. Before entering the testing space, they were instructed about how to use the blindfolded-walking response to indicate target distances—specifically, they were instructed to lower a blindfold after viewing the target, and then indicate the target’s remembered location by walking without vision and stopping at the location. Participants were outfitted with goggles, eye patch, blindfold, and hearing protectors, around the corner from the outdoor context and outside of the room when testing was indoors. Participants were led without vision into each context and maneuvered to the viewing position. In all conditions, the duration of viewing was controlled by a large black poster board held about 40 cm from the participant’s face. The head was held steady in all conditions. This procedure was employed to prevent visual exploration when viewing was restricted. At the beginning of each trial, the participant was instructed to raise the blindfold, and the experimenter then lowered the board to initiate viewing. The duration of viewing was approximately 5 s. The experimenter raised the board to terminate viewing and instructed the participant to close his or her eyes and lower the blindfold. The experimenter then said “Clear” to indicate that the target was removed and the participant should commence walking. Walking was accomplished without assistance and without error feedback.

In the restricted-view condition, viewing was through the aperture, but head direction was held steady. A head-angle calibration procedure was employed to ensure that participants began each trial with the target in view (i.e., so that head movement was not required to find the target). Outside the laboratory and before the experiment began, a marker was placed on a pole oriented vertically and positioned 0.5 m in front of the participant. On the basis of the participant’s eye height, the marker was positioned so that it would correspond to the gaze angle centered for the range of object distances employed. Participants then adjusted the goggles and their head angles to center the marker in their restricted FOV. At the beginning of each trial, the participants were instructed to raise the blindfold and open their eyes. With the posterboard obstructing their view, they were told to find the marker on the stick and then hold their head steady. The experimenter then put the stick to the side and lowered the posterboard to initiate viewing. In the full-view condition, vision was not restricted and participants were free to move their gaze about the scene, though head position was held steady throughout the trial.

Data analysis

The data were subjected to a mixed analysis of variance (ANOVA), with block order as a between-group variable, and context (indoor, outdoor), FOV (full, restricted view), and distance as within-subjects variables. Table 1 summarizes the data in terms of response sensitivity and bias. Response sensitivity represents the degree to which response distance differed systematically with differences in the distance of the target. It was calculated as the slope for each observer’s best-fitting regression line in each condition. Generally speaking, as differences in response sensitivity between conditions become more pronounced, the likelihood increases of a reliable Condition × Distance interaction in the ANOVAs presented below. Bias represents the overall tendency toward under- or overestimation. Table 1 reports the bias (i.e., mean error) as a percentage of the mean of the actual distances.
Table 1

Means and between-subjects standard errors for sensitivity (given by the slope relating response distance to target distance) and bias (given by the percent mean signed error across distances) in Experiments 1 and 2

 

Sensitivity

Bias

Exp. 1 (Blindfolded Walking) – Indoor First

 Indoor, restricted view

0.75 (0.07)

–25.8 (5.5)

 Indoor, full view

0.99 (0.06)

–12.7 (4.2)

 Outdoor, restricted view

0.99 (0.06)

–3.0 (6.1)

 Outdoor, full view

1.16 (0.08)

–3.1 (5.0)

Exp. 1 (Blindfolded Walking) – Outdoor First

 Indoor, restricted view

0.88 (0.09)

–13.0 (4.1)

 Indoor, full view

1.17 (0.08)

–0.5 (5.7)

 Outdoor, restricted view

1.07 (0.13)

8.5 (6.5)

 Outdoor, full view

1.23 (0.09)

0.1 (5.4)

Exp. 2 (Size Gesture*) – Indoor First

 Indoor, restricted view

0.46 (0.13)

–26.5 (9.6)

 Indoor, full view

0.86 (0.12)

–14.3 (8.5)

 Outdoor, restricted view

0.47 (0.11)

–33.1 (9.5)

 Outdoor, full view

0.71 (0.09)

–15.2 (11.2)

Exp. 2 (Size Gesture*) – Outdoor First

 Indoor, restricted view

0.62 (0.08)

–14.9 (4.9)

 Indoor, full view

1.01 (0.08)

1.1 (4.2)

 Outdoor, restricted view

0.36 (0.08)

–13.9 (6.3)

 Outdoor, full view

0.80 (0.12)

–2.6 (5.1)

Both experiments used carpet as the indoor floor covering. *These data represent perceived target distances along the ground, as inferred from size gesture responses using Eq. 1.

Results

Mauchly’s tests indicated that the sphericity assumption was violated for all contrasts involving the distance variable [all χ 2s (20) > 32.40, all ps < .041]. In the ANOVA results reported below, degree-of-freedom corrections were applied for these contrasts using Greenhouse–Geisser estimates of sphericity (εs = .447 [Distance], .730 [Context × Distance], .711 [FOV × Distance], and .706 [Context × FOV × Distance]).

Because we found no main effect of block order and no block order interactions (all ps > .10), collapsed data are depicted in Fig. 2. A main effect of distance can be seen, F(2.68, 69.79) = 224.41, p < .001, η p 2 = .896, indicating that responses were strongly sensitive to the physical target distances. Responses were greater in outdoor than in indoor contexts, F(1, 26) = 83.17, p < .001, η p 2 = .762, and this was qualified by a reliable Context × FOV interaction, F(1, 26) = 15.92, p < .001, η p 2 = .380. Pairwise comparisons (two-tailed t tests, with Bonferroni correction for multiple comparisons) showed that all pairs of contrasts were reliably different, ps < .0001, except for the comparison between the outdoor-full and outdoor-restricted conditions, p = .027, which fell above the Bonferroni-adjusted alpha level of .008. The means underlying this interaction were indoor–restricted = 3.63 m, indoor–full = 4.20 m, outdoor–restricted = 4.62 m, and outdoor–full = 4.43 m.
Fig. 2

Mean response distance as a function of target distance and field of view in Experiment 1. Performance in the indoor and outdoor contexts is shown on the left and right, respectively. Error bars show ±1 standard error; the thick lines show the best-fitting linear functions through the mean data. Carpet was the floor covering in the “indoor” viewing context

No main effect of FOV emerged, F(1, 26) = 2.55, p = .123, η p 2 = .089. There were, however, reliable Context × Distance, F(4.38, 113.84) = 3.36, p = .010, η p 2 = .115, and FOV × Distance, F(4.26, 110.90) = 6.30, p < .001, η p 2 = .195, interactions. We found no other reliable interactions (all Fs < 2.98, all ps > .096).

Discussion

Experiment 1 produced three key findings. First, in keeping with our first prediction, an effect of environmental context was observed in the restricted-view condition. Specifically, when the goggles were worn, participants made greater distance responses outdoors than indoors. This outcome provides clear support for the idea that one’s assumptions about the (unseen) environmental context play a role in performance, because the cues were similarly reduced in both viewing contexts. Second, contrary to our second prediction, no evidence emerged that there was more underestimation outdoors than indoors in the full-FOV condition. In fact, the opposite was true. Third, an effect of FOV restriction was observed only when indoors. This suggests that visual contextual cues have more influence indoors than outdoors.

One interpretation of these findings is that the performance differences between contexts with restricted view arose because observers were operating under different sets of assumptions. Indoors, the top-down assumption (under FOV restriction) was that the space would be small because it was known to be indoors, and the distance judgments were biased accordingly. When the FOV restriction was removed, distance judgments were larger, perhaps because the visual cues specified a larger space than had been assumed. Outdoors, the top-down assumption was of a large space, and the judgments were biased accordingly. When the FOV restriction was removed, the judgments did not change appreciably, perhaps because the visual cues specified an environment similar to the assumed size.

Although this interpretation can account for the observed data and maintains a role for top-down assumptions about the environment size on perceived egocentric distance, an alternative account remains viable. This alternative posits that the targets were in fact perceived closer outdoors than indoors in the full cue condition, in keeping with some past work (Andre & Rogers, 2006; Da Silva, 1985; Teghtsoonian & Teghtsoonian, 1970), but that this perceptual bias toward underestimation was offset by a response bias to walk farther when observers assumed they were in a large, outdoor environment. Analogously, the selective undershooting with restricted viewing indoors reflects a reluctance to produce the full extent of the intended walk without vision when the room size is unknown. This view predicts that evidence of context effects would be sensitive to the particular response modes used. More specifically, the blindfolded-walking response, because walking is constrained by the physical size of the environment, should exhibit larger differences between the indoor and outdoor environments; response modes that are not constrained by the environment size (e.g., verbal estimates of distance or distance estimates inferred from manual size judgments) should exhibit smaller context effects, if any. We tested this idea in Experiments 24. To foreshadow our results, these experiments provided evidence for the alternative interpretation—that is, that top-down knowledge of environmental size influenced the blindfolded-walking responses at the response execution stage, rather than influencing perception per se.

Experiment 2

Earlier, we posited that assumptions about the size of the environment have an impact on perceived distance. However, because indoor environments are enclosed spaces, an alternate possibility is that blindfolded-walking judgments are generally smaller indoors, particularly when the room size is uncertain. Consistent with this idea, we have run a number of studies indoors with viewing durations that arguably prohibited an appreciation of the room size, and these studies have generally shown underestimation at all distances (e.g., Gajewski et al., 2010; Gajewski, Philbeck, et al., 2014; Gajewski, Wallin, & Philbeck, 2014). Undershooting might occur, for example, if people are hesitant to walk indoors without vision because they are concerned about bumping into walls—walls that observers might reasonably assume to be fairly close if they are uncertain of the actual room size. If observers are concerned about bumping into a wall, their responses for farther targets—the is, those closer to the walls—might exhibit disproportionately more underestimation than responses for closer targets, perhaps resulting in a compressive, nonlinear pattern of results. To our knowledge, no studies have reported such a tendency. A complication is that some earlier studies involving blindfolded walking provided observers with several minutes’ exposure of walking without vision under the guidance of an experimenter, ostensibly to reduce any hesitation that observers might have felt to walk without vision (e.g., Philbeck & Loomis, 1997; Rieser, Pick, Ashmead, & Garing, 1995; Steenhuis & Goodale, 1988). Empirically, this practice does tend to reduce undershooting (Elliott, 1987), although subsequent work has suggested that the mechanism is likely due to perceptuomotor recalibration rather than decreased hesitancy (Philbeck, Woods, Arthur, & Todd, 2008). This means that some work involving blindfolded walking indoors might have failed to exhibit undershooting because a predisposition to underwalk when indoors was cancelled out by a tendency to walk farther due to perceptuomotor recalibration. Taken together, then, the evidence suggests that concerns about safety are unlikely to explain underwalking when indoors.

Nevertheless, there is a suggestion in the literature that walked distances indoors (even in multicue viewing contexts) can be somewhat shorter than responses for the same target distances in outdoor contexts because of different response attitudes or strategies (Iosa, Fusco, Morone, & Paolucci, 2012). Even if participants are not explicitly concerned about the wall—the experimenters do, after all, ensure their safety—the possibility remains that participants are less inhibited or constrained in their walking when outdoors. Larger, faster, and more confident strides might translate to the response to distance being elevated to some degree across the entire manipulated distance range. Whatever the source, rather than influencing the perception of distance, assumptions about size of the space might instead impact performance at a postperceptual output stage.

Experiment 2 investigated this possibility by closely matching the design of Experiment 1, but with a response mode that was not constrained by the size of the environmental context. In Experiment 2 we used the size gesture technique, in which participants indicated the physical size (the width) of an object by spreading their hands apart accordingly. According to the size–distance invariance hypothesis (see Kilpatrick & Ittelson, 1953), perceived distance may be inferred from indications of object size by using Eq. 1:
$$ \mathrm{perceived}\ \mathrm{distance}={\left(\mathrm{physical}\ \mathrm{distance}/\mathrm{physical}\ \mathrm{size}\right)}^{\ast }\ \mathrm{indicated}\ \mathrm{size}. $$
(1)

This equation is agnostic as to the source of errors in perceived distance. Thus, in principle it can be applied under a variety of conditions—for example, conditions that restrict distance and depth cues, as well as conditions that elicit misperception of distance by altering the perceived slant of the ground plane (Wu, Ooi, & He, 2004). There has been debate as to the extent to which this relationship holds true (e.g., Epstein et al., 1961; Haber & Levin, 2001; Mon-Williams & Tresilian, 1999), but many studies have shown at least some behavioral linkage between indications of distance and size (for a review, see Sedgwick, 1986). As such, even if this relationship does not hold exactly, the size gesture response provides some indication of changes in perceived distance between conditions, which is the focus of the present experiment. Critically, responses with this response mode should not be constrained by knowledge about the proximity of the back wall in the indoor context. If target distances indoors were perceived to be closer indoors when the context was obscured, presumably due to top-down assumptions about the environmental context, the FOV restriction should have an effect here similar to the one in Experiment 1. Conversely, if the context effect under FOV restriction that we observed in Experiment 1 was due to response (output) biases related to the blindfolded-walking response, the context effect should disappear in Experiment 2, in which a response mode was used that was not constrained by the environment size.

Method

Participants

Twenty-eight new individuals from the George Washington University community participated in the experiment. The data from two participants were not included in the analyses because on several trials an inferred ground distance could not be defined—the size gesture in these cases suggested an eye-to-target distance that was less than the observer’s eye height. This exclusion yielded a final sample size of 26. All of the participants were 18 to 30 years of age, reported normal or corrected-to-normal vision, and were paid $10/h or participated in exchange for course credit.

Stimuli and apparatus

The targets were bright yellow, rectangular sheets of foam placed on the floor at one of seven distances (3.0, 3.5, 4.0, 4.5, 5.0, 5.5, and 6.0 m) from the observer. Here, the physical size of the targets varied systematically with distance, which held the angular size of the target constant (subtending approximately 0.52 × 3.67 deg of visual angle) and provided a stimulus dimension to drive the physical size judgments. All trials were conducted in the same contexts as in Experiment 1 (see Fig. 1a and b) and with the same goggles, eye patch, blindfold, and hearing protectors.

Design and procedure

As in Experiment 1, environmental context (indoor vs. outdoor) and FOV (restricted view vs. full view) were manipulated within subjects. All participants were tested in both contexts, but with context blocked and context order counterbalanced; the restricted-view condition was always administered first within each context. The general procedure was the same as in Experiment 1. Here, however, the response mode was blindfolded size gesturing. Participants were informed that the targets would vary in terms of both size and distance, but that participants should always indicate the targets’ size. When responding, participants began with their hands flat together and spread them in parallel to indicate the width of the target object. This was accomplished without vision. The experimenter then measured and recorded the space between the hands. Participants practiced the response outside the laboratory with two desktop objects and were allowed to see their responses once they were completed. Size gestures during the experimental trials were performed without error feedback, and target distance was randomized within each block.

Data analysis

There are several ways to approach the analysis of size gesture data. Our primary interest was in the use of the size response as an indirect measure of perceived distance (Gogel, 1984; Loomis & Philbeck, 2008; Rand, Tarampi, Creem-Regehr & Thompson, 2011; Sedgwick, 1986). For ease of exposition and ready comparison between the response modes, we considered distance inferred from the size gesture. Using Eq. 1, above, we computed the perceived eye-to-target distance from the target’s physical size, its physical egocentric distance, and the participant’s size gesture response. Assuming that the observer knows the target is on the ground, we can then trigonometrically extract the perceived distance along the ground from the apparent eye-to-target distance. This approach was advantageous because effects on response sensitivity and bias could be compared more readily with more direct indications of distance (e.g., blindfolded walking and verbal reports). Critically, the patterns of statistical results generally did not differ from those based on direct analysis of the actual size responses. This was the case despite the fact that one participant had to be excluded from the inferred distance analysis. Several of this participant’s size responses were extraordinarily small, which suggested apparent distances that were smaller than the participant’s eye height and rendered ground distance undefined.

Results

As in Experiment 1, Mauchly’s tests indicated that the sphericity assumption was violated for all contrasts involving the distance variable [all χ 2s (20) > 30.94, all ps < .058]. In the ANOVA results reported below, degree-of-freedom corrections were applied for these contrasts using Greenhouse–Geisser estimates of sphericity (εs = .416 [distance], .668 [Context × Distance], .612 [FOV × Distance], and .674 [Context × FOV × Distance]).

We again observed no main effect of block order and no block order interactions (all ps > .144), so collapsed data are depicted in Fig. 3. Likewise, we again found a main effect of distance, F(2.49, 59.89) = 92.11, p < .001, η p 2 = .793. In contrast to Experiment 1, here there were no main effect of context, F(1, 24) = 1.22, p = .280, η p 2 = .048, and no Context × FOV interaction, F(1, 24) = 0.01, p = .934, η p 2 < .001. Also unlike Experiment 1, Experiment 2 revealed a main effect of FOV, F(1, 24) = 36.36, p < .001, η p 2 = .602, which was qualified by a reliable FOV × Distance interaction, F(3.67, 88.16) = 11.56, p < .001, η p 2 = .325. This interaction reflected a tendency toward reduced response sensitivity and smaller responses in restricted-view conditions (see Table 1). No other interactions were reliable (all Fs < 2.26, all ps > .068).
Fig. 3

Mean response distance (inferred from size gestures) as a function of target distance and field of view in Experiment 2. Performance in the indoor and outdoor contexts is shown on the left and right, respectively. Error bars show ±1 standard error; the thick lines show the best-fitting linear functions through the mean data. Carpet was the floor covering in the “indoor” viewing context

Discussion

Unlike in the results of Experiment 1, in which the walked indications of target distance were generally smaller indoors than outdoors, here we found no difference in the responses indoors versus outdoors. In addition, there was no evidence that this tendency was manifested differently under restricted versus full FOV. Put differently, FOV-restricted viewing had similar effects on the size gesture responses indoors versus outdoors. This contrasts with the findings of Experiment 1, in which an effect of FOV restriction was only found in the indoor environment.

Although fully accounting for the differences between Experiments 1 and 2 is beyond the scope of this article, we propose that the blindfolded-walking and size gesture responses were subject to different response-specific output biases. According to this view, the targets in both experiments were in fact perceived closer under FOV restriction, in keeping with past work (Creem-Regehr, Willemsen, Gooch, & Thompson, 2005; Wu et al., 2004). In Experiment 1, however, this perceptual bias toward underestimation under FOV restriction was offset by a response bias to walk somewhat farther when observers assumed they were outdoors, perhaps because they felt less constrained by the environment. This trade-off effectively nullified the behavioral manifestation of the perceptual underestimation caused by FOV restriction. In Experiment 2, a similar response bias was not manifested in the size gesture responses, presumably because the size gestures were not constrained by the physical size of the environment in the way that blindfolded-walking responses were.

Experiment 3

Experiments 1 and 2 provided conflicting evidence for our original hypothesis, namely that one’s top-down knowledge or assumptions about the environmental context might influence perceived distance under conditions of perceptual uncertainty: Experiment 1, using blindfolded walking, produced an effect of environmental context under FOV restriction, whereas Experiment 2, using size gestures, did not. As we have argued, restricting the FOV with a small aperture occludes most of the bottom-up (visual) cues that act to differentiate between indoor and outdoor environments. In Experiments 1 and 2, however, the visible ground plane textures differed between environments, with the indoor ground being covered by low-pile carpet and the outdoor ground being covered by grass. In principle, these visible differences (rather than top-down knowledge) could have been sufficient to drive the indoor–outdoor response differences in Experiments 1 and 2 in unforeseen ways—for example, by altering the optical slant of the ground plane—and possibly in a response-specific manner (He, Wu, Ooi, Yarbrough, & Wu, 2004). The best test of a possible influence of top-down knowledge of environment size on distance judgments would entail having exactly equal visible ground textures between indoor and outdoor environments. This ideal is very difficult to attain in real environments. In Experiment 3, however, we nevertheless attempted to come closer to this ideal by using an artificial-turf ground covering for the indoor environment. As in Experiment 1, we used blindfolded walking as the response mode, but here we focused on the restricted FOV. These conditions would provide a particularly well-controlled opportunity to observe influences of top-down knowledge of environment size under the blindfolded-walking response mode, in that the visual similarity of the stimulus and its immediate surroundings seen through the aperture would be more nearly equated across environments. If we found an effect of environmental context similar to the one in the restricted-view conditions of Experiment 1 (i.e., more undershooting indoors), this would suggest that the effect does not depend crucially on the visual properties of the ground covering, and would therefore more strongly implicate top-down environmental knowledge. Experiment 4, below, would address the separate issue of whether such an effect of environmental context reflects a bona fide impact on perceived egocentric distances or instead is due to top-down influences on response (output) biases.

Method

Participants

Fifty new individuals from the George Washington University community participated in the experiment. All of the participants were 18 to 23 years of age, reported normal or corrected-to-normal vision, and participated in exchange for course credit.

Stimuli and apparatus

The targets were three thin rectangular boards painted bright yellow. The boards were of three sizes (10.5 × 14.9, 11.9 × 17.9, and 13.8 × 20.7 cm) and were placed on the ground at one of seven distances (3.0, 3.5, 4.0, 4.5, 5.0, 5.5, and 6.0 m). On each trial, the long axis of the target was randomly oriented to be either parallel to the line of sight or in a frontoparallel plane. Thus, size varied, but not systematically with distance. All trials were conducted within the same indoor context employed in Experiments 1 and 2, except that here artificial turf grass was placed on the floor of the laboratory (see Fig. 1a). The aim was to approximate the visual texture of the local ground surface that had been seen through the aperture outdoors in Experiments 1 and 2. All trials were conducted with the same goggles, blindfold, and hearing protectors as in those previous experiments. Because there was no full-view condition in this experiment, the goggle aperture was used to impose monocular vision, and no eye patch was used.

Design and procedure

The environmental context (indoor vs. outdoor) was manipulated between groups. All trials used the restricted-view condition. As in Experiment 1, participants indicated distance via blindfolded-walking responses. The procedure was otherwise matched to those of Experiments 1 and 2.

Results

Mauchly’s test indicated that the sphericity assumption was violated for the distance contrast [χ 2(20) = 69.99, p < .001]. In the ANOVA results reported below, degree-of-freedom correction was applied for this contrast using the Greenhouse–Geisser estimate of sphericity (ε = .709). The data from Experiment 3 are shown in Fig. 4. Parameter estimates for all conditions are reported in Table 2.
Fig. 4

Mean response distance as a function of target distance and viewing context in Experiment 3. Distance estimates were obtained via blindfolded walking. Error bars show ±1 standard error; the thick lines show the best-fitting linear functions through the mean data. Artificial turf was the floor covering in the “indoor” viewing context. Restricted-FOV viewing was used throughout this experiment; there was no full-FOV condition

Table 2

Means and between-subjects standard errors for sensitivity (given by the slope relating response distance to target distance) and bias (given by the percent mean signed error across distances) in Experiments 3 and 4

 

Sensitivity

Bias

Exp. 3 (Blindfolded Walking) – Restricted View

 Indoor

0.75 (0.07)

–25.8 (5.5)

 Outdoor

0.99 (0.06)

–3.0 (6.1)

Exp. 4 – Outdoor First – Restricted View

 Indoor, verbal

0.71 (0.06)

–40.1 (4.6)

 Indoor, distance inferred from size*

1.01 (0.13)

1.5 (7.3)

 Outdoor, verbal

0.67 (0.07)

–41.8 (4.4)

 Outdoor, distance inferred from size*

0.85 (0.09)

5.3 (6.6)

Exp. 4 – Indoor First – Restricted View

 Indoor, verbal

0.76 (0.08)

–44.0 (3.3)

 Indoor, distance inferred from size*

0.86 (0.17)

–4.9 (7.0)

 Outdoor, verbal

0.79 (0.07)

–33.8 (4.8)

 Outdoor, distance inferred from size*

0.92 (0.10)

1.4 (7.5)

Both experiments used artificial turf as the indoor floor covering. *These data represent perceived target distances along the ground, as inferred from size gesture responses using Eq. 1.

We found a main effect of context, F(1, 48) = 34.73, p < .001, η p 2 = .420, such that responses indoors were smaller than those outdoors (means 3.28 vs. 4.58 m, respectively). There was also a main effect of distance, F(4.25, 204.22) = 67.73, p < .001, η p 2 = .585, but no Distance × Context interaction, F(4.25, 204.22) = 1.85, p = .117, η p 2 = .037.

Discussion

The results were qualitatively similar to those in the restricted-view conditions of Experiment 1, with responses in the outdoor environment being generally larger than those in the indoor environment. Interestingly, the pattern was similar even though the indoor ground covering was very different in Experiment 3 from that in Experiment 1. It is nearly impossible to completely match a grassy ground texture in real-world environments; even if the exact physical texture were used, differences in lighting could still change the visual texture. Nevertheless, given the reasonable assumption that the ground texture between environments would be more nearly matched here than in Experiment 1, the results indicate that top-down knowledge about the environment size likely played a dominant role in driving the indoor–outdoor effects in Experiment 3. A remaining issue is whether this top-down knowledge caused a perceptual rescaling of the environment, or whether it influenced the blindfolded-walking responses primarily at the output stage—for example, due to participants feeling constrained or hesitant to walk indoors when the environment was assumed to be small, or to an opposite tendency to walk farther outdoors when the environment was assumed to afford more space. This is a difficult question to answer definitively, but Experiment 4 we attempted to address the issue by replicating Experiment 3 with two response modes that were not constrained by the size of the environment. If top-down knowledge about environmental size primarily influenced the perception of egocentric distances in Experiments 1 and 3, similar effects of context under FOV restriction should be found using two responses that were unconstrained by the physical size of the environment. If the context effects in Experiments 1 and 3 operated primarily at the response (output) stage, the context effects should disappear, or at least be diminished, when using these two response types.

Experiment 4

The primary goal of Experiment 4 was to augment the size gesture responses of Experiment 2 with verbal reports of target distance. Including both response types here provides some continuity with Experiment 2, by repeating the usage of size gesture responses, and also the opportunity to explore the generalizability of the size gesture data to another response mode that is not constrained by the action space. Using verbal report of egocentric distance judgments has a rich empirical history in research (Da Silva, 1985), so using this response mode would provide an empirical link to the extensive past literature. Accordingly, Experiment 4 replicated the design of Experiment 3 but used the verbal report and size gesture response modes instead of blindfolded walking. We again used artificial turf as the ground covering indoors, to more nearly equate the similarity of the target stimulus and its immediate visual surround as seen through the FOV-restricting aperture. If the context effect from Experiment 3 was due to an output-level response tendency associated with the blindfolded-walking response, rather than to a genuine perceptual context effect elicited by top-down assumptions about the environment size, we predict that no context effect should emerge in Experiment 4, when the response modes were not constrained by the physical size of the environment.

Method

Participants

Fifty new individuals from the George Washington University community participated in the experiment. All of the participants were 18 to 30 years of age, reported normal or corrected-to-normal vision, and were paid $10/h or participated in exchange for course credit.

Stimuli and apparatus

The targets were three thin rectangular boards painted bright yellow on one side and bright orange on the other. The boards were of three sizes (10.5 × 14.9, 11.9 × 17.9, and 13.8 × 20.7 cm) and were placed on the ground at one of seven distances (3.0, 3.5, 4.0, 4.5, 5.0, 5.5, and 6.0 m). A relatively minor methodological difference from Experiment 3 is that here, the targets were angled to stand roughly perpendicular to the observer’s line of sight. As in Experiment 3, on each trial the long axis of the target was randomly oriented to be either parallel to the line of sight or in a frontoparallel plane, so that size varied, but not systematically with distance. All trials were conducted within the same indoor and outdoor contexts employed in Experiments 13, except that here, as in Experiment 3, artificial turf grass was placed on the floor of the laboratory. All trials were conducted with the same goggles, blindfold, and hearing protectors. As in Experiment 3, monocular vision was imposed by using the aperture goggles in all conditions, and thus no eye patch was required.

Design and procedure

Participants were assigned to one of two groups: the indoor-first group or the outdoor-first group. All trials employed the restricted-view condition. Target color (orange or yellow) varied between blocks and was counterbalanced. This was done to minimize biases toward producing the same responses between blocks, which might arise if participants assumed that the exact same stimulus parameters were being used in both contexts. Participants gave verbal estimates of the distance to the targets in the metric that was most familiar (most typically in feet). Prior to testing and outside of the testing spaces, the experimenter presented the observers with a yardstick and stood 3 feet shoulder-to-shoulder from them to refresh in their mind on the size of the unit of measure. Participants gave verbal and size gesture responses on every trial. The procedure was otherwise matched to those of Experiments 13.

Results

The data from Experiment 4 are shown in Fig. 5. Verbal and size gesture responses were analyzed in separate ANOVAs. The verbal analysis showed a Block Order × Context interaction (see below), suggesting that the responses in the context experienced in Block 2 were systematically influenced by the context experienced in the preceding block. For this reason, Fig. 5 plots first-block data only, to show responses that are not influenced by these carryover effects. Parameter estimates for all conditions are reported in Table 2.
Fig. 5

Mean response distance as a function of target distance and viewing context in Experiment 4. Distance estimates obtained via verbal reports are shown in the left panel; distance estimates inferred from size gestures are shown in the right panel. Error bars show ±1 standard error; the thick lines show the best-fitting linear functions through the mean data. Artificial turf was the floor covering in the “indoor” viewing context. Restricted-FOV viewing was used throughout this experiment; there was no full-FOV condition. The data shown are from the first block of trials only; see the text for details

Verbal distance estimates

Mauchly’s tests showed that the sphericity assumption was violated for the distance and Distance × Context contrasts [both χ 2s (20) > 40.79, both ps < .004]. In the ANOVA results, the degrees of freedom for these contrasts were corrected using Greenhouse–Geisser estimates of sphericity (εs = .646 [distance] and .802 [Context × Distance]). Although no main effect of block order was apparent, F(1, 48) = 0.22, p = .641, η p 2 = .005, we did observe a reliable Context × Block Order interaction, F(1, 48) = 4.83, p = .033, η p 2 = .091. Aside from a main effect of distance, however, F(3.87, 185.92) = 119.25, p < .001, η p 2 = .713, there were no other reliable effects or interactions (all Fs < 2.93, all ps > .093).

Distance inferred from size gestures

Two data points were eliminated before analysis because an inferred ground distance could not be defined—the size gestures in these cases suggested an eye-to-target distance that was less than the observer’s eye height. Mauchly’s test indicated that the sphericity assumption was violated for the distance and Distance × Context contrasts [both χ 2s (20) > 35.51, both ps < .018]. In the ANOVA results, the degrees of freedom for these contrasts were corrected using Greenhouse–Geisser estimates of sphericity (εs = .682 [distance] and .805 [Context × Distance]). The only contrast to reach significance in this analysis was distance, F(4.09, 188.30) = 48.73, p < .001, η p 2 = .514 (all other Fs < 1.78, all other ps > .189).

Discussion

A benefit of manipulating indoor versus outdoor context within subjects is that this increases statistical power by leveraging the repeated measures within participants. Here, however, block order interacted with context for the verbal responses, such that the verbal responses were greater in the outdoor context, but only when the indoor block was administered first. This indicates that some unintended between-block influences were operating when verbal responses were used. One possibility is that participants remembered their specific verbal responses or the range of their verbal responses from the first block and explicitly scaled their responses in the second block with reference to their responses in the first. Size gesture and blindfolded-walking responses might be less susceptible to this kind of effect, to the extent that these action-based responses are less verbalizable, and perhaps thus less memorable between blocks. Additional work would be required in order to interpret these block order effects and determine their source, but exploring these issues is beyond the scope of the present article. Focusing on first-block performance, with context varying between groups, provides a means of assessing context effects without any carryover effects from a previous block. The two context groups consisted of 25 participants each, thereby approximately equaling the statistical power of the within-subjects context manipulation in Experiment 2 (n = 28, total). In the first-block data, verbal responses did not differ reliably between the indoor and outdoor contexts, F(1, 48) = 0.90, p = .765, η p 2 = .002, and similarly, we found no effects of context with the size gesture responses, F(1, 47) = 0.66, p = .420, η p 2 = .014. The fact that a robust context effect emerged in Experiment 3, with the blindfolded-walking responses, but no evidence of context effects in Experiment 4, with either verbal or size gesture responses, suggests that output-level response biases were the source of the environmental context effects in Experiment 3 (and in Exp. 1).

The size gesture responses here appeared to differ somewhat from those collected in Experiment 2, with the gestured sizes being closer to accurate in Experiment 4, on average, than in Experiment 2. The explanation for this apparent difference is unclear. One methodological difference was that in Experiment 4, the physical target sizes were paired randomly with target distance—thus yielding a variety of angular target sizes; by contrast, in Experiment 2, the angular target size was kept constant across target distances. The lack of variation in angular sizes in Experiment 2 may have reduced the range of size gesture responses relative to Experiment 4. The floor coverings also differed between experiments (carpet in Exp. 2 vs. turf in Exp. 4). Whatever the explanation for the apparent difference in the size gesture responses might be, the comparison of interest in both experiments was the within-subjects comparison in responses between the indoor and outdoor contexts. In both experiments, size gesture responses did not differ reliably between the indoor and outdoor contexts.

General discussion

At the outset, we suggested that assumptions or knowledge about the size of the surrounding environment might play some role in scaling judgments of object distances within the environment. We set out to test this idea in the context of real indoor and outdoor contexts. In the present series of experiments, we found clear evidence of environmental context effects. Interestingly, however, these effects were strongly dependent on the response modality, as well as on whether a full or restricted FOV was provided. We will organize our discussion of context effects around the FOV manipulation.

Context effects under restricted FOV

We assumed that the top-down effects associated with knowledge about the size of the environment would be minimal when visual (data-driven) information was abundant. Accordingly, to equate visual information between environments as much as possible and create conditions favorable for top-down assumptions or knowledge to be manifest, in some conditions we restricted observers’ FOV using an aperture that occluded all but the target and its immediately surrounding ground surface. The environmental context effects under these restricted-view conditions depended strongly upon response mode. In Experiments 1 and 3, using the blindfolded-walking technique, distances were underestimated indoors relative to outdoors. In Experiment 2, in which distances were inferred using the size gesture technique, responses did not differ between indoor and outdoor environments. In Experiment 4, using a different ground covering indoors, we again found no difference between indoor and outdoor performance in either size gesture or verbal responses.

It is important to note that artificial turf and real grass do not appear completely identical; even if the ground covering were completely identical, differences in lighting between the environments could create differences in visible texture that might be relevant for distance and layout perception. Nevertheless, on the assumption that the artificial turf and grass ground coverings provided a reasonable match of the visual cues (including local ground texture) across environments, Experiment 4 was our strongest test of the possible influence of top-down environmental assumptions or knowledge on visually perceived distance. This study showed, however, that even under these matched conditions there was no apparent influence of this kind of top-down effect—a result confirmed with two very different response modes (size gestures and verbal reports). Although we tested two relatively common settings, it remains possible that top-down environmental context effects might manifest under other circumstances (e.g., much larger differences in setting size or the target distances).

Experiment 3 used conditions identical to those in Experiment 4, including a close matching of the visible ground textures across environments, but used blindfolded walking rather than verbal or size gesture responses. The blindfolded-walking responses in Experiment 3 did show evidence of top-down knowledge of the environmental size, such that distances were underestimated more while indoors. We interpret the differing results between Experiments 3 and 4 as stemming from response-specific postperceptual (output) effects. According to this view, knowledge of the environmental context had very little impact on the underlying perceived distances across contexts. When blindfolded walking was used to judge distances, observers presumably were aware that they were indoors but otherwise had no basis for knowing how much space was available for them to walk without vision. The fact that participants generally underwalked under these circumstances may reflect a tendency to be conservative when producing walked distances, owing to uncertainty about the room size. This interpretation is supported by the lack of an indoor–outdoor difference in Experiment 1, in which full views of the environments were provided; presumably, the full FOV minimized uncertainty about the room size (see also Andre & Rogers, 2006). Also, the indoor undershooting disappeared in Experiment 4 using two response types (verbal and size gesture responses) that were not constrained by the environment size.

Context effects under full FOV

In general, the pattern of indoor versus outdoor effects that we saw in Experiments 1 and 2 under full FOVs mirrored the effects found in past studies (Andre & Rogers, 2006; Lappin et al., 2006; Witt et al., 2007), such that target distances were judged to be somewhat larger indoors than the same physical distances outdoors, at least for distances inferred from size gestures; even with blindfolded-walking responses, however, the response sensitivity slopes were higher indoors than outdoors. Our data support the assumption that data-driven visual information is generally stronger and/or more abundant in indoor than in outdoor environments. Indoor and outdoor environments are distinguished by a great many distinct visual differences (e.g., visible ground texture, the availability of linear perspective or binocular disparity cues, location of the visible horizon, etc.), but as yet the relative importance of these various differences is unknown and remains an important topic for future research.

Effect of FOV restriction

Although the foregoing discussion may give the impression that the pattern of undershooting indoors under the blindfolded-walking response mode stems purely from a response bias (e.g., walking less to avoid bumping into an unseen wall), it is important to point out that the results of our FOV manipulation in Experiments 1 and 2 point to an additional perceptual component. Specifically, restricting the FOV when indoors resulted in similar decrements in performance for both blindfolded-walking judgments and distances inferred from size gestures. This result generally confirms those from past work (Creem-Regehr et al., 2005; Wu et al., 2004). This pattern is also understandable from an informational perspective, because restricting the FOV often has the effect of occluding the visual information that specifies the scale of the environment (e.g., texture gradient along the ground surface; He et al., 2004; Li & Durgin, 2012; Sinai, Ooi, & He, 1998). Taken together, these results suggest that the indoor blindfolded walking undershooting under FOV restriction stems from both perceptual underestimation and response bias. The outdoor blindfolded walking in Experiment 1 was apparently unaffected by FOV restriction. Our view is that some perceptual underestimation did indeed occur for these outdoor walking responses, in keeping with the outdoor size gesture responses in Experiment 2 and in past work (Creem-Regehr et al., 2005; Wu et al., 2004). These responses did not show evidence of perceptual underestimation, however, because blindfolded walking outdoors was effectively released from the walking-specific response bias present when indoors.

In our Experiment 2, size gesture responses were better-scaled under full FOV in both indoor and outdoor environments. In Experiment 1, however, we found no effect of FOV restriction in blindfolded-walking judgments conducted outside; this apparently contrasts with Wu et al. (2004), who found that blindfolded-walking responses outside after FOV-restricted target viewing were scaled more poorly than those under full-FOV target viewing. We have also found no effect of completely occluding the nearby ground surface in indoor environments (Gajewski, Wallin, & Philbeck, 2014). The source of these apparent discrepancies with earlier work performed by Wu et al. (2004) remains unclear. The answer may be related to uncontrolled cross-experiment variations in the availability of other visual information lying outside the nearby ground surface that could be used to scale visual space. In past work, we have speculated that the global visual information specifying the size of the environment (i.e., coming from outside the nearby ground plane) might be used to scale visual space, perhaps in addition to information in the nearby ground surface (Gajewski, Philbeck, et al., 2014; Gajewski, Wallin, & Philbeck, 2014). If this is true, the visual system’s reliance on relatively local ground plane cues versus more global environment size cues might itself depend upon the environmental context, with the nearby ground plane taking on relatively more weight in outdoor environments, which provide fewer planar surfaces in depth, fewer texture gradients, less linear perspective, and so forth. More research will be required in order to address this issue.

Conclusions

Taken together, these results confirmed past work showing that environmental context influences distance judgments. We drew a distinction between context effects stemming from visual (data-driven) information between environments and effects stemming from top-down assumptions or knowledge about the environmental context. Our results speak to the relative importance of these two potential sources in producing effects of environmental context in distance judgments. In our experiments, top-down assumptions about the context played a role particularly for blindfolded-walking responses when participants were unaware of the exact size of the environment but likely assumed it to be relatively small. When the responses were not constrained by the size of the environment, however, we found little evidence for a role of top-down assumptions in the relatively common experimental contexts that we tested.

Notes

Author note

The research reported in this publication was supported by the National Eye Institute of the National Institutes of Health under Award No. R01EY021771. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This research was also supported by National Science Foundation Graduate Fellowship DGE-1246908 to C.P.W.

References

  1. Andre, J., & Rogers, S. (2006). Using verbal and blind-walking distance estimates to investigate the two visual systems hypothesis. Perception & Psychophysics, 68, 353–361. doi: https://doi.org/10.3758/BF03193682 CrossRefGoogle Scholar
  2. Barlow, H. B. (1958). Temporal and spatial summation in human vision at different background intensities. Journal of Physiology, 141, 337–350.CrossRefPubMedPubMedCentralGoogle Scholar
  3. Creem-Regehr, S. H., Willemsen, P., Gooch, A. A., & Thompson, W. B. (2005). The influence of restricted viewing conditions on egocentric distance perception: Implications for real and virtual indoor environments. Perception, 34, 191–204. doi: https://doi.org/10.1068/p5144 CrossRefPubMedGoogle Scholar
  4. Cutting, J. E., & Vishton, P. M. (1995). Perceiving layout and knowing distances: The integration, relative potency, and contextual use of different information about depth. In W. Epstein & S. J. Rogers (Eds.), Perception of space and motion (Handbook of Perception and Cognition, 2nd ed., pp. 71–117). San Diego, CA: Academic Press.Google Scholar
  5. Da Silva, J. A. (1985). Scales of perceived egocentric distance in a large open field: Comparison of three psychophysical methods. American Journal of Psychology, 98, 119–144.Google Scholar
  6. Elliott, D. (1987). The influence of walking speed and prior practice on locomotor distance estimation. Journal of Motor Behavior, 19, 476–485.CrossRefPubMedGoogle Scholar
  7. Epstein, W., Park, J., & Casey, A. (1961). The current status of the size-distance hypotheses. Psychological Bulletin, 58, 491–514.CrossRefPubMedGoogle Scholar
  8. Foley, J. M. (1977). Effect of distance information and range on two indices of visually perceived distance. Perception, 6, 449–460.CrossRefPubMedGoogle Scholar
  9. Foley, J. M. (1991). Binocular space perception. In D. Regan (Ed.), Binocular vision (Vol. 9, pp. 75–92). Boca Raton, FL: CRC Press.Google Scholar
  10. Gajewski, D. A., Philbeck, J. W., Pothier, S., & Chichka, D. (2010). From the most fleeting of glimpses: On the time course for the extraction of distance information. Psychological Science, 21, 1446–1453. doi: https://doi.org/10.1177/0956797610381508 CrossRefPubMedPubMedCentralGoogle Scholar
  11. Gajewski, D. A., Philbeck, J. W., Wirtz, P. W., & Chichka, D. (2014). Angular declination and the dynamic perception of egocentric distance. Journal of Experimental Psychology: Human Perception and Performance, 40, 361–377.PubMedGoogle Scholar
  12. Gajewski, D. A., Wallin, C. P., & Philbeck, J. W. (2014). Gaze behavior and the perception of egocentric distance. Journal of Vision, 14(1), 20:1–19. doi: https://doi.org/10.1167/14.1.20 CrossRefPubMedPubMedCentralGoogle Scholar
  13. Geisler, W. S., & Kersten, D. (2002). Illusions, perception and Bayes. Nature Neuroscience, 5, 508–510.CrossRefPubMedGoogle Scholar
  14. Gogel, W. C. (1984). The role of perceptual interrelations in figural synthesis. In P. C. Dodwell & T. Caelli (Eds.), Figural synthesis (pp. 31–82). Hillsdale, NJ: Erlbaum.Google Scholar
  15. Haber, R. N., & Levin, C. A. (2001). The independence of size perception and distance perception. Perception & Psychophysics, 63, 1140–1152.CrossRefGoogle Scholar
  16. He, Z. J., Wu, B., Ooi, T. L., Yarbrough, G., & Wu, J. (2004). Judging egocentric distance on the ground: Occlusion and surface integration. Perception, 33, 789–806. doi: https://doi.org/10.1068/p5256a CrossRefPubMedGoogle Scholar
  17. Iosa, M., Fusco, A., Morone, G., & Paolucci, S. (2012). Walking there: Environmental influence on walking-distance estimation. Behavioural Brain Research, 226, 124–132.CrossRefPubMedGoogle Scholar
  18. Kilpatrick, F., & Ittelson, W. (1953). The size-distance invariance hypothesis. Psychological Review, 60, 223–231.CrossRefPubMedGoogle Scholar
  19. Lappin J. S., Shelton A. L., & Rieser J. J. (2006). Environmental context influences visually perceived distance. Perception & Psychophysics, 68, 571–581. doi: https://doi.org/10.3758/BF03208759 CrossRefGoogle Scholar
  20. Li, Z., & Durgin, F. H. (2012). A comparison of two theories of perceived distance on the ground plane: The angular expansion hypothesis and the intrinsic bias hypothesis. i-Perception, 3, 368–383.CrossRefPubMedPubMedCentralGoogle Scholar
  21. Loomis, J. M., da Silva, J. A., Fujita, N., & Fukusima, S. S. (1992). Visual space perception and visually directed action. Journal of Experimental Psychology: Human Perception and Performance, 18, 906–921. doi: https://doi.org/10.1037/0096-1523.18.4.906 PubMedGoogle Scholar
  22. Loomis, J. M., & Philbeck, J. W. (2008). Measuring spatial perception with spatial updating and action. In R. L. Klatzky, B. MacWhinney, & M. Behrmann (Eds.), Embodiment, ego-space, and action (pp. 1–44). New York, NY: Psychology Press.Google Scholar
  23. Lunsman, M., Edwards, J. D., Andel, R., Small, B. J., Ball, K. K., & Roenker, D. L. (2008). What predicts changes in useful field-of-view test performance? Psychology and Aging, 23, 917–927. doi: https://doi.org/10.1037/a0013466 CrossRefPubMedPubMedCentralGoogle Scholar
  24. Mon-Williams, M., & Tresilian, J. R. (1999). The size-distance paradox is a cognitive phenomenon. Experimental Brain Research, 126, 578–582.CrossRefPubMedGoogle Scholar
  25. Philbeck, J. W., & Loomis, J. M. (1997). Comparison of two indicators of perceived egocentric distance under full-cue and reduced-cue conditions. Journal of Experimental Psychology: Human Perception and Performance, 23, 72–85. doi: https://doi.org/10.1037/0096-1523.23.1.72 PubMedGoogle Scholar
  26. Philbeck, J. W., & Witt, J. K. (2015). Action-specific Influences on perception and post-perceptual processes: Present controversies and future directions. Psychological Bulletin, 141, 1120–1144.CrossRefPubMedPubMedCentralGoogle Scholar
  27. Philbeck, J. W., Woods, A. J., Arthur, J., & Todd, J. (2008). Progressive locomotor recalibration during blind walking. Perception & Psychophysics, 70, 1459–1470. doi: https://doi.org/10.3758/PP.70.8.1459 CrossRefGoogle Scholar
  28. Rand, K. M., Tarampi, M. R., Creem-Regehr, S. H., & Thompson, W. B. (2011). The importance of a visual horizon for distance judgments under severely degraded vision. Perception, 40, 143–154.CrossRefPubMedPubMedCentralGoogle Scholar
  29. Rieser, J. J., Pick, H. L., Jr., Ashmead, D. H., & Garing, A. E. (1995). Calibration of human locomotion and models of perceptual–motor organization. Journal of Experimental Psychology: Human Perception and Performance, 21, 480–497. doi: https://doi.org/10.1037/0096-1523.21.3.480 PubMedGoogle Scholar
  30. Sedgwick, H. A. (1986). Space perception. In K. R. Boff, L. Kaufman, & J. P. Thomas (Eds.), Handbook of perception and human performance: Vol. 1. Sensory processes and perception (pp. 21.21–21.57). New York, NY: Wiley.Google Scholar
  31. Sinai, M. J., Ooi, T. L., & He, Z. J. (1998). Terrain influences the accurate judgement of distance. Nature, 395, 497–500.CrossRefPubMedGoogle Scholar
  32. Steenhuis, R. E., & Goodale, M. A. (1988). The effects of time and distance on accuracy of target-directed locomotion: Does an accurate short-term memory for spatial location exist? Journal of Motor Behavior, 20, 399–415.CrossRefPubMedGoogle Scholar
  33. Teghtsoonian, R., & Teghtsoonian, M. (1970). Scaling apparent distance in natural outdoor settings. Psychonomic Science, 21, 215–216.CrossRefGoogle Scholar
  34. Toye, R. C. (1986). The effect of viewing position on the perceived layout of space. Perception & Psychophysics, 40, 85–92. doi: https://doi.org/10.3758/BF03208187 CrossRefGoogle Scholar
  35. Vishton, P. M., Rea, J. G., Cutting, J. E., & Nuñez, L. N. (1999). Comparing effects of the horizontal–vertical illusion on grip scaling and judgment: Relative versus absolute, not perception versus action. Journal of Experimental Psychology: Human Perception and Performance, 25, 1659–1672. doi: https://doi.org/10.1037/0096-1523.25.6.1659 PubMedGoogle Scholar
  36. Wagner, M. (1985). The metric of visual space. Perception & Psychophysics, 38, 483–495.CrossRefGoogle Scholar
  37. Watson, A. B. (1986). Temporal sensitivity. In K. R. Boff, L. Kaufman, & J.P. Thomas (Eds.), Handbook of perception and human performance: Volume 1. Sensory processes and perception (pp. 6.1–6.43). New York, NY: Wiley.Google Scholar
  38. Witt, J. K., Stefanucci, J. K., Riener, C. R., & Proffitt, D. R. (2007). Seeing beyond the target: Environmental context affects distance perception. Perception, 36, 1752–1768.CrossRefPubMedGoogle Scholar
  39. Wu, B., He, Z. J., & Ooi, T. L. (2007). The linear perspective information in ground surface representation and distance judgment. Perception & Psychophysics, 69, 654–672.CrossRefGoogle Scholar
  40. Wu, B., Ooi, T. L., & He, Z. J. (2004). Perceiving distance accurately by a directional process of integrating ground information. Nature, 428, 73–77. doi: https://doi.org/10.1038/nature02350 CrossRefPubMedGoogle Scholar
  41. Yang, Z., & Purves, D. (2003). A statistical explanation of visual space. Nature Neuroscience, 6, 632–640.CrossRefPubMedGoogle Scholar

Copyright information

© The Psychonomic Society, Inc. 2017

Authors and Affiliations

  • John W. Philbeck
    • 1
  • Daniel A. Gajewski
    • 1
  • Sandra Mihelič Jaidzeka
    • 1
  • Courtney P. Wallin
    • 1
  1. 1.Department of PsychologyGeorge Washington UniversityWashingtonUSA

Personalised recommendations